Newly diagnosed Type 1 Diabetes (T1D) patients often struggle to obtain effective Blood Glucose (BG) prediction models due to the lack of sufficient BG data from Continuous Glucose Monitoring (CGM), presenting a significant "cold start" problem in patient care. Utilizing population models to address this challenge is a potential solution, but collecting patient data for training population models in a privacy-conscious manner is challenging, especially given that such data is often stored on personal devices. Considering the privacy protection and addressing the "cold start" problem in diabetes care, we propose "GluADFL", blood Glucose prediction by Asynchronous Decentralized Federated Learning. We compared GluADFL with eight baseline methods using four distinct T1D datasets, comprising 298 participants, which demonstrated its superior performance in accurately predicting BG levels for cross-patient analysis. Furthermore, patients' data might be stored and shared across various communication networks in GluADFL, ranging from highly interconnected (e.g., random, performs the best among others) to more structured topologies (e.g., cluster and ring), suitable for various social networks. The asynchronous training framework supports flexible participation. By adjusting the ratios of inactive participants, we found it remains stable if less than 70% are inactive. Our results confirm that GluADFL offers a practical, privacy-preserved solution for BG prediction in T1D, significantly enhancing the quality of diabetes management.

Nucleic acid amplification is a crucial method in infectious disease diagnostic and identification of cancer mutations that allows for informed treatment. The CMOS-fabricated Ion-Sensitive Field-Effect Transistor (ISFET) can be exploited at the Point-of-Care as part of a handheld Lab-on-Chip (LoC) device thanks to its pH sensitivity to achieve real-time sensing of the hydrogen release during amplification. Current approaches involve off-chip processing of the sensor data, introducing challenges due to the amount of data to be transmitted and analysed, and to ensure patient privacy in a procedure that involves multiple data transfers. This paper proposes a pixel architecture for real-time in-pixel classification of the sensor data. A 4x4 ISFET array is shown, with a design that allows for full scalability to a much larger device. The system is developed in TSMC 65nm technology, resulting in a total pixel area of 51.4x49.54μm. Classification of positive and negative experiments is simulated with ideal and real experimental data.

BACKGROUND: Blood-related infections are a significant concern in healthcare. They can lead to serious medical complications and even death if not promptly diagnosed and treated. Throughout time, medical research has sought to identify clinical factors and strategies to improve the management of these conditions. The increasing adoption of electronic health records has led to a wealth of electronically available medical information and predictive models have emerged as invaluable tools. This manuscript offers a detailed survey of machine-learning techniques used for the diagnosis and prognosis of bacteraemia, bloodstream infections, and sepsis shedding light on their efficacy, potential limitations, and the intricacies of their integration into clinical practice. METHODS: This study presents a comprehensive analysis derived from a thorough search across prominent databases, namely EMBASE, Google Scholar, PubMed, Scopus, and Web of Science, spanning from their inception dates to October 25, 2023. Eligibility assessment was conducted independently by investigators, with inclusion criteria encompassing peer-reviewed articles and pertinent non-peer-reviewed literature. Clinical and technical data were meticulously extracted and integrated into a registry, facilitating a holistic examination of the subject matter. To maintain currency and comprehensiveness, readers are encouraged to contribute manuscript suggestions and/or reports for integration into this living registry. RESULTS: While machine learning (ML) models exhibit promise in advanced disease stages such as sepsis, early stages remain underexplored due to data limitations. Biochemical markers emerge as pivotal predictors during early stages such as bacteraemia, or bloodstream infections, while vital signs assume significance in sepsis prognosis. Integrating temporal trend information into conventional machine learning models appears to enhance performance. Unfortunately, sequential deep learning models face challenges, showing minimal performance improvements and significant drops in external datasets, potentially due to learning missing patterns within the scarce data available rather than understanding disease dynamics. Real-life implementation receives limited attention, as meeting design requirements proves challenging within existing healthcare infrastructure. The data collected in an event-based fashion during clinical practice is insufficient to fully harness the potential of these data-hungry models. Despite limitations, opportunities abound in leveraging flexible models and exploiting real-time non-invasive data collection technologies such as wearable devices or microneedles. Addressing research gaps in early disease stages, harnessing patient history data often underused, and embracing continual diagnostics beyond treatment initiation are crucial for improving healthcare decision-making support and adoption across the entire management pathway. CONCLUSIONS: This comprehensive survey illuminates the landscape of ML applications in blood-related infection management, offering insights for future research and clinical practice. Implementing clinical ML-based clinical decision support systems requires balancing research with practical considerations. Current methodologies often lead to complex models lacking transparency and practical validation. Integration into healthcare systems faces regulatory, privacy, and trust challenges. Clear presentations and adherence to standards are essential to boost confidence in machine learning models for real-world healthcare applications.

Accurate prediction of future blood glucose (BG) levels can effectively improve BG management for people living with type 1 or 2 diabetes, thereby reducing complications and improving quality of life. The state of the art of BG prediction has been achieved by leveraging advanced deep learning methods to model multimodal data, i.e., sensor data and self-reported event data, organized as multi-variate time series (MTS). However, these methods are mostly regarded as "black boxes" and not entirely trusted by clinicians and patients. In this paper, we propose interpretable graph attentive recurrent neural networks (GARNNs) to model MTS, explaining variable contributions via summarizing variable importance and generating feature maps by graph attention mechanisms instead of post-hoc analysis. We evaluate GARNNs on four datasets, representing diverse clinical scenarios. Upon comparison with fifteen well-established baseline methods, GARNNs not only achieve the best prediction accuracy but also provide high-quality temporal interpretability, in particular for postprandial glucose levels as a result of corresponding meal intake and insulin injection. These findings underline the potential of GARNN as a robust tool for improving diabetes care, bridging the gap between deep learning technology and real-world healthcare solutions.

Real-time continuous glucose monitoring (CGM), augmented with accurate glucose prediction, offers an effective strategy for maintaining blood glucose levels within a therapeutically appropriate range. This is particularly crucial for individuals with type 1 diabetes (T1D) who require long-term self-management. However, with extensive glycemic variability, developing a prediction algorithm applicable across diverse populations remains a significant challenge. Leveraging meta-learning for domain generalization, we propose GPFormer, a Transformer-based zero-shot learning method designed for multi-horizon glucose prediction. We developed GPFormer on the REPLACE-BG dataset, comprising 226 participants with T1D, and proceeded to evaluate its performance using three external clinical datasets with CGM data. These included the OhioT1DM dataset, a publicly available dataset including 12 T1D participants, as well as two proprietary datasets. The first proprietary dataset included 22 participants, while the second contained 45 participants, encompassing a diverse group with T1D, type 2 diabetes, and those without diabetes, including patients admitted to hospitals. These four datasets include both outpatient and inpatient settings, various intervention strategies, and demographic variability, which effectively reflect real-world scenarios of CGM usage. When compared with a group of machine learning baseline methods, GPFormer consistently demonstrated superior performance and achieved the lowest root mean square error for all the evaluated datasets up to a prediction horizon of two hours. These experimental results highlight the effectiveness and generalizability of the proposed model across a variety of populations, demonstrating its substantial potential to enhance glucose management in a wide range of practical clinical settings.

Photoplethysmography (PPG) is a non-invasive optical technique used to measure cardiovascular activity in both clinical and consumer applications. PPG signals are notoriously susceptible to noise and motion artefacts and in applications where high-frequency PPG signals are required, these issues are even more pronounced. In this work, we propose a two-stage algorithm to process these PPG waveforms without prior knowledge of their noise characteristics. To evaluate its performance, we utilized a set of PPG signal quality indices used to classify waveform quality. We found that using a Chebyshev type-II band-pass filter with cutoff frequencies of 0.35 Hz for high pass and 10 Hz for low pass performs best across six measured Signal Quality Indices (SQIs), achieving a 12% improvement in the dynamic time-warping (DTW) SQI. Furthermore, using a modified Savitzky-Golay smoothing algorithm with a frequency domain-derived window parameter leads to additional improvements in SQI-derived features of 14.09% and 12.22% in DTW and skewness respectively. Our proposed approach provides a universal solution for processing high-frequency raw PPG recordings and prepares them for the extraction of morphological features.

In this work, we propose a novel approach to integrate a scalable compressed sensing methodology in the analogue domain with a CMOS ISFET array. A current conveyor is employed with a switched capacitor to encode the output current from each ISFET sensor to a corresponding charge onto a capacitor, following by a pseudo-random non-zero diagonal sampling matrix that is generated by Linear Feedback Shift Registers (LSFR) for array sampling. The design also features a 12-bit Successive Approximation Register (SAR) ADC, enabling power efficient conversions at 50 KSamples/s using a 1.25 MHz clock, with an ENOB of 10.3. The 32 × 32 array is divided into 16 clusters, each containing 64 pixels arranged in an 8 × 8 configuration serving as a compressed sensing unit block. The overall system is designed under a 65 nm process occupying a silicon area of 0.375 mm2. It operates at a programmable frame rate of 30 - 240 fps, with an overall power consumption of 17.13 -117.23 μW, and a lowest energy per pixel of 394 pJ in compressed sensing mode. We verify the performance of the system with a PSNR comparison for image quality under two scenarios where CS is either enabled or disabled.

A portable and low-cost platform for continuous dengue patient monitoring will be presented. The front end of the system leverages PPG for non-invasive acquisition of key physiological parameters such as haematocrit levels and blood pressure changes. The edge computation capability of the ESP32-S3 enables the high data rate to be processed on-chip, reducing backend computation and network load. We expect this platform can serve as a cost-effective alternative to dengue patient monitoring in LMICs, reducing clinician workloads and improving patient outcomes.

