Large language models (LLMs) hold significant promise in clinical practice, yet their real-world adoption is constrained by their propensity to produce erroneous and occasionally harmful outputs, particularly in the intricate domain of rare diseases (RDs). This study introduces RDguru, a conversational intelligent agent leveraging the LangChain framework and powered by GPT-3.5-turbo. RDguru offers a comprehensive suite of functionalities, encompassing evidence-traceable knowledge Q&A and professional medical consultations for differential diagnosis (DDX), integrating authoritative knowledge sources and reliable tools. A novel multi-source fusion diagnostic model, rooted in deep Q-network, amalgamates three diagnostic recommendation strategies (GPT-4, PheLR, and phenotype matching) to enhance diagnostic recall during medical consultations. Through tailored tools and advanced algorithms for retrieval-augmented generation, RDguru excels in knowledge Q&A, automated phenotype annotation, and RD DDX. A multi-aspect Q&A analysis demonstrates RDguru outperforms ChatGPT in generating descriptions aligned with authoritative knowledge, quantified by ROUGE scores, GPT-4-based automatic rating, and RAGAs evaluation metrics. Testing on 238 published RD cases reveals that RDguru's top 5 multi-source fusion diagnoses recapture 63.87% of actual diagnoses, marking a 5.47% improvement over the state-of-the-art diagnostic method PheLR. Furthermore, RDguru's consultation strategy proves effective in eliciting diagnostically beneficial phenotypes and refining the prioritization of genuine diagnoses through multi-round phenotype-orient questioning. Evaluations against established benchmarks and real-world patient data demonstrate RDguru's efficacy and reliability, highlighting its potential to enhance clinical decision-making in the realm of RDs.

Type 2 diabetes (T2D) is a worldwide chronic disease that is difficult to cure and causes a heavy social burden. Early prediction of T2D can effectively identify high-risk populations and facilitate earlier implementation of appropriate preventive interventions. Various early prediction models for T2D have been proposed. However, these methods do not consider the following factors: 1) health examination records (HER) containing health information before diagnosis; 2) rating information containing clinical knowledge; and 3) local and global information of time-series features. These diagnostically relevant factors need to be considered. It is challenging due to irregular and multivariate time series. In this paper, we propose the multi-feature map integrated attention model (MFMAM) for early diabetes prediction using HER. Specifically, HER is converted into the multi-feature map to capture local and global volatility, as well as the sequence order of high-dimensional features. We concatenate rating indicators to introduce clinical knowledge. In addition, considering missing and temporal patterns, we utilize missing and time embedding to learn the complex transition of health status. We adopt attention mechanisms to capture essential features (channels) and time points (spatial). To evaluate the proposed model, we conducted experiments on real-world long-term HER. The results demonstrated that MFMAM outperformed baseline models on tasks of varying sequence lengths and prediction windows. Moreover, we applied our designs to baseline models, and their performance was considerably improved. The proposed model contributes to the short-term and long-term early prediction of T2D in individuals with varying information richness.

Although transcranial ultrasound plane-wave imaging (PWI) has promising clinical application prospects, studies have shown that variable speed-of-sound (SoS) would seriously damage the quality of ultrasound images. The mismatch between the conventional constant velocity assumption and the actual SoS distribution leads to the general blurring of ultrasound images. The optimization scheme for reconstructing transcranial ultrasound image is often solved using iterative methods like full-waveform inversion. These iterative methods are computationally expensive and based on prior magnetic resonance imaging (MRI) or computed tomography (CT) information. In contrast, the multi-stencils fast marching (MSFM) method can produce accurate time travel maps for the skull with heterogeneous acoustic speed. In this study, we first propose a convolutional neural network (CNN) to predict SoS maps of the skull from PWI channel data. Then, use these maps to correct the travel time to reduce transcranial aberration. To validate the performance of the proposed method, numerical, phantom and intact human skull studies were conducted using a linear array transducer (L11-5v, 128 elements, pitch = 0.3 mm). Numerical simulations demonstrate that for point targets, the lateral resolution of MSFM-restored images increased by 65%, and the center position shift decreased by 89%. For the cyst targets, the eccentricity of the fitting ellipse decreased by 75%, and the center position shift decreased by 58%. In the phantom study, the lateral resolution of MSFM-restored images was increased by 49%, and the position shift was reduced by 1.72 mm. This pipeline, termed AutoSoS, thus shows the potential to correct distortions in real-time transcranial ultrasound imaging, as demonstrated by experiments on the intact human skull.

