Deep Learning in Heart Sound Analysis: From Techniques to Clinical Applications.

Importance: Heart sound auscultation is a routinely used physical examination in clinical practice to identify potential cardiac abnormalities. However, accurate interpretation of heart sounds requires specialized training and experience, which limits its generalizability. Deep learning, a subset of machine learning, involves training artificial neural networks to learn from large datasets and perform complex tasks with intricate patterns. Over the past decade, deep learning has been successfully applied to heart sound analysis, achieving remarkable results and accumulating substantial heart sound data for model training. Although several reviews have summarized deep learning algorithms for heart sound analysis, there is a lack of comprehensive summaries regarding the available heart sound data and the clinical applications. Highlights: This review will compile the commonly used heart sound datasets, introduce the fundamentals and state-of-the-art techniques in heart sound analysis and deep learning, and summarize the current applications of deep learning for heart sound analysis, along with their limitations and areas for future improvement. Conclusions: The integration of deep learning into heart sound analysis represents a significant advancement in clinical practice. The growing availability of heart sound datasets and the continuous development of deep learning techniques contribute to the improvement and broader clinical adoption of these models. However, ongoing research is needed to address existing challenges and refine these technologies for broader clinical use.

Screening Tool for Paroxysmal Atrial Fibrillation Based on a Deep-Learning Algorithm Using Printed 12-Lead Electrocardiographic Records during Sinus Rhythm.

Background: Recent advancements in artificial intelligence (AI) have significantly improved atrial fibrillation (AF) detection using electrocardiography (ECG) data obtained during sinus rhythm (SR). However, the utility of printed ECG (pECG) records for AF detection, particularly in developing countries, remains unexplored. This study aims to assess the efficacy of an AI-based screening tool for paroxysmal AF (PAF) using pECGs during SR. Methods: We analyzed 5688 printed 12-lead SR-ECG records from 2192 patients admitted to Beijing Chaoyang Hospital between May 2011 to August 2022. All patients underwent catheter ablation for PAF (AF group) or other electrophysiological procedures (non-AF group). We developed a deep learning model to detect PAF from these printed SR-ECGs. The 2192 patients were randomly assigned to training (1972, 57.3% with PAF), validation (108, 57.4% with PAF), and test datasets (112, 57.1% with PAF). We developed an applet to digitize the printed ECG data and display the results within a few seconds. Our evaluation focused on sensitivity, specificity, accuracy, F1 score, the area under the receiver-operating characteristic curve (AUROC), and precision-recall curves (PRAUC). Results: The PAF detection algorithm demonstrated strong performance: sensitivity 87.5%, specificity 66.7%, accuracy 78.6%, F1 score 0.824, AUROC 0.871 and PRAUC 0.914. A gradient-weighted class activation map (Grad-CAM) revealed the model's tailored focus on different ECG areas for personalized PAF detection. Conclusions: The deep-learning analysis of printed SR-ECG records shows high accuracy in PAF detection, suggesting its potential as a reliable screening tool in real-world clinical practice.

An artificial intelligence-enabled electrocardiogram algorithm for the prediction of left atrial low-voltage areas in persistent atrial fibrillation.

OBJECTIVES: We aimed to construct an artificial intelligence-enabled electrocardiogram (ECG) algorithm that can accurately predict the presence of left atrial low-voltage areas (LVAs) in patients with persistent atrial fibrillation. METHODS: The study included 587 patients with persistent atrial fibrillation who underwent catheter ablation procedures between March 2012 and December 2023 and 942 scanned images of 12-lead ECGs obtained before the ablation procedures were performed. Artificial intelligence-based algorithms were used to construct models for predicting the presence of LVAs. The DR-FLASH and APPLE clinical scores for LVA prediction were calculated. We used a receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis to evaluate model performance. RESULTS: The data obtained from the participants were split into training (n = 469), validation (n = 58), and test sets (n = 60). LVAs were detected in 53.7% of all participants. Using ECG alone, the deep learning algorithm achieved an area under the ROC curve (AUROC) of 0.752, outperforming both the DR-FLASH score (AUROC = 0.610) and the APPLE score (AUROC = 0.510). The random forest classification model, which integrated a probabilistic deep learning model and clinical features, showed a maximum AUROC of 0.759. Moreover, the ECG-based deep learning algorithm for predicting extensive LVAs achieved an AUROC of 0.775, with a sensitivity of 0.816 and a specificity of 0.896. The random forest classification model for predicting extensive LVAs achieved an AUROC of 0.897, with a sensitivity of 0.862, and a specificity of 0.935. CONCLUSION: The deep learning model based exclusively on ECG data and the machine learning model that combined a probabilistic deep learning model and clinical features both predicted the presence of LVAs with a higher degree of accuracy than the DR-FLASH and the APPLE risk scores.