BRAF p.V600E mutations are key biomarkers for colorectal cancer (CRC) which are associated with poor patient prognosis and response to EGFR treatment. This paper demonstrates a proof-of-concept study for a portable Lab-on-Chip (LoC) device integrated with Ion-Sensitive Field-Effect Transistor (ISFET) sensors in detecting BRAF p.V600E biomarkers with high accuracy and speed, for Point-of-Care (PoC) testing of CRC. Wild-type and mutant-type copies of the BRAF gene were successfully distinguished using chip-compatible loop-mediated isothermal amplification (LAMP) reactions. Optimisation of the LAMP assay targeting BRAF p.V600E was performed and tested using qPCR instrumentation as a gold standard and benchmark for the LoC, exhibiting improved limits of detection (LODs) at 102copies/µL in under 15 minutes. A bespoke signal processing methodology was also developed using the Convolutional Neural Network EEGNet to classify nucleic acid amplification experiments on ISFET arrays, achieving an accuracy of over 95% and designed to be easily transferable to novel DNA/RNA targets. The findings show evidence of the potential to employ ISFET sensing in PoC diagnostics for CRC, ultimately bridging the gap in accessibility in limited resource settings.

PIK3CA p.E545K mutation is a well-studied breast cancer biomarker with a clinical significance as a therapeutic target, particularly with the use of the small molecule inhibitor Alpelisib. An issue with detecting this mutation and other mutations in this exon is that 98% of its sequence homology is identical to a pseudogene, a non-functional non-coding gene found in chromosome 22. This paper aims to use ISFET enabled Lab-on-Chip (LoC) technology, coupled with a well-studied isothermal amplification method (LAMP), as a means to distinguish wild-type (WT), mutant (MT) and pseudogene (PG) copies of DNA. In an age where affordability and accessibility of diagnostic tests is of crucial importance, the use of CMOS technology offers a great potential as an alternative for regular near-patient molecular testing using liquid biopsies. Bespoke primer design is tested and optimised with synthetic DNA to achieve high specificity and low sensitivity (100 copies). The primer efficiencies were also tested on our in-house LoC system showing near identical results to those obtained from a qPCR instrument. Spiking experiments were also conducted, where mixed populations of WT and MT were tested to assess the primers' abilities to estimate the variant allele frequency (VAF) of p.E545K and to mimic clinical scenarios. The results continue to depict how LoC technology in partnership with LAMP detection can be used in a liquid biopsy setting to detect blood DNA biomarkers to assist better patient stratification.

This work presents a chemical readout system designed in TSMC 180 nm technology. The proposed design has an input range of 0-1.8 V, linearity (R2) of over 0.997, high sensitivity of 600 KHz/pH, a maximum frame rate of 1.4 μ$s$and a small chip area. The readout system includes an Ion-Sensitive Field Effect Transistor (ISFET) front-end that works in the saturation region, trans-linear circuits for linearity enhancement, and a CCO (Current Controlled Oscillator)-based ADC as an analogue to digital converter. This system was designed to provide a good balance between input range, linearity, and silicon area. The proposed architecture is capable of compensating for 400 mV of trapped charge by changing the biasing current of the lineariser as a universal quadratic equation solver.

This paper presents an experimental characterisation of the pH dependence of drift on ISFET sensors. Experiments are run on an array of over 4000 sensors fabricated in commercial CMOS technology with pH buffer solutions of known pH. A mathematical model is built and the fitted coefficients are compared between experiments where the pH is constant and where pH changes. An exponential and a linear model are compared, as well as different metrics for coefficient comparison. It is shown that a dependence can be found between drift and pH variation when drift rate exceeds$15\mu V/s$. This serves as a first step towards the development of a new metric to aid classification of nucleic acid amplification experiments.

This paper presents a winner-take-all (WTA) approach for implementing background inhibition in neuromorphic Ion-Sensitive Field-Effect Transistor (ISFET) arrays. The integration of WTA, integrate and fire (I&F) and CMOS-based electrochemical readout paves the way for the next generation of Lab-on-chip (LoC) platforms to diagnose and classify infectious diseases using sensor learning. The integration of the WTA in individual pixels allows for spatial adaptive filtering which can help eliminate the dynamic background due to ion accumulation at the gate of the sensors. The readout is done through address-event representation (AER) to enable ultra-low power data acquisition. The cluster implementation makes the design scalable for implementation as part of a large-scale integrated sensor. The paper proposes a novel ultra-low powered approach where the pixel power consumption ranges from 171.6nW to 410.9nW with an expected sensitivity of 20.2 KHz/dpH to 29.1 KHz/dpH. The sensor array is implemented in TSMC 0.18$\mu\mathrm{m}$.

This paper demonstrates a novel framework utilising artificial neural networks (ANNs) to identify electrochemical signals, estimate drift and perform signal extraction for ISFET sensors. We propose a neural network based on the combination of Multi-Layer Perceptrons (MLPs) and Gated Recurrent Units (GRUs), to aid the analysis of chemical reactions for ISFETs by identifying the reaction origin and compensating for drift in real time. The model is trained and tested using Keras on an artificial dataset, achieving a reaction classification accuracy of 89.71 % with an average delay of 15.73s. We have also implemented the proposed model on an FPGA through high-level synthesis (HLS) with a tunable latency of 56254 clock cycles under an 100 MHz clock. This work paves the way for enhancing biosensors with ANNs, where a smart electrochemical imager with integrated edge processing can be implemented for various biomedical applications.

Deep learning models have achieved the state of the art in blood glucose (BG) prediction, which has been shown to improve type 1 diabetes (T1D) management. However, most existing models can only provide single-horizon prediction and face a variety of real-world challenges, such as lacking hardware implementation and interpretability. In this work, we introduce a new deep learning framework, the edge-based temporal fusion Transformer (E-TFT), for multi-horizon BG prediction, and implement the trained model on a customized wristband with a system on a chip (Nordic nRF52832) for edge computing. E-TFT employs a self-attention mechanism to extract long-term temporal dependencies and enables post-hoc explanation for feature selection. On a clinical dataset with 12 T1D subjects, it achieved a mean root mean square error of 19.09 ± 2.47 mg/dL and 32.31 ± 3.79 mg/dL for 30 and 60-minute prediction horizons, respectively, and outperformed all the considered baseline methods, such as N-BEATS and N-HiTS. The proposed model is effective for multi-horizon BG prediction and can be deployed on wearable devices to enhance T1D management in clinical settings.

Blood glucose (BG) prediction is essential to the success of glycemic control in type 1 diabetes (T1D) management. Empowered by the recent development of the Internet of Medical Things (IoMT), continuous glucose monitoring (CGM) and deep learning technologies have been demonstrated to achieve the state of the art in BG prediction. However, it is challenging to implement such algorithms in actual clinical settings to provide persistent decision support due to the high demand for computational resources, while smartphone-based implementations are limited by short battery life and require users to carry the device. In this work, we propose a new deep learning model using an attention-based evidential recurrent neural network and design an IoMT-enabled wearable device to implement the embedded model, which comprises a low-cost and low-power system on a chip to perform Bluetooth connectivity and edge computing for real-time BG prediction and predictive hypoglycemia detection. In addition, we developed a smartphone app to visualize BG trajectories and predictions, and desktop and cloud platforms to backup data and fine-tune models. The embedded model was evaluated on three clinical data sets including 47 T1D subjects. The proposed model achieved superior performance of root mean square error (RMSE), mean absolute error, and glucose-specific RMSE, and obtained the best accuracy for hypoglycemia detection when compared with a group of machine learning baseline methods. Moreover, we performed hardware-in-the-loop in silico trials with ten virtual T1D adults to test the whole IoMT system with predictive low-glucose management, which significantly reduced hypoglycemia and improved BG control.

Data-driven approaches for molecular diagnostics are emerging as an alternative to perform an accurate and inexpensive multi-pathogen detection. A novel technique called Amplification Curve Analysis (ACA) has been recently developed by coupling machine learning and real-time Polymerase Chain Reaction (qPCR) to enable the simultaneous detection of multiple targets in a single reaction well. However, target classification purely relying on the amplification curve shapes currently faces several challenges, such as distribution discrepancies between different data sources of synthetic DNA and clinical samples (i.e., training vs testing). Optimisation of computational models is required to achieve higher performance of ACA classification in multiplex qPCR through the reduction of those discrepancies. Here, we proposed a novel transformer-based conditional domain adversarial network (T-CDAN) to eliminate data distribution differences between the source domain (synthetic DNA data) and the target domain (clinical isolate data). The labelled training data from the source domain and unlabelled testing data from the target domain are fed into the T-CDAN, which learns both domains' information simultaneously. After mapping the inputs into a domain-irrelevant space, T-CDAN removes the feature distribution differences and provides a clearer decision boundary for the classifier, resulting in a more accurate pathogen identification. Evaluation of 198 clinical isolates containing three types of carbapenem-resistant genes (blaNDM,blaIMPandblaOXA-48) illustrates a curve-level accuracy of 93.1% and a sample-level accuracy of 97.0% using T-CDAN, showing an accuracy improvement of 20.9% and 4.9% respectively, compared with previous methods. This research emphasises the importance of deep domain adaptation to enable high-level multiplexing in a single qPCR reaction, providing a solid approach to extend qPCR instruments' capabilities without hardware modification in real-world clinical applications.

Mealtime insulin dosing is a major challenge for people living with type 1 diabetes (T1D). This task is typically performed using a standard formula that, despite containing some patient-specific parameters, often leads to sub-optimal glucose control due to lack of personalization and adaptation. To overcome the previous limitations here we propose an individualized and adaptive mealtime insulin bolus calculator based on double deep Q-learning (DDQ), which is tailored to the patient thanks to a personalization procedure relying on a two-step learning framework. The DDQ-learning bolus calculator was developed and tested using the UVA/Padova T1D simulator modified to reliably mimic real-world scenarios by introducing multiple variability sources impacting glucose metabolism and technology. The learning phase included a long-term training of eight sub-population models, one for each representative subject, selected thanks to a clustering procedure applied to the training set. Then, for each subject of the testing set, a personalization procedure was performed, by initializing the models based on the cluster to which the patient belongs. We evaluated the effectiveness of the proposed bolus calculator on a 60-day simulation, using several metrics representing the goodness of glycemic control, and comparing the results with the standard guidelines for mealtime insulin dosing. The proposed method improved the time in target range from 68.35% to 70.08% and significantly reduced the time in hypoglycemia (from 8.78% to 4.17%). The overall glycemic risk index decreased from 8.2 to 7.3, indicating the benefit of our method when applied for insulin dosing compared to standard guidelines.

