Long-term cuffless monitoring of blood pressure (BP) is of immense value in the diagnosis and prevention of cardiovascular diseases (CVD). While there are critical needs for wearable cuffless BP monitoring devices in personalized CVD care and hypertension management, current technologies face challenges in long-term accuracy which limits significantly its clinical applicability. To maintain the accuracy of cuffless blood pressure forecasting, frequent calibration is a vital necessity. However, conventional BP calibration approaches are inherently inconvenient and time-consuming. In this work, we address the challenge by developing a novel auto-calibration system enabling long-term BP estimation. The innovation of this approach combines depth-resolved vascular characteristics extracted from multi-wavelength photoplethysmographic (MWPPG) measurements at fingertip, with an adaptive cross-modal knowledge distillation algorithm for continuous refinement of blood pressure estimation. We have evaluated the system on the CAS-BP database of over a thousand subjects using various calibration strategies. The results have shown that frequent calibration can effectively enhance the accuracy of long-term blood pressure estimation. Compared to traditional manual calibration, the proposed auto-calibration method can perform frequent calibration of the model without requiring the true BP values. Furthermore, under the condition of daily auto-calibration, the system attains an optimal Mean Absolute Error (MAE) of 3.77 mmHg and 2.62 mmHg for systolic BP (SBP) and diastolic BP (DBP) respectively, significantly outperforming the performance achieved through traditional calibration methods. This work provides a novel conceptual framework to guide the calibration of cuffless wearable devices for BP monitoring.

Heart failure (HF) is a complex and heterogeneous syndrome caused by diverse factors, such as atrial fibrillation, diabetes, and pulmonary hypertension. The intricate pathophysiology of HF, coupled with variability in patient demographics and presentations, poses significant challenges to the effectiveness of existing deep learning models in HF risk prediction. In this paper, we propose a novel deep neural network, PDSNet, which leverages a new dual patient-disease spatial similarity strategy to improve HF risk prediction using short electronic health records. First, we develop ontology graphs to capture hierarchical relationships between patients based on HF-related symptoms and causes; Then, a bipartite graph model is utilised to learn spatial similarities between patients with similar hospital visit histories; Finally, we design a transformer-based architecture to integrate both temporal and spatial dynamics for predicting future hospital visits associated with HF risk. We benchmarked the PDSNet on predicting HF risk for 7,346 patients from the MIMIC-III dataset. Compared to seven state-of-the-art deep learning methods, our PDSNet model achieved improvements of 2%-12% in the area under the curve (AUC) score and 3%-18% in the F1 score. These findings highlight the promising potential of our proposed PDSNet to provide accurate and robust HF risk predictions, paving the way for efficient clinical decision support and personalized HF management.

Non-negative matrix factorization (NMF), widely used in motor neuroscience for identifying muscle synergies from electromyographical signals (EMGs), extracts non-negative synergies and is yet unable to identify potential negative components (NegCps) in synergies underpinned by inhibitory spinal interneurons. To overcome this constraint, we propose to utilize rectified latent variable model (RLVM) to extract muscle synergies. RLVM uses an autoencoder neural network, and the weight matrix of its neural network could be negative, while latent variables must remain non-negative. If inputs to the model are EMGs, the weight matrix and latent variables represent muscle synergies and their temporal activation coefficients, respectively. We compared performances of NMF and RLVM in identifying muscle synergies in simulated and experimental datasets. Our simulated results showed that RLVM performed better in identifying muscle-synergy subspace and NMF had a good correlation with ground truth. Finally, we applied RLVM to a previously published experimental dataset comprising EMGs from upper-limb muscles and spike recordings of spinal premotor interneurons (PreM-INs) collected from two macaque monkeys during grasping tasks. RLVM and NMF synergies were highly similar, but a few small negative muscle components were observed in RLVM synergies. The muscles with NegCps identified by RLVM exhibited near-zero values in their corresponding synergies identified by NMF. Importantly, NegCps of RLVM synergies showed correspondence with the muscle connectivity of PreM-INs with inhibitory muscle fields, as identified by spike-triggered averaging of EMGs. Our results demonstrate the feasibility of RLVM in extracting potential inhibitory muscle-synergy components from EMGs.