Phenotype-based prioritization of candidate genes and diseases has become a well-established approach for multi-omics diagnostics of rare diseases. Most current algorithms exploit semantic analysis and probabilistic statistics based on Human Phenotype Ontology and are commonly superior to naive search methods. However, these algorithms are mostly less interpretable and do not perform well in real clinical scenarios due to noise and imprecision of query terms, and the fact that individuals may not display all phenotypes of the disease they belong to. We present a Phenotype-driven Likelihood Ratio analysis approach (PheLR) assisting interpretable clinical diagnosis of rare diseases. With a likelihood ratio paradigm, PheLR estimates the posterior probability of candidate diseases and how much a phenotypic feature contributes to the prioritization result. Benchmarked using simulated and realistic patients, PheLR shows significant advantages over current approaches and is robust to noise and inaccuracy. To facilitate clinical practice and visualized differential diagnosis, PheLR is implemented as an online web tool (https://phelr.nbscn.org).

Deep learning based computer-aided diagnosis technology demonstrates an encouraging performance in aspect of polyp lesion detection on reducing the miss rate of polyps during colonoscopies. However, to date, few studies have been conducted for tracking polyps that have been detected in colonoscopy videos, which is an essential and intuitive issue in clinical intelligent video analysis task (e.g. lesion counting, lesion retrieval, report generation). In the paradigm of conventional tracking-by-detection system, detection task for lesion localization is separated from the tracking task for cropped lesions re-identification. In the multi object tracking problem, each target is supposed to be tracked by invoking a tracker after the detector, which introduces multiple inferences and leads to external resource and time consumption. To tackle these problems, we proposed a plug-in module named instance tracking head (ITH) for synchronous polyp detection and tracking, which can be simply inserted into object detection frameworks. It embeds a feature-based polyp tracking procedure into the detector frameworks to achieve multi-task model training. ITH and detection head share the model backbone for low level feature extraction, and then low level feature flows into the separate branches for task-driven model training. For feature maps from the same receptive field, the region of interest head assigns these features to the detection head and the ITH, respectively, and outputs the object category, bounding box coordinates, and instance feature embedding simultaneously for each specific polyp target. We also proposed a method based on similarity metric learning. The method makes full use of the prior boxes in the object detector to provide richer and denser instance training pairs, to improve the performance of the model evaluation on the tracking task. Compared with advanced tracking-by-detection paradigm methods, detectors with proposed ITH can obtain comparative tracking performance but approximate 30% faster speed. Optimized model based on Scaled-YOLOv4 detector with ITH illustrates good trade-off between detection (mAP 91.70%) and tracking (MOTA 92.50% and Rank-1 Acc 88.31%) task at the frame rate of 66 FPS. The proposed structure demonstrates the potential to aid clinicians in real-time detection with online tracking or offline retargeting of polyp instances during colonoscopies.

OBJECTIVE: Acute kidney injury (AKI) is a common complication after pediatric cardiac surgery, and the early detection of AKI may allow for timely preventive or therapeutic measures. However, current AKI prediction researches pay less attention to time information among time-series clinical data and model building strategies that meet complex clinical application scenario. This study aims to develop and validate a model for predicting postoperative AKI that operates sequentially over individual time-series clinical data. MATERIALS AND METHODS: A retrospective cohort of 3386 pediatric patients extracted from PIC database was used for training, calibrating, and testing purposes. A time-aware deep learning model was developed and evaluated from 3 clinical perspectives that use different data collection windows and prediction windows to answer different AKI prediction questions encountered in clinical practice. We compared our model with existing state-of-the-art models from 3 clinical perspectives using the area under the receiver operating characteristic curve (ROC AUC) and the area under the precision-recall curve (PR AUC). RESULTS: Our proposed model significantly outperformed the existing state-of-the-art models with an improved average performance for any AKI prediction from the 3 evaluation perspectives. This model predicted 91% of all AKI episodes using data collected at 24 h after surgery, resulting in a ROC AUC of 0.908 and a PR AUC of 0.898. On average, our model predicted 83% of all AKI episodes that occurred within the different time windows in the 3 evaluation perspectives. The calibration performance of the proposed model was substantially higher than the existing state-of-the-art models. CONCLUSIONS: This study showed that a deep learning model can accurately predict postoperative AKI using perioperative time-series data. It has the potential to be integrated into real-time clinical decision support systems to support postoperative care planning.