Recent advancements in hybrid closed-loop systems, also known as the artificial pancreas (AP), have been shown to optimize glucose control and reduce the self-management burdens for people living with type 1 diabetes (T1D). AP systems can adjust the basal infusion rates of insulin pumps, facilitated by real-time communication with continuous glucose monitoring. Deep reinforcement learning (DRL) has introduced new paradigms of basal insulin control algorithms. However, all the existing DRL-based AP controllers require extensive random online interactions between the agent and environment. While this can be validated in T1D simulators, it becomes impractical in real-world clinical settings. To this end, we propose an offline DRL framework that can develop and validate models for basal insulin control entirely offline. It comprises a DRL model based on the twin delayed deep deterministic policy gradient and behavior cloning, as well as off-policy evaluation (OPE) using fitted Q evaluation. We evaluated the proposed framework on an in silico dataset generated by the UVA/Padova T1D simulator, and the OhioT1DM dataset, a real clinical dataset. The performance on the in silico dataset shows that the offline DRL algorithm significantly increased time in range while reducing time below range and time above range for both adult and adolescent groups. Then, we used the OPE to estimate model performance on the clinical dataset, where a notable increase in policy values was observed for each subject. The results demonstrate that the proposed framework is a viable and safe method for improving personalized basal insulin control in T1D.

Time series data generated by continuous glucose monitoring sensors offer unparalleled opportunities for developing data-driven approaches, especially deep learning-based models, in diabetes management. Although these approaches have achieved state-of-the-art performance in various fields such as glucose prediction in type 1 diabetes (T1D), challenges remain in the acquisition of large-scale individual data for personalized modeling due to the elevated cost of clinical trials and data privacy regulations. In this work, we introduce GluGAN, a framework specifically designed for generating personalized glucose time series based on generative adversarial networks (GANs). Employing recurrent neural network (RNN) modules, the proposed framework uses a combination of unsupervised and supervised training to learn temporal dynamics in latent spaces. Aiming to assess the quality of synthetic data, we apply clinical metrics, distance scores, and discriminative and predictive scores computed by post-hoc RNNs in evaluation. Across three clinical datasets with 47 T1D subjects (including one publicly available and two proprietary datasets), GluGAN achieved better performance for all the considered metrics when compared with four baseline GAN models. The performance of data augmentation is evaluated by three machine learning-based glucose predictors. Using the training sets augmented by GluGAN significantly reduced the root mean square error for the predictors over 30 and 60-minute horizons. The results suggest that GluGAN is an effective method in generating high-quality synthetic glucose time series and has the potential to be used for evaluating the effectiveness of automated insulin delivery algorithms and as a digital twin to substitute for pre-clinical trials.

This paper introduces a linear weighted integrate-and-fire (I&F) neuron architecture for Ion-Sensitive Field-Effect Transistors (ISFETs). It contributes to the next generation of neuromorphic lab-on-chip (LoC) platforms with the aim to integrate electrochemical sensors with neural networks on-silicon to compensate for sensor non-idealities. The neuron consists of a readout circuit, a linear voltage-controlled weighting circuit, and a robust I&F circuit with low power consumption. Notably, the readout in the neuron circuit achieves linear conversion of input voltage to output current at low power. This work also presents a cluster architecture for spatial correlation of trapped charge effects and process variations. The calibration system integrated in each cluster is realized using a current bit cell chain inspired from the current mode algorithmic ADC. The compensation range for each pixel ranges from −157.7 mV to 128.1 mV. The system is implemented as a $25\times 20$ cluster array, and the sensitivity of each cluster is 10.92 kHz/pH.

Blood glucose prediction algorithms are key tools in the development of decision support systems and closed-loop insulin delivery systems for blood glucose control in diabetes. Deep learning models have provided leading results among machine learning algorithms to date in glucose prediction. However these models typically require large amounts of data to obtain best personalised glucose prediction results. Multitask learning facilitates an approach for leveraging data from multiple subjects while still learning accurate personalised models. In this work we present results comparing the effectiveness of multitask learning over sequential transfer learning, and learning only on subject-specific data with neural network and support vector regression. The multitask learning approach shows consistent leading performance in predictive metrics at both short-term and long-term prediction horizons. We obtain a predictive accuracy (RMSE) of 18.8 ±2.3, 25.3 ±2.9, 31.8 ±3.9, 41.2 ±4.5, 47.2 ±4.6 mg/dL at 30, 45, 60, 90, and 120 min prediction horizons respectively, with at least 93% clinically acceptable predictions using the Clarke Error Grid (EGA) at each prediction horizon. We also identify relevant prior information such as glycaemic variability that can be incorporated to improve predictive performance at long-term prediction horizons. Furthermore, we show consistent performance - ≤ 5% change in both RMSE and EGA (Zone A) - in rare cases of adverse glycaemic events with 1-6 weeks of training data. In conclusion, a multitask approach can allow for deploying personalised models even with significantly less subject-specific data without compromising performance.

A wireless power management unit is presented in this work. The system size of 0.4mm2with the low power consumption of 87μW allow it to be applied in implantable biomedical applications and ex-vivo systems. The proposed system optimises the conventional slew-rate detection low dropout (LDO) circuit by using a self-tuned controlling block that reduces the response time to 100ns. The static power required for this LDO is reduced to 2μW. The band gap reference (BGR) is designed with a high power supply rejection ratio (PSRR) of 60dB at 100kHz, along with an output reference voltage deviation of 1mV, to deal with the power supply ripple caused by the load shift keying (LSK). The RF power is transmitted at 433MHz and the internal power management circuit is able to provide a 1.4V stable DC supply for a front end on-chip sensor.

This paper presents a novel ISFET architecture with in-pixel ADC for large-scale integration and on-chip computational capabilities. Each pixel is composed by a comparator connected to a set of memory elements, storing in-pixel each conversion. Using this sensing scheme, the entire frame of ISFET pixels is acquired and stored in one single ADC cycle, eliminating the need for global or column-level ADCs and making this architecture scalable to larger arrays without instrumentation overhead. To enable compensation of ISFET non-idealities such as trapped charge, a novel gate-bootstrapping mechanism is introduced, enhancing the ADC dynamic range without additional circuitry and accommodating a trapped charge compensation range of 4.45 V. Fabricated in standard 180nm CMOS technology, the system is composed by a 16x16 ISFET array, a global DAC and a Digital Control Unit that enables operation and off-chip communication. The entire frame is acquired in 51.2 μs with a pixel power consumption of 10.15 μW, while storing the result in-pixel and eliminating the need for off-chip memories.

This paper presents a multi-sensing pixel capable of sensing ion concentration, temperature, moisture and light from the same spatial point. The presented pixel is composed of an ISFET chemical sensor, a light sensor, a moisture sensor and a temperature sensor, all implemented in standard CMOS technology in a compact pixel. All sensing modalities are acquired simultaneously and modulated onto a single waveform using multi-frequency PWM modulation, overcoming the associated transmission overhead. Furthermore, ISFET non-idealities such as trapped charge are compensated through a global DAC, balancing the input linear OTA, while the light sensor dark current is mitigated in-pixel by subtracting its influence using a reference photodiode. The proposed pixel occupies 50×50 μm2, and it is integrated as part of a 4×4 multi-sensing array, along with on-chip biasing, instrumentation and digital control. To the best of our knowledge, this architecture represents the first multi- sensing architecture capable of extracting four features from the same spatial point simultaneously, enabling ISFET-based platforms to sense both the chemical signal and the external conditions influencing its measurement.

This paper presents a USB 3.0 based readout platform for real-time ion imaging applications. The front end utilizes an ion-imaging array containing 16 k ISFET pixels to capture ion diffusion at 6100 fps. Operating at 200 MHz, the chip streams ion information at a data rate of 762.94 Mb/s. The backend readout system employs a FIFO-to-USB bridge (FT601Q) that supports Super Speed (5Gbps) to perform real-time data streaming and operates in bulk transfer mode to ensure data integrity. Implemented on a Xilinx Virtex UltraScale+ FPGA, the system involves a high-throughput data path through the on-board DDR4 SDRAMs, based on which a ring buffer is designed to provide the 2 GB buffering capacity. The readout system can operate in real time regardless of the USB glitches encountered on the Windows OS when involving data polling. In addition, a simple differencing algorithm with threshold detection is integrated into the back end for dynamic frame rate operation. The proposed readout system performs real-time ion imaging at high speed as well as streams the collected images for visualization within a latency of 25 ms, achieving state-of-the-art performance for high-speed Lab-on-Chip applications.

This work describes the implementation of SPACEMan, a wireless electrochemical system with concurrent potentiometric and amperometric sensing that can be utilised for saliva, sweat or point of care diagnostics. This system is designed with the vision of simpler interfaces for biofluid analysis. With a complete system-on-chip including electrochemical sensing, power management and data transmission, conventional interfaces like wirebonds will no longer be required in post-processing steps. The proposed architecture consists of a sensor front-end with four electrodes for concurrent amperometric and potentiometric sensing. This front-end outputs square wave signals mixed together with varying frequencies dependent on the sensed input, with the output type switchable with a state machine. A power management system consisting of a low dropout regulator (LDO) band gap reference (BGR), and a rectifier bridge is utilised for supplying power from an inductive link at 433MHz. Sensor data is transmitted wirelessly to a base station using LSK (Load-Shift Keying). The sensor front-end consumes 18μW, which the power management system more than adequately provides. The core area of the electronics without the coil is a conservative size of 0.41mm2.

In order to normalise measured concentrations of biomarkers in sweat, many of which are sweat rate dependent, a multi-modal analysis which measures flow rate alongside biomarker concentration is required. This paper presents an overview of current methods, as well as a detailed design of a calorimetric system to measure the flow rate of sweat, including the fluidic channel, signal conditioning circuits, and power regulation. Through simulation of the full system using LTSpice and SOLIDWORKS FlowSimulation, the final design achieved a minimum resolution of 0.85 pL/min, as well as an average error of 1.87% across the 0.5 to 20 pL/min measurement range.

This paper presents a 4 channel ASIC for sEMG sensing with in-built muscle fatigue and activity feature extraction. Each channel filters and conditions the electrode signal in parallel, while extracting key features for Low Back Pain (LBP) fatigue monitoring and forecasting - Zero Crossing rate and Root Mean Square through sEMG Envelope. The channels are integrated with a Transimpedance Amplifier, an 10-Bit ADC and a Digital Control Unit to digitise and enable transmission of extracted features. Fabricated in TSMC 180nm, these channels present a compact form factor (90μm× 630μm,) and a low power consumption (42.61 μw), ideal characteristic for wearable devices utilised for long-term monitoring of activities.