Data-driven inverse design of flexible pressure sensors.

Artificial skins or flexible pressure sensors that mimic human cutaneous mechanoreceptors transduce tactile stimuli to quantitative electrical signals. Conventional trial-and-error designs for such devices follow a forward structure-to-property routine, which is usually time-consuming and determines one possible solution in one run. Data-driven inverse design can precisely target desired functions while showing far higher productivity, however, it is still absent for flexible pressure sensors because of the difficulties in acquiring a large amount of data. Here, we report a property-to-structure inverse design of flexible pressure sensors, exhibiting a significantly greater efficiency than the conventional routine. We use a reduced-order model that analytically constrains the design scope and an iterative "jumping-selection" method together with a surrogate model that enhances data screening. As an exemplary scenario, hundreds of solutions that overcome the intrinsic signal saturation have been predicted by the inverse method, validating for a variety of material systems. The success in property design on multiple indicators demonstrates that the proposed inverse design is an efficient and powerful tool to target multifarious applications of flexible pressure sensors, which can potentially advance the fields of intelligent robots, advanced healthcare, and human-machine interfaces.

Association of Regular Opioid Use With Incident Dementia and Neuroimaging Markers of Brain Health in Chronic Pain Patients: Analysis of UK Biobank.

OBJECTIVES: We aimed to investigate the association of regular opioid use, compared with non-opioid analgesics, with incident dementia and neuroimaging outcomes among chronic pain patients. DESIGN: The primary design is a prospective cohort study. To triangulate evidence, we also conducted a nested case-control study analyzing opioid prescriptions and a cross-sectional study analyzing neuroimaging outcomes. SETTING AND PARTICIPANTS: Dementia-free UK Biobank participants with chronic pain and regular analgesic use. MEASUREMENTS: Chronic pain status and regular analgesic use were captured using self-reported questionnaires and verbal interviews. Opioid prescription data were obtained from primary care records. Dementia cases were ascertained using primary care, hospital, and death registry records. Propensity score-matched Cox proportional hazards analysis, conditional logistic regression, and linear regression were applied to the data in the prospective cohort, nested case-control, and cross-sectional studies, respectively. RESULTS: Prospective analyses revealed that regular opioid use, compared with non-opioid analgesics, was associated with an increased dementia risk over the 15-year follow-up (Hazard ratio [HR], 1.18 [95% confidence interval (CI): 1.08-1.30]; Absolute rate difference [ARD], 0.44 [95% CI: 0.19-0.71] per 1000 person-years; Wald χ2 = 3.65; df = 1; p <0.001). The nested case-control study suggested that a higher number of opioid prescriptions was associated with an increased risk of dementia (1 to 5 prescriptions: OR = 1.21, 95% CI: 1.07-1.37, Wald χ2 = 3.02, df = 1, p = 0.003; 6 to 20: OR = 1.27, 95% CI: 1.08-1.50, Wald χ2 = 2.93, df = 1, p = 0.003; more than 20: OR = 1.43, 95% CI: 1.23-1.67, Wald χ2 = 4.57, df = 1, p < 0.001). Finally, neuroimaging analyses revealed that regular opioid use was associated with lower total grey matter and hippocampal volumes, and higher white matter hyperintensities volumes. CONCLUSION: Regular opioid use in chronic pain patients was associated with an increased risk of dementia and poorer brain health when compared to non-opioid analgesic use. These findings imply a need for re-evaluation of opioid prescription practices for chronic pain patients and, if further evidence supports causality, provide insights into strategies to mitigate the burden of dementia.