Healthcare process leaves patient treatment trajectory (PTT), described as a sequence of interdependent clinical events affiliated with a large volume of longitudinal therapy and treatment information. Predicting the future clinical event in PTT, as a vital and essential task for providing insights into the entire treatment trajectory, can serve as an efficient and proactive altering service for health service delivery. However, it is challenging because there are long-term dependencies between clinical events, which are irregularly distributed along the temporal axis with varying time intervals. This characteristic inevitably impedes the performance of clinical event prediction (CEP) using the existing approaches. To address this challenge, we propose a novel approach to learn representative and discriminative PTT features for CEP. In detail, multivariate Hawkes process (HP) is adopted to uncover the mutual excitation intensities between clinical event pairs in an interpretable manner. Thereafter, the captured spontaneous and interactional intensities of events are incorporated into recurrent neural networks (RNN) to encode PTT in latent representations, while jointly performing the CEP task based on the extracted trajectory representations. We evaluate the performance of the proposed approach on a real clinical dataset consisting of 13,545 visits of 2,102 heart failure patients. Compared to state-of-the-art methods, our best model achieves 6.4% and 4.1% AUC performance gains on three-months and one-year CEP tasks, respectively. The experimental results demonstrate that the proposed approach outperforms state-of-the-art models in CEP, and can be profitably exploited as a basis for PTT analysis and optimization.

In recent years, it has been demonstrated that checklists can improve patient safety significantly. To facilitate the effective use of checklists in daily practice, both the medical community and the informatics community propose to implement checklists in dynamic checklist applications that can be integrated into the clinical workflow and that is specific to the patient context. However, it is difficult to develop such applications because they are tightly intertwined with the content of specific checklists. We propose a platform that enables access to dynamic checklist applications by configuring the infrastructures provided in the platform. Then, the applications can be developed without time-consuming programming work. We define a number of design criteria regarding point of care and clinical processes by analyzing the existing checklist applications and the lessons learned from implementations. Then, by applying rule-based clinical decision support and workflow management technologies, we design technical mechanisms to satisfy the design criteria. A dynamic checklist application platform is designed based on these mechanisms. Finally, we build a platform in various design cycle iterations, driven by multiple clinical cases. By applying the platform, we develop nine comprehensive dynamic checklist applications with 242 dynamic checklists. The results demonstrate both the feasibility and the overall generic nature of the proposed approach. We propose a novel platform for configuring dynamic checklist applications. This platform satisfies the general requirements and can be easily configured to satisfy different scenarios in which safety checklists are used.

Clinical pathways (CPs) are popular healthcare management tools to standardize care and ensure quality. Analyzing CP compliance levels and variances is known to be useful for training and CP redesign purposes. Flexible semantics of the business process model and notation (BPMN) language has been shown to be useful for the modeling and analysis of complex protocols. However, in practical cases one may want to exploit that CPs often have the form of task-time matrices. This paper presents a new method parsing complex BPMN models and aligning traces to the models heuristically. A case study on variance analysis is undertaken, where a CP from the practice and two large sets of patients data from an electronic medical record (EMR) database are used. The results demonstrate that automated variance analysis between BPMN task-time models and real-life EMR data are feasible, whereas that was not the case for the existing analysis techniques. We also provide meaningful insights for further improvement.

Clinical safety checklists receive much research attention since they can reduce medical errors and improve patient safety. Computerized checklist support systems are also being developed actively. Such systems should individualize checklists based on information from the patient’s medical record while also considering the context of the clinical workflows. Unfortunately, the form definitions, database queries and workflow definitions related to dynamic checklists are too often hard-coded in the source code of the support systems. This increases the cognitive effort for the clinical stakeholders in the design process, it complicates the sharing of dynamic checklist definitions as well as the interoperability with other information systems. In this paper, we address these issues by contributing the DCCSS meta-model which enables the model-based development of dynamic checklist support systems. DCCSS was designed as an incremental extension of standard meta-models, which enables the reuse of generic model editors in a novel setting. In particular, DCCSS integrates the Business Process Model and Notation (BPMN) and the Guideline Interchange Format (GLIF), which represent best of breed languages for clinical workflow modeling and clinical rule modeling respectively. We also demonstrate one of the use cases where DCCSS has already been applied in a clinical setting.

It has recently been demonstrated that checklists can enable significant improvements to patient safety. However, their clinical acceptance is significantly lower than expected. This is due to the lack of good support systems. Specifically, support systems are too static: this holds for paper-based support as well as for electronic systems that digitize paper-based support naively. Both approaches are independent from clinical process and clinical context. In this paper, we propose a process-oriented and context-aware dynamic checklist support system: Trace book. This system supports the execution of complex clinical processes and rules involving data from Electronic Medical Record systems. Workflow activities and forms are specific to individual patients based on clinical rules and they are dispatched to the right user automatically based on a process model. Besides describing the Trace book functionality in general, this paper demonstrates the support system specifically on an example application that we are preparing for a controlled clinical evaluation. At last we discuss the limitations of Trace book.