We have developed a CMOS System-on-Chip capable of concurrent DNA detection and actuation towards ultrafast and accurate molecular diagnostics. The system relies on a sensor array to perform electrochemical imaging by combining potentiometric ion-sensitive field-effect transistor (ISFET) sensing with impedance spectroscopy to concurrently detect DNA molecules and protons generated during an amplification reaction at high resolution. The impedance sensing electrodes are also used to conduct DNA manipulation using dielectrophoresis (DEP), further improving performance. The system achieves a maximum electric field strength of 2.5MV/m, generated with an AC voltage input of 1.8V, which is suitable for a battery supplied portable device. The sensing system made of EIS and ISFET front-ends are implemented in 0.18 μm CMOS technology. The ISFET sensing system presents a readout sensitivity of up to 129 mV (with a gain of 15 dB) with programmable controlled gate voltage to compensate non-idealities. Lastly, the EIS system is able to detect an impedance up to 100 MO within a frequency range between 100Hz-100kHz, presenting a dynamic range of 84.1 dB. The integrated systems achieve dual sensing and actuation for electrochemical DNA sensing with improved performance.

This paper presents a novel Lab-on-Chip ion imaging platform with programmable gain and array-level offset compensation. An array of 128 × 128 ISFET pixels are employed as the sensing front end, followed by a two-step column parallel readout circuit. The offset introduced by trapped charge and drift before any chemical event, is stored and fed back to the programmable gain instrumentation amplifier for compensation and signal amplification. A column-parallel 8-bit single slope ADC and 8-bit R-2R DAC are designed to achieve real-time array-level correlated double sampling, which also enables new possibilities to maximise sensitivity using off-chip image processing techniques. The system operates in real-time at a frame rate of up to 1000fps, with a maximum effective sensitivity of 1V/pH. Designed in TSMC 0.18 BCD process, the chip occupies a die area of 2.3 mm × 4.5 mm. We anticipate that this work would become a next generation solution for revealing inperceptible ion interactions in various biomedical applications.

Real-time blood glucose (BG) prediction can enhance decision support systems for insulin dosing such as bolus calculators and closed-loop systems for insulin delivery. Deep learning has been proven to achieve state-of-the-art performance in BG prediction. However, it is usually seen as a very computationally expensive approach, hence difficult to implement in wearable medical devices such as transmitters in continuous glucose monitoring (CGM) systems. In this work, we introduce a novel deep learning framework to predict BG levels with the edge inference on a microcontroller unit embedded in a low- power system. By using glucose measurements from a CGM sensor and a recurrent neural network that builds on long-short term memory, the personalized models achieves state-of-the-art performance on a clinical data set obtained from 12 subjects with T1D. In particular, the proposed method achieves an average root mean square error of 19.10 ± 2.04 for a 30-minute prediction horizon (PH) and 32.61 ± 3.45 for a 60-minute PH with high clinical accuracy. Notably, the framework has been optimized to achieve a minimum use of hardware resources (34KB FLASH and 1KB SRAM) as well as an execution time of 22 ms for low power operations (8 μW). The presented system has the potential to be implemented in wearable medical devices for diabetes care (CGM and insulin pumps) and to be integrated within an Internet of Things platform.

People with Type 1 diabetes (T1D) require regular exogenous infusion of insulin to maintain their blood glucose concentration in a therapeutically adequate target range. Although the artificial pancreas and continuous glucose monitoring have been proven to be effective in achieving closed-loop control, significant challenges still remain due to the high complexity of glucose dynamics and limitations in the technology. In this work, we propose a novel deep reinforcement learning model for single-hormone (insulin) and dual-hormone (insulin and glucagon) delivery. In particular, the delivery strategies are developed by double Q-learning with dilated recurrent neural networks. For designing and testing purposes, the FDA-accepted UVA/Padova Type 1 simulator was employed. First, we performed long-term generalized training to obtain a population model. Then, this model was personalized with a small data-set of subject-specific data. In silico results show that the single and dual-hormone delivery strategies achieve good glucose control when compared to a standard basal-bolus therapy with low-glucose insulin suspension. Specifically, in the adult cohort (n = 10), percentage time in target range 70, 180 mg/dL improved from 77.6% to 80.9% with single-hormone control, and to 85.6% with dual-hormone control. In the adolescent cohort (n = 10), percentage time in target range improved from 55.5% to [Formula: see text] with single-hormone control, and to 78.8% with dual-hormone control. In all scenarios, a significant decrease in hypoglycemia was observed. These results show that the use of deep reinforcement learning is a viable approach for closed-loop glucose control in T1D.

Diabetes is a chronic metabolic disorder that affects an estimated 463 million people worldwide. Aiming to improve the treatment of people with diabetes, digital health has been widely adopted in recent years and generated a huge amount of data that could be used for further management of this chronic disease. Taking advantage of this, approaches that use artificial intelligence and specifically deep learning, an emerging type of machine learning, have been widely adopted with promising results. In this paper, we present a comprehensive review of the applications of deep learning within the field of diabetes. We conducted a systematic literature search and identified three main areas that use this approach: diagnosis of diabetes, glucose management, and diagnosis of diabetes-related complications. The search resulted in the selection of 40 original research articles, of which we have summarized the key information about the employed learning models, development process, main outcomes, and baseline methods for performance evaluation. Among the analyzed literature, it is to be noted that various deep learning techniques and frameworks have achieved state-of-the-art performance in many diabetes-related tasks by outperforming conventional machine learning approaches. Meanwhile, we identify some limitations in the current literature, such as a lack of data availability and model interpretability. The rapid developments in deep learning and the increase in available data offer the possibility to meet these challenges in the near future and allow the widespread deployment of this technology in clinical settings.

This paper presents a method for detection of ESR1 p.E380Q, a common Breast Cancer (BC) mutation, using an ISFET (Ion-Sensitive Field-Effect Transistor) based Lab-on-Chip (LoC) platform. The LoC contains an ISFET array that can detect pH changes during DNA amplification, specifically Loop-Mediated Isothermal Amplification (LAMP). Synthetic ESR1 DNA was detected in a comparison pH-LAMP assay, carried out on the LoC platform as well as a conventional qPCR instrument. Positive detection of the allele arises due to bespoke allele-specific primers that target one base-pair difference between the wild-type and mutant alleles. The LoC and qPCR demonstrate comparable results detecting the mutant allele with mutant primers in around 25 minutes. The sensing microchip technology coupled with the molecular methods of isothermal chemistries and primer design allow this platform to be tested at a Point-of-Care setting for breast cancer patients, offering mutational tracking platform of circulating tumour DNA in liquid biopsies to assist patient stratification and allow tailored treatments.

Primers carry unique genetic information that allows them to be used as specific probes in various applications in the field of diagnostics and particularly in cancer, where the need for accurate treatment selection is crucial. Rapid and affordable detection of cancer specific targets, such as allele-specific single-nucleotide mutational changes, are of great need to improve treatment efficacy and guide clinical use. Detecting these mutations in isothermal conditions creates the opportunity to develop cost-efficient diagnostic platforms for cancer treatment. In this paper, a novel python script was constructed to develop a simple and personalised software tool that could automate the design of allele-specific primers in isothermal conditions using design parameters, such as free energies and annealing temperatures. The scripts' primers were compared to manually designed probes that were experimentally tested on a variant (ESR1 p.E380Q), commonly present in metastatic breast cancer, showcasing the applicability of the method and the potential for the script to be used as part of an automated software to design allele-specific assays for Lab-on-Chip platforms in cancer diagnostics.

This paper presents an Ion-to-Frequency ISFET architecture capable of operating at low power supplies, targeting fully-digital ultra-low power applications. The ISFET, biased in weak inversion, encodes the pH concentration as a frequency modulated digital signal by controlling the polarity and discharge rate of a capacitor. This architecture is implemented using digital gates operating at low voltages, achieving minimal power consumption and addressing the need for scalability to deep sub-micron technologies. Implemented in TSMC 0.18μm standard CMOS technology, each pixel occupies 20 μm × 21 μm and consumes a maximum of 33 pW operated at 0.2V while preserving a large sensitivity of 287 Hz/pH at a center frequency of 697Hz. Simulation results indicate low power consumption with a compact pixel size, becoming an efficient solution for the next-gen of portable and wearable applications.

A multi-sensor pixel is presented using ion-sensitive field-effect transistors (ISFETs) as ion sensors, photodiodes as optical sensors and MOSFETs as temperature sensors. The pixel is inspired from an Active Pixel Sensor (APS) topology previously reported as an ISFET front-end. Current mirrors and switches are used to multiplex sensors and encode the output in the time domain for in-pixel quantisation. A novel temperature compensation method, based on temperature coefficient cancellation, is implemented based on bandgap and translinear circuits. The non-idealities of ISFETs, such as drift and trapped charge, are compensated by source voltage modulation. The pixel is implemented using TSMC 0.18 μm CMOS technology and occupies a 38 μm × 36 μm area with an ultra-low power consumption of 33.96nW associated with weak inversion operation. The simulated results demonstrate a high pH sensitivity of 33.96 ns/pH and an ultra-stable temperature variation between 63 fA/°C and 2.3 pA/°C.

The ion-sensitive field-effect transistor (ISFET) has emerged as an ideal candidate to carry out point-of-care diagnosis through detection of hydrogen ions produced during amplification of pathogenic DNA. This group has previously hypothesised that ISFET-based DNA detection could be enhanced by positioning DNA close to the ISFET surface using electrodes that carry out low-power manipulation of DNA through dielectrophoresis (DEP). This paper provides a proof-of-concept for an ISFET-DEP system for DNA detection, combining information gathered from the literature and FEM electric field simulations to test the validity of the hypothesis and highlight key challenges in implementation. An electrode system was designed with sets of line and rectangular castellated interdigitated electrodes that produced simulated maximum electric field strengths between 2.58 - 7.70 × 105Vm-1and is capable of trapping DNA with applied voltages as low as (0.50 ± 0.05)V. Using this design, a 180nm CMOS prototype was developed for preliminary testing, containing three 5 × 2 arrays of 20μm × 20μm ISFETs as part of an 800μm × 400μm chip.