Application of artificial intelligence in the diagnosis and treatment of cardiac arrhythmia.

The rapid growth in computational power, sensor technology, and wearable devices has provided a solid foundation for all aspects of cardiac arrhythmia care. Artificial intelligence (AI) has been instrumental in bringing about significant changes in the prevention, risk assessment, diagnosis, and treatment of arrhythmia. This review examines the current state of AI in the diagnosis and treatment of atrial fibrillation, supraventricular arrhythmia, ventricular arrhythmia, hereditary channelopathies, and cardiac pacing. Furthermore, ChatGPT, which has gained attention recently, is addressed in this paper along with its potential applications in the field of arrhythmia. Additionally, the accuracy of arrhythmia diagnosis can be improved by identifying electrode misplacement or erroneous swapping of electrode position using AI. Remote monitoring has expanded greatly due to the emergence of contactless monitoring technology as wearable devices continue to develop and flourish. Parallel advances in AI computing power, ChatGPT, availability of large data sets, and more have greatly expanded applications in arrhythmia diagnosis, risk assessment, and treatment. More precise algorithms based on big data, personalized risk assessment, telemedicine and mobile health, smart hardware and wearables, and the exploration of rare or complex types of arrhythmia are the future direction.

Relationship between HLA genetic variations, COVID-19 vaccine antibody response, and risk of breakthrough outcomes.

The rapid global distribution of COVID-19 vaccines, with over a billion doses administered, has been unprecedented. However, in comparison to most identified clinical determinants, the implications of individual genetic factors on antibody responses post-COVID-19 vaccination for breakthrough outcomes remain elusive. Here, we conducted a population-based study including 357,806 vaccinated participants with high-resolution HLA genotyping data, and a subset of 175,000 with antibody serology test results. We confirmed prior findings that single nucleotide polymorphisms associated with antibody response are predominantly located in the Major Histocompatibility Complex region, with the expansive HLA-DQB1*06 gene alleles linked to improved antibody responses. However, our results did not support the claim that this mutation alone can significantly reduce COVID-19 risk in the general population. In addition, we discovered and validated six HLA alleles (A*03:01, C*16:01, DQA1*01:02, DQA1*01:01, DRB3*01:01, and DPB1*10:01) that independently influence antibody responses and demonstrated a combined effect across HLA genes on the risk of breakthrough COVID-19 outcomes. Lastly, we estimated that COVID-19 vaccine-induced antibody positivity provides approximately 20% protection against infection and 50% protection against severity. These findings have immediate implications for functional studies on HLA molecules and can inform future personalised vaccination strategies.

Medication Usage Record-Based Predictive Modeling of Neurodevelopmental Abnormality in Infants under One Year: A Prospective Birth Cohort Study.

Early identification of children with neurodevelopmental abnormality is a major challenge, which is crucial for improving symptoms and preventing further decline in children with neurodevelopmental abnormality. This study focuses on developing a predictive model with maternal sociodemographic, behavioral, and medication-usage information during pregnancy to identify infants with abnormal neurodevelopment before the age of one. In addition, an interpretable machine-learning approach was utilized to assess the importance of the variables in the model. In this study, artificial neural network models were developed for the neurodevelopment of five areas of infants during the first year of life and achieved good predictive efficacy in the areas of fine motor and problem solving, with median AUC = 0.670 (IQR: 0.594, 0.764) and median AUC = 0.643 (IQR: 0.550, 0.731), respectively. The final model for neurodevelopmental abnormalities in any energy region of one-year-old children also achieved good prediction performance. The sensitivity is 0.700 (IQR: 0.597, 0.797), the AUC is 0.821 (IQR: 0.716, 0.833), the accuracy is 0.721 (IQR: 0.696, 0.739), and the specificity is 0.742 (IQR: 0.680, 0.748). In addition, interpretable machine-learning methods suggest that maternal exposure to drugs such as acetaminophen, ferrous succinate, and midazolam during pregnancy affects the development of specific areas of the offspring during the first year of life. This study established predictive models of neurodevelopmental abnormality in infants under one year and underscored the prediction value of medication exposure during pregnancy for the neurodevelopmental outcomes of the offspring.