Clinical pathways leave traces, described as event sequences with regard to a mixture of various latent treatment behaviors. Measuring similarities between patient traces can profitably be exploited further as a basis for providing insights into the pathways, and complementing existing techniques of clinical pathway analysis (CPA), which mainly focus on looking at aggregated data seen from an external perspective. Most existing methods measure similarities between patient traces via computing the relative distance between their event sequences. However, clinical pathways, as typical human-centered processes, always take place in an unstructured fashion, i.e., clinical events occur arbitrarily without a particular order. Bringing order in the chaos of clinical pathways may decline the accuracy of similarity measure between patient traces, and may distort the efficiency of further analysis tasks. In this paper, we present a behavioral topic analysis approach to measure similarities between patient traces. More specifically, a probabilistic graphical model, i.e., latent Dirichlet allocation (LDA), is employed to discover latent treatment behaviors of patient traces for clinical pathways such that similarities of pairwise patient traces can be measured based on their underlying behavioral topical features. The presented method provides a basis for further applications in CPA. In particular, three possible applications are introduced in this paper, i.e., patient trace retrieval, clustering, and anomaly detection. The proposed approach and the presented applications are evaluated via a real-world dataset of several specific clinical pathways collected from a Chinese hospital.

Clinical protocols are a popular instrument to document how clinicians are expected to behave under specific conditions. Protocols are typically based on internationally peer reviewed clinical guidelines as well as on hospital-local agreements. Existing techniques for monitoring protocol adherence only support protocol descriptions involving simple sequences and local decision rules. As care and cure processes are becoming increasingly complex, the need for more advanced techniques naturally emerges. In this paper we present a novel approach to defining and monitoring complex clinical protocols. By using BPMN to document protocols we enable the concise specification of protocols that involve multiple stakeholders that operate in parallel and under uncertainty. Uncertainty relates to the fact that protocols may involve complex loops and choices. While this specification style was becoming increasingly popular in the literature and practice of hospital management and operations management in general, corresponding conformance analysis techniques were still lacking. This paper contributes the first such technique and evaluate it on a complex compliance pattern from the cardiology domain.

Resource allocation, as an integral part of business process management (BPM), is more widely acknowledged by its importance for process-aware information systems. Despite the industrial need for efficient and effective resource allocation in BPM, few scientifically-grounded approaches exist to support these initiatives. In this paper, a new approach of resource allocation optimization is proposed, built on the concepts that is part of an operation-oriented view on process optimization. Essentially, the proposed approach automatically generates a specific task operation model (TOM) for a particular business process. In addition, in order to support end users in making sensible resource allocations, an ant colony optimization-based algorithm is presented, which makes it possible to search an optimal task operation path on the generated TOM. This allows one to suggest how a business user should efficiently allocate resources to perform the tasks of a particular process case. The feasibility of the presented approach is demonstrated by a simulation experiment. The experimental results show that the proposed approach outperforms reasonable heuristic approaches to satisfy process performance goals, and it is possible to improve the current state of BPM.

Microarray technology has been increasingly recognized as a powerful means for monitoring the expression levels of thousands of genes simultaneously. Microarray image processing is an essential aspect of microarray experiment, of which gridding is thought to be the most important step of spot recognition. Many times, microarray image gridding requires assisted intervention to achieve the acceptable accuracy. In this paper, an automatic microarray image gridding algorithm was presented by using image projection vectors together with power spectrum model. For obtaining grid position, the image projection vectors were utilized by adequately considering the grid parameters. On the other hand, as a preprocessing procedure of microarray gridding, detection of the grid rotation was involved in our study by using power spectrum analyses of the image projection vectors. Our approach has been evaluated by three different microarray datasets. Experimental comparisons with up-to-date approaches by using both synthetic and real image data are demonstrated. The gridding result was shown to be very accurate, and able to provide correct gridding dataset for the downstream microarray analyses. In summary, our study demonstrated the combination of image projection vectors with power spectrum model as a powerful strategy for microarray image gridding.

In order to improve the performance of mass segmentation on mammograms, an intelligent algorithm is proposed in this paper. It establishes two mass models to characterize the various masses, and the ones in the denser tissue are represented with Model I, while the ones in the fatty tissue are represented with Model II. Then, it uses iterative thresholding to extract the suspicious area, as well as the rough regions of those masses matching Model II, and applies a DWT-based technique to locate those masses matching Model I, which are hidden in the high gray-level intensity and contrast area. A region growing process restricted by Canny edge detection is subsequently used to segment the rough regions of those masses matching Model I, and finally snakes are carried out to find all the mass regions roughly extracted above. Thirty patient cases with 60 mammograms and 107 masses were used for evaluation, and the experimental result has demonstrated the algorithm's better performance over the conventional methods.