This paper demonstrates a novel readout platform for ISFET-based ion imagers which is capable of performing high-throughput data acquisition and real-time monitoring on high-speed chemical reactions. The front end employs a 128×128 array of integrated ISFET pH sensors fabricated in unmodified CMOS process. The array operates at a frame rate of up to 3000 fps for detecting hydrogen ion diffusion, generating a maximum data stream of 491.52 Mbps. A digital readout system consisting of a readout module for data buffering, an AXI master controller for accessing on-board DDR3 memory and a PCIe subsystem for transmitting data packets is implemented on an Alinx AX7103 development board to link the chip and the PC. The platform capabilities are demonstrated with a real-time ion imaging experiment, by observing the diffusion of NaOH pills in water within 320 ms, visualized on screen with a latency of 0.15 s. Lastly, different image processing algorithms including Gaussian, Bilateral and Non-local Mean are evaluated for noise reduction and an accelerator for the optimum filter is implemented for real-time ion-imaging.

DAPPER is a front end system capable of simultaneous amperometric and potentiometric sensing proposed for low-power multi-parameter analysis of bio-fluids such as saliva. The system consists of two oscillator circuits, generating a frequency relative to their sensed current and voltage signals. These signals are then mixed together to produce a single channel output that can be transmitted through backscattering (load-shift keying). The entire system consumes 40μW from a 1.4V supply. The linear ranges of potentiometry and amperometry circuits are 0.4V - 1V and 250pA - 5.6μA (87dB), and their input referred noise is 1.7μV and 44.6fA, respectively.

We design an Ion-Sensitive Field-Effect (ISFET) array leveraging on the two successful fields of neuromorphic electronics and chemical sensing to encode the signal in spikes and perform sensor processing between neighbouring pixels. The array is structured as clusters integrating 4 × 4 pixels with sensor compensation, taking advantage of spatial correlation of sensor non-idealities. The offset compensation is capable of calibrating in a range of 662 mV. The system shows a robust, scalable and power efficient architecture with a sensitivity ranging from 2.56 MHz/pH to 3.38 MHz/pH. The pixel occupies an area of 30μm × 24μm, and the cluster area is 205 μm × 205 μm. The layout of each pixel is spread out with digital blocks embedded in-between, which improves signal coupling by enlarging the chemical sensing area of each pixel. The system readout implements address event representation (AER) for triggering the outputs.

Widening the range of targets for ion-sensitive field-effect transistors (ISFETs) fabricated in unmodified CMOS technology has been enabled by the deposition of polymeric ionophore membranes at the surface, requiring specific sensor training. We present a novel ISFET array with on-chip multiplexing capabilities to perform offline training and real-time sensing on a single substrate, enabling analogue averaging for low noise sensing and reducing post-processing. The analogue front-end pixels rely on a modular current-mode spatial averaging (CMSA) circuit, producing an averaged sensor output per cluster based on a switching matrix integrated within the array. At the output stage, a weak inversion active resistor implements an ELIN system to guarantee linearity, reaching an expected sensitivity of 160 mV/pH with a gain of 3, when designed using a 0.18 μm standard CMOS technology. Based on calibration data for the ion targets, an on-chip training algorithm determines the sensitivity of each pixel to each ion, identifies sensing regions corresponding to polymeric membranes at the surface and sets the pixel connectivity through the switching matrix.

In type 1 diabetes management, maintaining nocturnal blood glucose within target range can be challenging. Although semi-automatic systems to modulate insulin pump delivery, such as low-glucose insulin suspension and the artificial pancreas, are starting to become a reality, their elevated cost and performance below user expectations is hindering their adoption. Hence, a decision support system that helps people with type 1 diabetes, on multiple daily injections or insulin pump therapy, to avoid undesirable overnight blood glucose fluctuations (hyper- or hypoglycaemic) is an attractive alternative. In this paper, we introduce a novel data-driven approach to predict the quality of overnight glycaemic control in people with type 1 diabetes by analyzing commonly gathered data during the day-time period (continuous glucose monitoring data, meal intake and insulin boluses). The proposed approach is able to predict whether overnight blood glucose concentrations are going to remain within or outside the target range, and therefore allows the user to take the appropriate preventive action (snack or change in basal insulin). For this purpose, a number of popular established machine learning algorithms for binary classification were evaluated and compared on a publicly available clinical dataset (i.e., OhioT1DM). Although there is no clearly superior classification algorithm, this study indicates that, by using commonly gathered data in type 1 diabetes management, it is possible to predict the quality of overnight glycaemic control with reasonable accuracy (AUC-ROC = 0.7).

Control of blood glucose is essential for diabetes management. Current digital therapeutic approaches for subjects with type 1 diabetes mellitus such as the artificial pancreas and insulin bolus calculators leverage machine learning techniques for predicting subcutaneous glucose for improved control. Deep learning has recently been applied in healthcare and medical research to achieve state-of-the-art results in a range of tasks including disease diagnosis, and patient state prediction among others. In this paper, we present a deep learning model that is capable of forecasting glucose levels with leading accuracy for simulated patient cases (root-mean-square error (RMSE) = 9.38 ± 0.71 [mg/dL] over a 30-min horizon, RMSE = 18.87 ± 2.25 [mg/dL] over a 60-min horizon) and real patient cases (RMSE = 21.07 ± 2.35 [mg/dL] for 30 min, RMSE = 33.27 ± 4.79% for 60 min). In addition, the model provides competitive performance in providing effective prediction horizon ([Formula: see text]) with minimal time lag both in a simulated patient dataset ([Formula: see text] = 29.0 ± 0.7 for 30 min and [Formula: see text] = 49.8 ± 2.9 for 60 min) and in a real patient dataset ([Formula: see text] = 19.3 ± 3.1 for 30 min and [Formula: see text] = 29.3 ± 9.4 for 60 min). This approach is evaluated on a dataset of ten simulated cases generated from the UVA/Padova simulator and a clinical dataset of ten real cases each containing glucose readings, insulin bolus, and meal (carbohydrate) data. Performance of the recurrent convolutional neural network is benchmarked against four algorithms. The proposed algorithm is implemented on an Android mobile phone, with an execution time of 6 ms on a phone compared to an execution time of 780 ms on a laptop.

For people with Type 1 diabetes (T1D), forecasting of blood glucose (BG) can be used to effectively avoid hyperglycemia, hypoglycemia and associated complications. The latest continuous glucose monitoring (CGM) technology allows people to observe glucose in real-time. However, an accurate glucose forecast remains a challenge. In this work, we introduce GluNet, a framework that leverages on a personalized deep neural network to predict the probabilistic distribution of short-term (30-60 minutes) future CGM measurements for subjects with T1D based on their historical data including glucose measurements, meal information, insulin doses, and other factors. It adopts the latest deep learning techniques consisting of four components: data pre-processing, label transform/recover, multi-layers of dilated convolution neural network (CNN), and post-processing. The method is evaluated in-silico for both adult and adolescent subjects. The results show significant improvements over existing methods in the literature through a comprehensive comparison in terms of root mean square error (RMSE) ([Formula: see text] mg/dL) with short time lag ([Formula: see text] minutes) for prediction horizons (PH) = 30 mins (minutes), and RMSE ([Formula: see text] mg/dL) with time lag ([Formula: see text] mins) for PH = 60 mins for virtual adult subjects. In addition, GluNet is also tested on two clinical data sets. Results show that it achieves an RMSE ([Formula: see text] mg/dL) with time lag ([Formula: see text] mins) for PH = 30 mins and an RMSE ([Formula: see text] mg/dL) with time lag ([Formula: see text] mins) for PH = 60 mins. These are the best reported results for glucose forecasting when compared with other methods including the neural network for predicting glucose (NNPG), the support vector regression (SVR), the latent variable with exogenous input (LVX), and the auto regression with exogenous input (ARX) algorithm.

This paper presents a novel 32×32 ISFET array for DNA amplification detection. Each pixel contains a highly-linear ISFET-based OTA and a sawtooth oscillator, converting the solution pH into a digital clock with chemically controlled duty cycle. By employing a differential measurement between a DAC-generated voltage and the pH solution on the OTA, a highly linear, PVT-insensitive response is achieved, while opening the possibility of real-time pixel-wise compensation of trapped charge and chemical drift. The proposed architecture achieves a sensitivity of 11.78%/pH while maintaining large linear dynamic range and a temperature sensitivity of 0.0033 pH/K. Implemented using 0.18μm standard CMOS technology, each pixel occupies 40 μm × 40 μm. This architecture paves the way towards a new generation of ISFET imagers, capable of learning from data and correcting their measurements in real time.

This paper introduces the use of bioelectronic medicine for glucose control in Type 1 diabetes. In particular, we present a new hybrid closed-loop glucose controller that regulates (i) the insulin and glucagon doses delivered using a pump and (ii) the value of insulin sensitivity of the patient, which would be modulated through electrical stimulation of the nervous system. The presented controller achieves improved glucose control with increased percentage of time of glucose levels within target and decreased hormonal delivery when compared with conventional glucose controllers. This work shows the potential of using bioelectronic modulation of insulin sensitivity for diabetes management.

A portable and low-cost system capable of high-speed ion-imaging using ISFET arrays will be demonstrated. The system takes advantage of the portability and processing power of Raspberry Pi, enabling high-speed measurements from a large-scale ISFET array. This array contains 64×200 ISFET pixels and is connected to a custom PCB that provides analogue-to-digital conversion as well as configuration signals. This platform offers a complete solution allowing ion-imaging and pH observations outside of the lab environment. We anticipate that this efficient approach holds significant potential for affordable ion-imaging systems, making them more readily available for further research.