Cardiac murmur grading and risk analysis of cardiac diseases based on adaptable heterogeneous-modality multi-task learning.

Cardiovascular disease (CVDs) has become one of the leading causes of death, posing a significant threat to human life. The development of reliable Artificial Intelligence (AI) assisted diagnosis algorithms for cardiac sounds is of great significance for early detection and treatment of CVDs. However, there is scarce research in this field. Existing research mainly faces three major challenges: (1) They mainly limited to murmur classification and cannot achieve murmur grading, but attempting both classification and grading may lead to negative effects between different multi-tasks. (2) They mostly pay attention to unstructured cardiac sound modality and do not consider the structured demographic modality, as it is difficult to balance the influence of heterogeneous modalities. (3) Deep learning methods lack interpretability, which makes it challenging to apply them clinically. To tackle these challenges, we propose a method for cardiac murmur grading and cardiac risk analysis based on heterogeneous modality adaptive multi-task learning. Specifically, a Hierarchical Multi-Task learning-based cardiac murmur detection and grading method (HMT) is proposed to prevent negative interference between different tasks. In addition, a cardiac risk analysis method based on Heterogeneous Multi-modal feature impact Adaptation (HMA) is also proposed, which transforms unstructured modality into structured modality representation, and utilizes an adaptive mode weight learning mechanism to balance the impact between unstructured modality and structured modality, thus enhancing the performance of cardiac risk prediction. Finally, we propose a multi-task interpretability learning module that incorporates an important evaluation using random masks. This module utilizes SHAP graphs to visualize crucial murmur segments in cardiac sound and employs a multi-factor risk decoupling model based on nomograms. And then we gain insights into the cardiac disease risk in both pre-decoupled multi-modality and post-decoupled single-modality scenarios, thus providing a solid foundation for AI assisted cardiac murmur grading and risk analysis. Experimental results on a large real-world CirCor DigiScope PCG dataset demonstrate that the proposed method outperforms the state-of-the-art (SOTA) method in murmur detection, grading, and cardiac risk analysis, while also providing valuable diagnostic evidence.

HLA-C*07:01 and HLA-DQB1*02:01 protect against white matter hyperintensities and deterioration of cognitive function: A population-based cohort study.

BACKGROUND: Neuroinflammation and aberrant immune regulation are increasingly implicated in the pathophysiology of white matter hyperintensities (WMH), an imaging marker of cerebrovascular pathologies and predictor of cognitive impairment. The role of human leukocyte antigen (HLA) genes, critical in immunoregulation and associated with susceptibility to neurodegenerative diseases, in WMH pathophysiology remains unexplored. METHODS: We performed association analyses between classical HLA alleles and WMH volume, derived from MRI scans of 38 302 participants in the UK Biobank. To identify independent functional alleles driving these associations, we conducted conditional forward stepwise regression and lasso regression. We further investigated whether these functional alleles showed consistent associations with WMH across subgroups characterized by varying levels of clinical determinants. Additionally, we validated the clinical relevance of the identified alleles by examining their association with cognitive function (n = 147 549) and dementia (n = 460 029) in a larger cohort. FINDINGS: Four HLA alleles (DQB1*02:01, DRB1*03:01, C*07:01, and B*08:01) showed an association with reduced WMH volume after Bonferroni correction for multiple comparisons. Among these alleles, DQB1*02:01 exhibited the most significant association (ß = -0.041, 95 % CI: -0.060 to -0.023, p = 1.04 × 10-5). Forward selection and lasso regression analyses indicated that DQB1*02:01 and C*07:01 primarily drove this association. The protective effect against WMH conferred by DQB1*02:01 and C*07:01 persisted in clinically relevant subgroups, with a stronger effect observed in older participants. Carrying DQB1*02:01 and C*07:01 was associated with higher cognitive function, but no association with dementia was found. INTERPRETATION: Our population-based findings support the involvement of immune-associated mechanisms, particularly both HLA class I and class II genes, in the pathogenesis of WMH and subsequent consequence of cognitive functions.