Training deep neural networks (DNNs) typically requires massive computational power. Existing DNNs exhibit low time and storage efficiency due to the high degree of redundancy. In contrast to most existing DNNs, biological and social networks with vast numbers of connections are highly efficient and exhibit scale-free properties indicative of the power law distribution, which can be originated by preferential attachment in growing networks. In this work, we ask whether the topology of the best performing DNNs shows the power law similar to biological and social networks and how to use the power law topology to construct well-performing and compact DNNs. We first find that the connectivities of sparse DNNs can be modeled by truncated power law distribution, which is one of the variations of the power law. The comparison of different DNNs reveals that the best performing networks correlated highly with the power law distribution. We further model the preferential attachment in DNNs evolution and find that continual learning in networks with growth in tasks correlates with the process of preferential attachment. These identified power law dynamics in DNNs can lead to the construction of highly accurate and compact DNNs based on preferential attachment. Inspired by the discovered findings, two novel applications have been proposed, including evolving optimal DNNs in sparse network generation and continual learning tasks with efficient network growth using power law dynamics. Experimental results indicate that the proposed applications can speed up training, save storage, and learn with fewer samples than other well-established baselines. Our demonstration of preferential attachment and power law in well-performing DNNs offers insight into designing and constructing more efficient deep learning.

This article investigates the use of acoustic signals recorded at the human wrist for hand gesture recognition. The prototype consists of 40 microphones to be worn at the wrist. The gesture recognition performance is evaluated through the identification of 36 gestures in American sign language (ASL), including 26 ASL alphabetical characters and 10 ASL numbers. The optimal area for sensor band placement (distal/proximal) is examined to reveal the location of the highest discrimination accuracy. Ten subjects are recruited to perform over ten trials for each set of hand gestures. Using mutual information-based feature selection methods, two time-domain features: difference absolute mean value and log-energy entropy, is selected from 35 commonly used time-domain features for the hand gesture classification. As a proof-of-concept, two common classification methods, linear discriminant analysis and support vector machine, is used to classify hand gestures from the acoustic measurements recorded by each subject. Results show the intrasubject average classification accuracy above 90% using the two features with all 40 microphones, while the average classification accuracy exceeding 84% is obtained using ten microphones. These results indicate that acoustic signatures from the human wrist can be utilized for hand gesture recognition, while the use of few, simple features, with low computational requirements is sufficient to characterize some hand gestures. The proposed technique demonstrates a promising means for developing a low-cost wearable hand gesture recognition device using microphones.

The recent breakthroughs in computer vision have benefited from the availability of large representative datasets (e.g. ImageNet and COCO) for training. Yet, robotic vision poses unique challenges for applying visual algorithms developed from these standard computer vision datasets due to their implicit assumption over non-varying distributions for a fixed set of tasks. Fully retraining models each time a new task becomes available is infeasible due to computational, storage and sometimes privacy issues, while naïve incremental strategies have been shown to suffer from catastrophic forgetting. It is crucial for the robots to operate continuously under open-set and detrimental conditions with adaptive visual perceptual systems, where lifelong learning is a fundamental capability. However, very few datasets and benchmarks are available to evaluate and compare emerging techniques. To fill this gap, we provide a new lifelong robotic vision dataset ("OpenLORIS-Object") collected via RGB-D cameras. The dataset embeds the challenges faced by a robot in the real-life application and provides new benchmarks for validating lifelong object recognition algorithms. Moreover, we have provided a testbed of 9 state-of-the-art lifelong learning algorithms. Each of them involves 48 tasks with 4 evaluation metrics over the OpenLORIS-Object dataset. The results demonstrate that the object recognition task in the ever-changing difficulty environments is far from being solved and the bottlenecks are at the forward/backward transfer designs. Our dataset and benchmark are publicly available at https://lifelong-robotic-vision.github.io/dataset/object.

Service robots should be able to operate autonomously in dynamic and daily changing environments over an extended period of time. While Simultaneous Localization And Mapping (SLAM) is one of the most fundamental problems for robotic autonomy, most existing SLAM works are evaluated with data sequences that are recorded in a short period of time. In real-world deployment, there can be out-of-sight scene changes caused by both natural factors and human activities. For example, in home scenarios, most objects may be movable, replaceable or deformable, and the visual features of the same place may be significantly different in some successive days. Such out-of-sight dynamics pose great challenges to the robustness of pose estimation, and hence a robot’s long-term deployment and operation. To differentiate the forementioned problem from the conventional works which are usually evaluated in a static setting in a single run, the term lifelong SLAM is used here to address SLAM problems in an ever-changing environment over a long period of time. To accelerate lifelong SLAM research, we release the OpenLORIS-Scene datasets. The data are collected in real-world indoor scenes, for multiple times in each place to include scene changes in real life. We also design benchmarking metrics for lifelong SLAM, with which the robustness and accuracy of pose estimation are evaluated separately. The datasets and benchmark are available online at lifelong-robotic-vision.github.io/dataset/scene.