This paper presents a 32×32 ISFET sensing array with in-pixel digitisation for pH sensing. The in-pixel digitisation is achieved using an inverter-based sensing pixel that is controlled by a triangular waveform. This converts the pH response of the ISFET into a time-domain signal whilst also increasing dynamic range and thus the ability to tolerate sensor offset. The pixels are interfaced to a 15-bit asynchronous column-wise time-to-digital converter (TDC), enabling fast sensor readout whilst using minimal silicon area. Parallel output of 32 TDC interfaces are serialised to achieve fast data though-put. This system is implemented in a standard 0.18 μm standard CMOS technology, with a pixel size of 26 μm × 26 μm and a TDC of 26 μm × 180 μm. Simulation results demonstrate that chemical sampling of up to 5k frames per second can be achieved with a clock frequency of 160 MHz and a TDC resolution of 190 ps. The total power consumption of the overall system is 7.34 mW.

In this paper, we show a compensation method for mismatch in large-scale ISFET arrays which is caused by the presence of trapped charge at the sensor's floating gate. To facilitate ISFET calibration, the Programmable-Gate method is used. We improve on a previously proposed gradient descent algorithm by making an a priori estimate of the calibration step size thus allowing to reduce the number of iterations to one. This is enabled by an initial characterisation of the programmable gate capacitor in order to determine the effect of the calibration signal on the pixel's output. Measured results of both approaches are presented using a 64×200 ISFET array with a parasitic PG capacitor located vertically inside the pixel stack such that pixel area is not compromised. Additionally, results are shown for two chips which correspond to different trapped charge spreads with both algorithms reducing the standard deviation of the trapped charge by an average of 78% and 66% respectively.

We present the next step of neuromorphic ISFET arrays with spike domain encoding and spatial device compensation. Each pixel provides a spiking signal with a frequency related to the pH in solution and expected sensitivity of 48.6 to 112.2 kHz/dpH. The array is arranged as clusters which use regulation to cancel undesirable sensor offset and then linear interpolation for temporal drift during the readout. The scheme relies on spatial correlation of ISFET behaviour which is demonstrated with a low standard deviation of 11.6 mV sensor offset, which is well in the pixel compensation range of ± 500 mV. On an array level, address-event representation is used for external signal handling, which enables low power and scalable throughput. The array chip is implemented in TSMC 0.18 μm CMOS technology.

This paper presents a fully integrated system-on-chip for ultra-high frame rate ion-imaging using a pH-sensing ISFET array. Linear pH-to-current conversion is achieved by operating the ISFET in velocity saturation which guarantees that the ion concentration in the chemical solution is linearly transduced to the output of the sensor. Implemented in a 3-Transistor (3-T) pixel for compactness, the ISFET also consists of a reset switch to compensate for sensor non-ideal effects such as trapped charge and drift. High speed readout is achieved using a current-mode signal processing pipeline while auto-zeroing is employed to reduce fixed pattern noise. The sensing array comprises 128 × 128 pixels and every row shares its own readout circuit followed by 128 row-parallel 1MS/sec single slope ADCs. Designed in standard TSMC 180nm CMOS process, the chip achieves 7800fps with 16k pixels and a silicon area of 2mm × 2mm, which is the fastest ISFET array reported in literature.

We present a novel System-on-Chip (SoC) design for DNA amplification and detection with the ability to electrostatically manipulate DNA molecules. The chip integrates an array of ISFET sensors for pH monitoring with heaters for temperature regulation, which enables amplification methods requiring thermal cycling or constant temperature. The sensing array comprises 1129 pixels, with each pixel based on an ISFET front-end with a programmable gate and implemented in a unity gain buffer configuration. The pH readout achieves a sensitivity of 196.5mV/pH with a gain of 30 and a frame rate of 80.5 fps. A temperature characterisation demonstrates that the system is capable of raising the temperature of a 10μL chemical solution to 96.44° C. Lastly, the on-chip interdigitated electrodes use a dielectrophoresis (DEP) technique to generate an AC sinusoidal electric field over a frequency range of 100Hz-1MHz with a simulated amplitude of approximately 100kVm-1to orient and immobilise DNA molecules at the ISFET sensing interface.

A mobile diagnostic system is demonstrated for the early detection of infectious disease outbreaks in remote areas. The system comprises an ISFET-based platform, an AndroidOS application running on a smartphone, and a cloud server. Incorporation of microfluidics on the 78×56 ISFET array permits on-board isothermal DNA amplification and detection. Each die is mounted on single-use cartridges. The platform is controlled by the app, which collects relevant data via Bluetooth to process through algorithms stored on the smartphone. Upon a positive result, a data package containing disease type, geographical location, and timestamp is sent to the cloud. Real-time monitoring of outbreaks to pandemics can be visualized accordingly.

This paper presents a portable low-power platform that aims to maintain glycaemic control in people with type 1 diabetes. It runs on dedicated hardware, and incorporates a bio-inspired controller that replicates the insulin-secreting physiology of the pancreatic beta cells. The controller runs on an embedded microchip and the system has been optimized for low-power operation. The low-power, hand-held unit (BiAP) interfaces to a continuous glucose sensor (Dexcom G5) and sends the required insulin dose to a subcutaneous pump (Tandem t:slim). An adaptive meal bolus calculator running on an iPhone is used to improve the controller. The system architecture is presented in this paper along with the results of a 4 day hardware-in-the-loop test. The results show that the BiAP platform is robust and capable of achieving good control with no missed glucose values or device disconnections. Finally, the hand-held unit achieves power efficient operation, lasting for 6 days during full control.

A novel 96-channel ASIC for simultaneous recording of sEMG signals for muscle fatigue monitoring is presented. Each channel is fully differential such that information regarding the median frequency (MDF) of the sEMG signal can be extracted which has been shown to correlate with muscle fatigue. The analogue front-end of each channel includes a tunable band pass filter and a programmable variable gain, consumes 8.4μW and occupies an area of 0.034mm2. Additionally, the ASIC incorporates compressed sensing (CS) to reduce the data transmitted from all channels into rates achievable using Bluetooth technology. Each on-chip CS module occupies 0.0085mm2and consumes 17μW per channel with a compression ratio of 10. The reconstructed signal achieves a SNRrecon of 10.2dB and a MARD of 1.7% for the MDF of the sEMG signal, validated using real sEMG recordings. The system architecture is designed in a modular way to ensure scalability for 100+ channels. As such, the system comprises 4 independent readout modules of 24-channels each which include a dedicated multiplexer, ADC and a compressed sensing module for each. The full ASIC is designed in a 0.35 μm CMOS process, occupies an area of 5.82mm2and has a total power consumption of 3.93mW.

This paper introduces the use of a diode-connected MOSFET as a temperature controlled switch for trapped charge cancellation of ISFET sensors. The current flowing through the reverse-biased diodes of 2.7 aA is negligible at low temperature but reaches 23 fA when heated to 100°C, which allows for low temperature readout and high temperature compensation of trapped charge. The diode-connected device is tied to the floating gate of the ISFET which is integrated as part of a source-follower readout. The in-pixel feedback loop uses a comparator to turn off the polysilicon heaters once the offset has been cancelled, triggering the readout by switching the op amp into a buffer configuration. Simulations show that the system is calibrated in 35s and the output sensitivity reaches 27.74 mV/pH. The pixel output then tracks any ionic change at its surface without sensor offset.

This paper presents a 128 × 128 ISFET array with current mode readout peripherals for real-time ion imaging. Current-mode operation is employed to achieve very high speed and frame rate and provide a linear mapping between the ion concentration at the sensing layer (typically hydrogen ions - pH) to the drain current of the device. To this effect, a single device biased in the triode region can serve as both the sensing and readout device in the pixel ensuring a very small area footprint per pixel. Compensating for known non-ideal effects of the ISFET, namely trapped charge and drift, is implemented by resetting the gate voltage whereas any additional circuit offsets are eliminated by auto-zeroing. Auto-zeroing and sampling takes place on a row-parallel basis which is then multiplexed to 8 current mode ADCs. The chip is designed in a standard 0.35um CMOS process, occupies an area of 2.6mm × 2.2mm and can achieve a frame rate of 3000 fps which is the highest in this process node. We anticipate that the proposed system will increase the current temporal limit of detection of chemical reactions and provide new insight into real-time ion dynamics.

We demonstrate a CMOS-based lab-on-chip platform which is capable of ion imaging to detect a variation in hydrogen, potassium and sodium ions. An ISFET array is used to detect a change in ion concentration with a calibration scheme to cancel the offset due to trapped charge. The Si3N4passivation layer confers an inherent sensitivity to the sensors, and additional polymer membranes are pipetted containing a potassium and sodium ionophore. An initial algorithm identifies the sensitivity of each pixel towards the target ions. The user can inject a solution with a given concentration of ions and observe the real-time output change of the array on a MATLAB interface. The display then provides an estimate of the target ion concentration.

We demonstrate a diagnostics platform which integrates an ISFET array and a temperature control loop for isothermal DNA detection. The controller maintains a temperature of 63°C to perform nucleic acid amplification which is detected by the on-chip sensors. The 32×32 ISFET array is first calibrated to cancel trapped charge and then measures the change in the pH of the reaction. The sensor data is sent to a microcontroller and the reaction is monitored in real-time using a MATLAB interface. Experiments confirm a change of 0.9 pH when tested for the presence of RNA associated with the Zika virus.

This paper presents a novel ISFET sensor architecture which uses Through-Silicon Vias (TSV) available in a standard CMOS process to serve as both the sensing surface to detect the release of ions and the reaction well which is necessary for DNA sequencing. Using TSVs ensures that no post-fabrication steps are needed to deposit the wells, on top of offering a very large sensing area compared to typical planar sensing areas. Due to its irregular geometry, we employ a finite element method to model the TSV which renders a very large equivalent passivation capacitance (Cpass = 4.06pf) as a consequence of its large sensing area and 3D structure. Therefore, we show how two typical ISFET pixels for ion sensing can be configured to use TSVs, offering better performance in terms of noise and input signal attenuation.

A fully-integrated differential biosensing platform on CMOS is presented for miniaturized enzyme-based electrochemical sensing. It enables sensor background current elimination and consists of a differential sensor array and a differential readout IC (DiRIC). The sensor array includes a four-electrode sensor for amperometric electrochemical sensing, as well as a differential ISFET-based pH sensor to calibrate the biosensors. The ISFET is biased in weak inversion and co-designed with DiRIC to enable pH measurements from 1 to 14 with resolution of 0.1 pH. DiRIC enables differential current measurement in the range of ±100 μA with more than 120dB dynamic range.

Type 1 diabetes mellitus (T1DM) is an autoimmune disease characterised by elevated blood glucose levels. Good glucose control is critical to reduce the incidence of complications and improving quality of life for people with T1DM. The Bio-inspired Artificial Pancreas (BiAP), a handheld closed-loop insulin delivery system based on beta-cell physiology, has been developed to automatically control blood glucose in people with T1DM. One of the biggest challenges when controlling blood glucose levels in T1DM is the ingestion of meals. We demonstrate a hardware-in-the-loop platform that allows announcing various meals through an iPad application to the BiAP system replicating real-time feeding. For this purpose, a mixed-meal model library was developed to challenge the BiAP controller using an in silico simulator.

This paper presents a compact and scalable architecture design of a multi-channel potentiostat. Utilizing a hybrid-multiplexed technique, the system is capable of driving multi-electrode array structures of large sizes with few readout channels. A 5-channel potentiostat with 80-electrode capability was fabricated into a portable 8.382×9.906 cm2PCB prototype using discrete components. It features a dynamic current range of 126dB and 3.3V single-supply operation. Controlled by a MATLAB graphical user interface, the system demonstrates realtime data acquisition and achieves similar performance to a commercial potentiostat based on electrochemical validation.

We demonstrate a portable PCB prototype of a potentiostat capable of driving an 80-electrode sensor array with 5 readout channels. Controlled by a MATLAB graphical user interface, the system demonstrates real-time data acquisition and achieves similar performance to a commercial potentiostat based on electrochemical validation.

In this paper, we present a novel bio-inspired CMOS-based hexagonal ISFET sensor which, compared to traditional devices, exhibit a more compact arrangement as part of large arrays and provides benefits in terms of capacitive attenuation. We classify these novel sensors as Enclosed Gate Transistors (EGTs) and provide a layout in standard AMS 0.35 μm technology. Electrical characteristics are derived theoretically for the device, including an effective W/L, and simulations for both AMS 0.18 and 0.35 μm CMOS processes are provided. The results indicate a decrease in parasitic gate capacitance between 20 % and 40%, highlighting the advantages in attenuation of respectively 0.36 dB and 0.71 dB for the smallest lengths of devices. The noise performance is also improved, with the input referred noise reduced by 1 or 2 % for each process. The devices were fabricated in standard 0.35 μm CMOS technology for future characterisation.

This paper presents an ISFET-based front-end to detect DNA homopolymerisms in ion-semiconductor sequencing. This is realised using a novel ISFET-based switched current integrator which provides discrete output levels proportional to the number of nucleotides incorporated by integrating signals from DNA polymerase reactions in real time. The output response is maintained for noise comparable to the input signal at 0.01pH peak to peak. The proposed circuit has a gain of 2.4μA/pH, σ ± 50nA at 5Hz sampling frequency and 1μA bias current.

This paper investigates ISFET circuits operating in current mode and demonstrates linear pH-to-current operation. The two operating regions that provide linear gate-voltage to drain-current operation are the triode and velocity saturation regions which have been compared with a focus on pH sensing applications. Subsequently, two sensor arrays optimised for operation in each region are proposed with simulations supporting feasible operation in the current domain. Furthermore, the paper demonstrates a novel 1T-per-pixel array with the ISFET serving as both the sensing and readout element in the pixel.

In this paper, we present a novel approach to ISFET arrays which allows the conception of ion imaging Lab-on-CMOS platforms. K+ and Na+ selective polymer membranes are deposited on the surface of the array so that each pixel is selective to a particular ionic species. An initial calibration produces an accurate mapping of the array in terms of ion-selective regions and determines the sensitivity of the membrane. The system exhibits K+ and Na+ sensitivities of respectively 51.2 mV/dec and 46.8 mV/dec, and demonstrates good discrimination of Potassium and Sodium ions for a common solution exposed to the chip, with a reported error lower than 1%. This ISFET-based tri-ion imaging array constitutes the basis for a portable integrated multi-ion platform.

People with type 1 diabetes (T1D) rely on exogenous insulin in order to bring blood glucose levels back to healthy levels after a meal. The amount of bolus insulin needed depends on various parameters and recent research has been focused on providing decision support for calculating clinically safe insulin doses. In this demonstration we present a mobile `Advanced Bolus Calculator for Diabetes' for insulin dosing decision support and its operation and notification service through a connected smartwatch. The smartwatch can be used to display glucose information, log diabetes related events and request insulin dose recommendations.

Electromyography (EMG) analysis can provide useful Information about a muscle's fatigue state by estimating the travelling speed of Action Potentials in muscle tissue. Bit-stream cross-correlation is a reliable and low complexity method for speed estimation. This paper presents a comparison and circuit implementation between three different bit-stream modulation method s suitable for sEMG signals. The modulators are simulated and evaluated using real retrospective sEMG signals and compared with a Matlab cross-correlation function.

This paper presents an FPGA based DNA comparison platform which can be run concurrently with the sensing phase of DNA sequencing and shortens the overall time needed for de novo DNA assembly. A hybrid overlap searching algorithm is applied which is scalable and can deal with incremental detection of new bases. To handle the incomplete data set which gradually increases during sequencing time, all-against-all comparisons are broken down into successive window-against-window comparison phases and executed using a novel dynamic suffix comparison algorithm combined with a partitioned dynamic programming method. The complete system has been designed to facilitate parallel processing in hardware, which allows real-time comparison and full scalability as well as a decrease in the number of computations required. A base pair comparison rate of 51.2 G/s is achieved when implemented on an FPGA with successful DNA comparison when using data sets from real genomes.

The papers in this special section are devoted to the topic of ways in which biomedical technologies and informatics are being used in the management of diabetes. It is estimated that 371 million people worldwide have diabetes and the number is increasing at an alarming rate. In addition to short-term symptoms, there are long-term macro and microvascular complications including cardiovascular disease, especially heart attacks and strokes, kidney failure, diabetic foot disease leading to gangrene and amputation, and blindness.

This paper presents the architecture and initial usability results of an advanced insulin bolus calculator for diabetes (ABC4D), which provides personalized insulin recommendations for people with diabetes by differentiating between various diabetes scenarios and automatically adjusting its parameters over time. The proposed platform comprises two main components: a smartphone-based patient platform allowing manual input of glucose and variables affecting blood glucose levels (e.g., meal carbohydrate content and exercise) and providing real-time insulin bolus recommendations; and a clinical revision platform to supervise the automatic adaptations of the bolus calculator parameters. The system implements a previously in silico validated bolus calculator algorithm based on case-based reasoning, which uses information from similar past events (i.e., cases) to suggest improved personalized insulin bolus recommendations and automatically learns from new events. Usability of ABC4D was assessed by analyzing the system usage at the end of a six-week pilot study (n = 10). Further feedback on the use of ABC4D has been obtained from each participant at the end of the study from a usability questionnaire. On average, each participant requested 115 ± 21 insulin recommendations, of which 103 ± 28 (90%) were accepted. The clinical revision software proposed a total of 754 case revisions, where 723 (96%) adaptations were approved by a clinical expert and updated in the patient platform.

A novel pixel architecture is presented to be part of a pH-based DNA microarray using ISFETs as chemical sensors. The design, based on APS architectures, performs pulse width modulation to encode the pH in time. It allows compensation for a major ISFET nonideality, offset in the threshold voltage, which is caused by trapped charge on the floating gate and the passivation layer of the ISFET. Also, the drop in sensitivity due to the passivation capacitance inherent to CMOS processes is attenuated. The system is implemented using a 0.35 fim standard CMOS technology. The pixel is shown to achieve a high and tunable accuracy of between 0.55 μs/dpH and 2.9 μs/dpH. Along with an estimation of the noise and the incidence of calibration, this leads to a resolution of approximately 20 mpH on a 1 pH-range. Trapped charge compensation proves to be effective up to an offset voltage of 1.25 V. The pixel is compact and reaches a total area of 16.5 μm × 16.25 μm.

Miniaturized and integrated onto smart clothing sensors are more likely to be accepted by patients, thus enhancing healthcare in both hospital and home environments. Our research focuses on the development of a patient-centred, multi-sensing platform for a smart garment for knee functional monitoring capable of assessing a patient's activities in their daily environment, providing clinicians with objective markers of performance and assisting in devising customised rehabilitation strategies.

This paper presents an advanced insulin bolus advisor for people with diabetes on multiple daily injections or insulin pump therapy. The proposed system, which runs on a smartphone, keeps the simplicity of a standard bolus calculator while enhancing its performance by providing more adaptability and flexibility. This is achieved by means of applying a retrospective optimization of the insulin bolus therapy using a novel combination of run-to-run (R2R) that uses intermittent continuous glucose monitoring data, and case-based reasoning (CBR). The validity of the proposed approach has been proven by in-silico studies using the FDA-accepted UVa-Padova type 1 diabetes simulator. Tests under more realistic in-silico scenarios are achieved by updating the simulator to emulate intrasubject insulin sensitivity variations and uncertainty in the capillarity measurements and carbohydrate intake. The CBR(R2R) algorithm performed well in simulations by significantly reducing the mean blood glucose, increasing the time in euglycemia and completely eliminating hypoglycaemia. Finally, compared to an R2R stand-alone version of the algorithm, the CBR(R2R) algorithm performed better in both adults and adolescent populations, proving the benefit of the utilization of CBR. In particular, the mean blood glucose improved from 166 ± 39 to 150 ± 16 in the adult populations (p = 0.03) and from 167 ± 25 to 162 ± 23 in the adolescent population (p = 0.06). In addition, CBR(R2R) was able to completely eliminate hypoglycaemia, while the R2R alone was not able to do it in the adolescent population.

This paper presents an automatic calibration system for ISFET chemical sensing arrays to tune out any mismatch in sensitivity. Through exploitation of the high frequency spectrum of the sensed signal which does not contain any chemical reaction information, local sensitivity is examined and then calibrated through a successive approximation control mechanism and an 8-bit variable gain amplifier to provide high accuracy tuning. The total calibration process contains three phases, peak detection, AGC through a SAR and finally sensor readout of the detected chemical signal. Designed in a typical 0.35μm CMOS process, the system is capable of compensating a sensitivity deviation of up to 24% in an ISFET array, constraining the error to just 1.5%.