Linear Dynamical Systems are widely used to study the underlying patterns of multivariate time series. A basic assumption of these models is that high-dimensional time series can be characterized by some underlying, low-dimensional and time-varying latent states. However, existing approaches to LDS modeling mostly learn the latent space with a prescribed dimensionality. When dealing with short-length high- dimensional time series data, such models would be easily overfitted. We propose Reduced-Rank Linear Dynamical Systems (RRLDS), to automatically retrieve the intrinsic dimensionality of the latent space during model learning. Our key observation is that the rank of the dynamics matrix of LDS captures the intrinsic dimensionality, and the variational inference with a reduced-rank regularization finally leads to a concise, structured, and interpretable latent space. To enable our method to handle count-valued data, we introduce the dispersion-adaptive distribution to accommodate over-/ equal-/ and under-dispersion nature of such data. Results on both simulated and experimental data demonstrate our model can robustly learn latent space from short-length, noisy, count-valued data and significantly improve the prediction performance over the state-of-the-art methods.

The brain encodes information by neural spiking activities, which can be described by time series data as spike counts. Latent Variable Models (LVMs) are widely used to study the unknown factors (i.e. the latent states) that are dependent in a network structure to modulate neural spiking activities. Yet, challenges in performing experiments to record on neuronal level commonly results in relatively short and noisy spike count data, which is insufficient to derive latent network structure by existing LVMs. Specifically, it is difficult to set the number of latent states. A small number of latent states may not be able to model the complexities of underlying systems, while a large number of latent states can lead to overfitting. Therefore, based on a specific LVMs called Linear Dynamical System (LDS), we propose a Reduced-Rank Linear Dynamical System (RRLDS) to estimate latent states and retrieve an optimal latent network structure from short, noisy spike count data. This framework estimates the model using Laplace approximation. To further handle count-valued data, we introduce the dispersion-adaptive distribution to accommodate over-/ equal-/ and under-dispersion nature of such data. Results on both simulated and experimental data demonstrate our model can robustly learn latent space from short-length, noisy, count-valued data and significantly improve the prediction performance over the state-of-the-art methods.

We propose an Internet of Things (IoT) framework for next-generation racket sports training. To validate its performance, a wireless wearable sensing device (WSD) based on microelectromechanical systems motion sensors was used to recognize different badminton strokes and classify skill levels from different badminton players. The system includes a customized sensor node for data collection, a mobile app, and a cloud-based data processing unit. The WSD developed is low-cost, easy-to-use, and computationally efficient compared to video-based methods for analyzing badminton strokes. It offers the advantage of dynamic monitoring of multiple players in indoor and outdoor environments. In this paper, we present the hardware design, mobile software implementation, and data processing algorithms of the system. Twelve right-handed male subjects wore the WSD on their wrists while each performed 30 trials of different strokes in a real badminton court. The results show that our system is capable of recognizing three different actions, i.e.,smashes,clears, anddrops, with an accuracy rate of 97%. The skill assessment function can differentiate between professional, subelite, and amateur players from their stroke performance. This IoT framework aims to change the way of racket sports training from experience-driven (subjective) to data-driven (objective), and which can be easily extended to analyze the motions and skill levels of players in other racket sports (e.g., tennis, table tennis, and squash) for training and/or practice.

The concept of muscle synergies has proven to be an effective method for representing patterns of muscle activation. The number of degrees of freedom to be controlled are reduced while also providing a flexible platform for producing detailed movements using synergies as building blocks. It has previously been shown that small components of movement are crucial to producing precise and coordinated movement. Methods which focus on the variance of the data make it possible to overlook these small components in the synergy extraction process. However, algorithms which address the inherent complexity in the neuromuscular system are lacking. To that end we propose a new nonnegative matrix factorization algorithm which employs a cross fuzzy entropy similarity measure, thus, extracting muscle synergies which preserve the complexity of the recorded muscular data. The performance of the proposed algorithm is illustrated on representative EMG data.

Conventional dimensionality reduction methods are not applicable for hybrid data as they require the data set to be pure numerical. In this study, the mutual information (MI)-based unsupervised feature transformation (UFT) method which can transform symbolic features into numerical features without information loss was integrated with principle component analysis (PCA) for dimensionality reduction of hybrid data. The NSL-KDD data set for internet intrusion detection was used to verify this integrated UFT+PCA method. The experimental results show that, the UFT+PCA can reduce the dimension and improve the classification accuracies of hybrid data effectively.

Dementia is one of the most common neurological disorders among the elderly. Identifying those who are of high risk suffering dementia is important for early diagnosis in order to slow down the disease progression and help preserve some cognitive functions of the brain. To achieve accurate classification, significant amount of subject feature information are involved. Hence identification of demented subjects can be transformed into a pattern classification problem. In this letter, we introduce a graph based semi-supervised learning algorithm for Medical Diagnosis by using partly labeled samples and large amount of unlabeled samples. The new method is derived by a compact graph that can well grasp the manifold structure of medical data. Simulation results show that the proposed method can achieve better sensitivities and specificities compared with other state-of-art graph based semi-supervised learning methods.

Normal and visually-impaired zebrafish larvae have differentiable light-induced locomotor response (LLR), which is composed of visual and non-visual components. It is recently demonstrated that differences in the acute phase of the LLR, also known as the visual motor response (VMR), can be utilized to evaluate new eye drugs. However, most of the previous studies focused on the average LLR activity of a particular genotype, which left information that could address differences in individual zebrafish development unattended. In this study, machine learning techniques were employed to distinguish not only zebrafish larvae of different genotypes, but also different batches, based on their response to light stimuli. This approach allows us to perform efficient high-throughput zebrafish screening with relatively simple preparations. Following the general machine learning framework, some discriminative features were first extracted from the behavioral data. Both unsupervised and supervised learning algorithms were implemented for the classification of zebrafish of different genotypes and batches. The accuracy of the classification in genotype was over 80 percent and could achieve up to 95 percent in some cases. The results obtained shed light on the potential of using machine learning techniques for analyzing behavioral data of zebrafish, which may enhance the reliability of high-throughput drug screening.

This study explores the feasibility of mental workload monitoring using a single-channel mobile EEG system. We investigated the modulation of frontal neural activity with respect to changes in mental workload levels induced by visual-motor tasks of varied difficulty. Using a computerized visual-motor task similar to mirror drawing, our work demonstrated that perceived difficulty was more dominated by the complexity of the path to be traced than the actual time taken to complete the task. EEG activities recorded from the forehead area at the beginning of each trial are positively correlated with overall perceived difficulty of the task. Results in this study suggest that frontal EEG spectra are significantly modulated by the changes in relative mental workload levels during a visual-motor task. Such finding shed light on the possibility of mental workload monitoring in daily life using a user-friendly mobile EEG system.

In this paper, we propose an FPGA-based hardware architecture for conducting real-time prediction of neural activity using a second-order generalized Laguerre-Volterra model (GLVM). This architecture serves as a rapid prototype of the prediction module of the future cognitive neural prosthetic device. We validate the functionality of the hardware model by utilizing the neuronal firing data of behaving rats trained to perform the delayed nonmatch-to-sample (DNMS) memory task.

Dementia is one of the most common neurological disorders among the elderly. Identifying those who are of high risk suffering dementia is important to the administration of early treatment in order to slow down the progression of dementia symptoms. However, to achieve accurate classification, significant amount of subject feature information are involved. Hence identification of demented subjects can be transformed into a pattern recognition problem with high-dimensional nonlinear datasets. In this paper, we introduce trace ratio linear discriminant analysis (TR-LDA) for dementia diagnosis. An improved ITR algorithm (iITR) is developed to solve the TR-LDA problem. This novel method can be integrated with advanced missing value imputation method and utilized for the analysis of the nonlinear datasets in many real-world medical diagnosis problems. Finally, extensive simulations are conducted to show the effectiveness of the proposed method. The results demonstrate that our method can achieve higher accuracies for identifying the demented patients than other state-of-art algorithms.