Electromyography analysis can provide information about a muscle's fatigue state. The Power Spectral Density (PSD) function of the EMG signal undergoes a progressive compression towards lower frequencies and change of shape during fatigue. These changes are best represented by the instantaneous median frequency (iMDF) of the myoelectric signal. This paper presents an analogue based system developed in CMOS that extracts muscle fatigue through iMDF estimation of the EMG signal. The complete circuit design was simulated and evaluated using real retrospective muscle fatigue data. Results show that the average error in iMDF estimation, assuming a maximum iMDF of 150 Hz, would be less than 6 Hz. The system has been implemented in a commercially available 0.35 μm CMOS technology requiring a total power consumption of 250 μW from a 3.3 V supply.

This paper identifies and addresses key design considerations and trade-offs in the implementation of a CMOS high-resolution microheater array for Lab-on-Chip (LOC) applications. Specifically, this is investigated in the context of facilitating the in vitro thermal stimulation of single neurons. The paper analyses the electro-thermal response (by means of COMSOL simulations) and reliability issues (such as melting and electromigration) of different microheater designs. The analysis shows that a small-area heater is more efficient in terms of power, but it has more reliability problems essentially due to electromigration effects. For the proposed heater designs, the expected lifetime is a few days (in continuous operation) in the worst scenario, which is still generally acceptable for LOC applications.

In this paper we propose a low-power, compact ISFET interface capable of overcoming problems of DC offset due to trapped charge and transcoductance reduction due to capacitive division, which commonly exist with implementation in CMOS. Through direct feedback to the floating gate and a low-leakage switching scheme, all the unwanted factors are eliminated while the output is capable of tracking a pH reaction which occurs at the sensing surface. The circuit has been designed to be an inherently constant-voltage-constant-current structure, avoiding any loss of sensitivity due to capacitive division. Additionally, a tradeoff between noise and stability is presented for optimum choice of capacitors. The proposed ISFET and interface has been designed in a typical 0.35 μm process, has an area 60 ×70um2and consumes 231nW of power, which facilitates it's implementation for large chemical sensing arrays.

This paper explores the potential of partitioning the dynamic programming algorithm to utilise the capabilities of FPGA platforms for parallel genome sequence comparison and assembly. We use this to solve the prefix-suffix approximate matching problem to find overlaps between DNA strands in a given sequence. This is achieved by partitioning the basic dynamic programming (DP) algorithm into a series of discrete sub-DP calculations. Analysis of the error rate as a result of this partitioning by applying random sequences as input data is shown, and simulation results confirm good matching with the original algorithm with an error of less that 1.5% in the worst case. Optimisation for array implementation in FPGA is shown and linear scalability for larger arrays is proven. Simulation results of the whole system confirm 98.8% similarity of the overlap adjacent matrix between error and error-free data.

This paper presents the implementation of a low-power analogue circuit, which replicates the granular release of insulin from beta-cell of the pancreas, to control the blood glucose. Results show that the circuit with a power consumption of 1.667 mW can achieve the same physiological responses with the designed model developed in Matlab. It is therefore expected that in a future implementation of the circuit in silicon, the same quality of blood glucose control can be achieved and therefore can be used as part of the Artificial Pancreas to support the treatment of Diabetes.

In-pixel differential measurement with calibration for chemical reaction occurrence detection, particularly for pH-based Single Nucleotide Polymorphism (SNP) detection and genotyping arrays is presented. The proposed work adjusts the offset of upto 130mV between the ISFETs and the REFET in a calibration step and cancels out common mode noise by differential pairs with a 63.5dB CMRR. Conventionally in such arrays Ion-sensitive Field-effect Transistors were used as sensors with interface circuitry, requiring conversion of the signal prior to processing. In this system they form part of the conditioning circuitry, improving the signal-to-noise ratio and easing the design constraints of intelligent large scale system-on-chips.

This paper introduces a CMOS based system that has been designed to allow parallel comparison of fragmented DNA sequences for on-chip assembly. The compatibility of different existing PC-based algorithms for implementation in CMOS is compared and the overlap-layout-consensus approach is found to be the most suitable one. The designed system comprises a scalable processing array capable of parallel computation, which allows identification of overlaps in DNA fragments in addition to error tolerance through dynamic programming. Analysis shows that there is a “pixel area vs computation time” trade-off when implementing such a parallel architecture. Results from a hypothetical assembly confirm good overlap detection and error tolerance, with up to 94% similarity in the detected overlaps, when the error is as much as 10%.

This preliminary study reports on the combined use of wireless accelerometers and wireless EMG sensors for monitoring walking patterns. The sensor data was analyzed in frequency domain through FFT, PSD and time-frequency spectrogram analysis. Accelerometer spectra was found to shift towards lower frequencies (;50 Hz) during asymmetric walking. Median frequency was used to quantify the spectral shifts. The combined wireless accelerometer/EMG system showed potential for discrimination between the normal and asymmetric walking.

This paper presents a CMOS-based lab-on-chip platform for combined magnetic manipulation and opto-chemical sensing. Within each pixel, a Programmable Gate (PG) ISFET chemical sensor is combined with an active pixel sensor, and is encompassed within an inductive coil. The integrated pixel is tesselated to form an 8 × 8 array. Fabricated in a commercially available 0.35 μm CMOS technology, the system can be used for simultaneous optical imaging and pH sensing, and includes auto- calibration mechanisms for eliminating sensor non-idealities. A spatiotemporal magnetic field pattern generator has also been embedded for micro-scale magnetic manipulation. Controlled via a MATLAB based graphical user interface, the system achieves real time data acquisition at 6 fps, a pH sensitivity of 57 mV/pH and demonstrates magnetic manipulation of micro-beads.

This paper presents an integrated silicon pancreatic islet to be used for the control algorithm of an artificial pancreas. The system replicates the temporal behaviour of the electrophysiology of the alpha and beta cells which are located in the islet of Langerhans in the pancreas, generating signals which can be used to infuse the two regulatory hormones, glucagon and insulin, used for tight glycemic control. The silicon islet can be easily programmed to choose between multiple bursting periods and can thus be tailored to each patients insulin sensitivity. Designed in a 0.35μm process, the system is fully integrated with an on-chip digitally tuned oscillator to drive the system and an algorithmic mode A/D converter to interface to an electrochemical glucose sensor.

This paper presents a novel CMOS based PG-ISFET (programmable gate-ion sensitive field effect transistor) and readout for compensation of large threshold voltages observed with ISFETs fabricated in a standard CMOS process. The proposed device uses a capacitively coupled floating gate to allow tunability of its operating point to counteract the presence of trapped charge, thus allowing operation within a tolerable gate voltage range. By using feedback, an adaptive readout has been designed, which allows integration of the device as well as cancellation of reduced sensitivity due to extra capacitance of the programmable gate. Fabricated in a 0.35 mum CMOS process, the device can compensate for a variation of up to 14.2 V for 1 muA using a 3.3 V supply.

This work is an investigation on the large threshold voltage variation exhibited in CMOS based ISFETs. This irregularity is thoroughly examined and is identified to be caused by mobile ionic charge that is induced in the sensing membrane when the membrane is in contact with the ionic-solution. This auxiliary charge increments the effective capacitance of the sensing membrane, causing irregular shifts in the characteristics of the devices. Several methods for overcoming this issue are addressed.

This paper presents a novel readout circuit for a pH sensitive programmable-gate ion-sensitive field effect transistor (PG-ISFET) to overcome bias issues due to threshold voltage variation and increase output-referred sensitivity. Compared to other commonly-used ISFET readouts, this circuit uses two extra programmable nodes which are driven by a feedback configuration. Using the device in a source follower configuration, one node is used to evaluate and cancel the offset of the intrinsic device while the other tracks and amplifies changes in pH. A sample and hold protocol has been developed to minimize the leakage effects and improve the pH sensing range. The system has been designed and fabricated in AMS 0.35 mum, to compensate for a threshold voltage variation of plusmn10.5 V and provide a pH sensitivity of 200 mV/pH.

A CMOS based ion sensitive chemical imager chip is presented for real time spatio- temporal monitoring of chemical activity. Ion-sensitive field effect transistors (ISFETs) implemented on an unmodified process (AMS 0.35 um) are used as the chemical input to the analogue circuits. A 3 x 11 array has been designed and fabricated for monitoring applications such as cell metabolism and growth. Chemical sensor information is encoded using integrate and fire neurons with address event representation (AER) used to fan out the data. Global adaptive biasing of pixels is used to conserve power and allow automatic functionality in changing buffer concentrations.

This demonstration presents the first bio-inspired approach to glucose management of Type-I diabetic patients using a real-time closed-loop insulin delivery system. The delivery system consists of a glucose biosensor, used with the silicon beta-cell to drive a motorized pump. Glucose-induced bursting of beta cells in the pancreas are used to control the insulin secretion in our bodies. A low-power implementation of these metabolic cells in silicon is achieved resulting in efficient glucose control.

This paper presents a novel interface for ion-sensitive field effect transistors (ISFET) in which operation at a fixed electrical bias is achieved by voltage clamping. The chosen topology provides pH-dependent current and voltage output signals to drive an appropriate output stage. The circuit can be operated in either strong or weak inversion, depending on the requirements of the application with a pseudo-differential ISFET-REFET topology to allow use of an on-chip reference electrode. Simulation results are shown herein for single-ended and differential implementations. For a single-ended front-end with 1nA bias current, the strong inversion implementation at 2.5V supply has a power consumption of 0.185mW at pH 7 and the weak inversion implementation at 1.5V supply has a power consumption of 13nW at pH 7. The circuit and ISFET-REFET pair have been fabricated in the UMC 0.25μm CMOS technology.

In this paper we present a biologically inspired circuit capable of implementing the membrane potential behaviour of the pancreatic beta cell. This is important as it is correlated with insulin release which is required for regulating blood glucose at a healthy level. Dysfunction of these cells is the main cause of various forms of diabetes. A nano-power glucose biosensor is used to provide a chemical input to the silicon based beta-cell. Measured results of up to 25 mM of glucose concentration are shown, as well as bursting behaviour from the circuit which matches well with physiological results