We present a parallelized and pipelined architecture for a generalized Laguerre-Volterra MIMO system to identify the time-varying neural dynamics underlying spike activities. The proposed architecture consists of a first stage containing a vector convolution and MAC (Multiply and Accumulation) component, a second stage containing a pre-threshold potential updating unit with an error approximation function component, and a third stage consisting of a gradient calculation unit. A flexible and efficient architecture that can accommodate different design speed requirements are generated. Simulation results are rigorously analyzed. A hardware IP library for versatile models and applications is proposed. The design runs on a Xilinx Virtex-6 FPGA and the processing core produces data samples at a maximum clock rate of 357MHz, which is 4.37 × 105times faster than the corresponding software model running on an AMD Pheono 9750 Quad Core Processor. It occupies 216,766 LUTs, maximum 12 block-RAMs, and 2016 DSP-blocks.

We present a full-parallelized and pipelined architecture for a generalized Laguerre-Volterra MIMO system to identify the time-varying neural dynamics underlying spike activities. The proposed architecture consists of a first stage containing a vector convolution and MAC (Multiply and Accumulation) component, a second stage containing a pre-threshold potential updating unit with an error approximation function component, and a third stage consisting of a gradient calculation unit. A flexible and efficient architecture that can accommodate different design speed requirements is generated. The design runs on a Xilinx Virtex-6 FPGA and the processing core produces data samples at a speed of 1.33×106data frames/sec, which is 3.1×103times faster than the corresponding C model running on an Intel i7-860 Quad Core Processor.

The successful development of neural prostheses requires an understanding of the neurobiological bases of cognitive processes, i.e., how the collective activity of populations of neurons results in a higher level process not predictable based on knowledge of the individual neurons and/or synapses alone. We have been studying and applying novel methods for representing nonlinear transformations of multiple spike train inputs (multiple time series of pulse train inputs) produced by synaptic and field interactions among multiple subclasses of neurons arrayed in multiple layers of incompletely connected units. We have been applying our methods to study of the hippocampus, a cortical brain structure that has been demonstrated, in humans and in animals, to perform the cognitive function of encoding new long-term (declarative) memories. Without their hippocampi, animals and humans retain a short-term memory (memory lasting approximately 1 min), and long-term memory for information learned prior to loss of hippocampal function. Results of more than 20 years of studies have demonstrated that both individual hippocampal neurons, and populations of hippocampal cells, e.g., the neurons comprising one of the three principal subsystems of the hippocampus, induce strong, higher order, nonlinear transformations of hippocampal inputs into hippocampal outputs. For one synaptic input or for a population of synchronously active synaptic inputs, such a transformation is represented by a sequence of action potential inputs being changed into a different sequence of action potential outputs. In other words, an incoming temporal pattern is transformed into a different, outgoing temporal pattern. For multiple, asynchronous synaptic inputs, such a transformation is represented by a spatiotemporal pattern of action potential inputs being changed into a different spatiotemporal pattern of action potential outputs. Our primary thesis is that the encoding of short-term memories into new, long-term memories represents the collective set of nonlinearities induced by the three or four principal subsystems of the hippocampus, i.e., entorhinal cortex-to-dentate gyrus, dentate gyrus-to-CA3 pyramidal cell region, CA3-to-CA1 pyramidal cell region, and CA1-to-subicular cortex. This hypothesis will be supported by studies using in vivo hippocampal multineuron recordings from animals performing memory tasks that require hippocampal function. The implications for this hypothesis will be discussed in the context of "cognitive prostheses"-neural prostheses for cortical brain regions believed to support cognitive functions, and that often are subject to damage due to stroke, epilepsy, dementia, and closed head trauma.

DNA electronics circuits require an efficient way to accurately position and individually manipulate DNA molecules. The recent development of atomic force microscopy (AFM) seems to be a promising solution. We have recently developed an AFM based augmented reality system. This new system can provide both real-time force feedback and real-time visual feedback during nanomanipulation. We have shown that nano-imprinting and manipulation of nano-particles and nano-rods can be easily performed under assistance of the augmented reality system. In this research, the system's ability is extended to manipulation of DNA molecules. Using a polynomial fitting method, the deformation of DNA molecules is displayed in real time in the augmented reality system during manipulation. Indeed, DNA molecules adopt many different structures including kinks, bends, bulges and distortions. These different structures and inappropriate physical contacts may result in the controversy of DNA conductivity reported over the last decade. The AFM based nanomanipulation system can be used either as a nanolithography tool to make small gap electrodes or a nanomanipulation tool to elongate, deform and cut DNA molecules. The measurement of the conductivity of DNA molecules in their different shapes and structures is a promising method to find conclusive evidences, which verify the electrical conductivity of DNA molecules.