Machine learning in predicting T-score in the Oxford classification system of IgA nephropathy.

Background: Immunoglobulin A nephropathy (IgAN) is one of the leading causes of end-stage kidney disease (ESKD). Many studies have shown the significance of pathological manifestations in predicting the outcome of patients with IgAN, especially T-score of Oxford classification. Evaluating prognosis may be hampered in patients without renal biopsy. Methods: A baseline dataset of 690 patients with IgAN and an independent follow-up dataset of 1,168 patients were used as training and testing sets to develop the pathology T-score prediction (T pre) model based on the stacking algorithm, respectively. The 5-year ESKD prediction models using clinical variables (base model), clinical variables and real pathological T-score (base model plus T bio), and clinical variables and T pre (base model plus T pre) were developed separately in 1,168 patients with regular follow-up to evaluate whether T pre could assist in predicting ESKD. In addition, an external validation set consisting of 355 patients was used to evaluate the performance of the 5-year ESKD prediction model using T pre. Results: The features selected by AUCRF for the T pre model included age, systolic arterial pressure, diastolic arterial pressure, proteinuria, eGFR, serum IgA, and uric acid. The AUC of the T pre was 0.82 (95% CI: 0.80-0.85) in an independent testing set. For the 5-year ESKD prediction model, the AUC of the base model was 0.86 (95% CI: 0.75-0.97). When the T bio was added to the base model, there was an increase in AUC [from 0.86 (95% CI: 0.75-0.97) to 0.92 (95% CI: 0.85-0.98); P = 0.03]. There was no difference in AUC between the base model plus T pre and the base model plus T bio [0.90 (95% CI: 0.82-0.99) vs. 0.92 (95% CI: 0.85-0.98), P = 0.52]. The AUC of the 5-year ESKD prediction model using T pre was 0.93 (95% CI: 0.87-0.99) in the external validation set. Conclusion: A pathology T-score prediction (T pre) model using routine clinical characteristics was constructed, which could predict the pathological severity and assist clinicians to predict the prognosis of IgAN patients lacking kidney pathology scores.

Estimating causal effects of physical disability and number of comorbid chronic diseases on risk of depressive symptoms in an elderly Chinese population: a machine learning analysis of cross-sectional baseline data from the China longitudinal ageing social survey.

OBJECTIVE: This study aimed to explore the causal effects of physical disability and number of comorbid chronic diseases on depressive symptoms in an elderly Chinese population. DESIGN, SETTING AND ANALYSIS: Cross-sectional, baseline data were obtained from the China Longitudinal Ageing Social Survey, a stratified, multistage, probabilistic sampling survey conducted in 2014 that covers 28 of 31 provincial areas in China. The causal effects of physical disability and number of comorbid chronic diseases on depressive symptoms were analysed using the conditional average treatment effect method of machine learning. The causal effects model's adjustment was made for age, gender, residence, marital status, educational level, ethnicity, wealth quantile and other factors. OUTCOME: Assessment of the causal effects of physical disability and number of comorbid chronic diseases on depressive symptoms. PARTICIPANTS: 7496 subjects who were 60 years of age or older and who answered the questions on depressive symptoms and other independent variables of interest in a survey conducted in 2014 were included in this study. RESULTS: Physical disability and number of comorbid chronic diseases had causal effects on depressive symptoms. Among the subjects who had one or more functional limitations, the probability of depressive symptoms increased by 22% (95% CI 19% to 24%). For the subjects who had one chronic disease and those who had two or more chronic diseases, the possibility of depressive symptoms increased by 13% (95% CI 10% to 15%) and 20% (95% CI 18% to 22%), respectively. CONCLUSION: This study provides evidence that the presence of one or more functional limitations affects the occurrence of depressive symptoms among elderly people. The findings of our study are of value in developing programmes that are designed to identify elderly individuals who have physical disabilities or comorbid chronic diseases to provide early intervention.

Home monitoring for clinically suspected obstructive sleep apnea in pregnancy.

STUDY OBJECTIVES: To determine if a home sleep apnea test (HSAT) using a type III portable monitor (PM), Nox-T3 (Nox Medical, Inc., Reykjavik, Iceland), detects obstructive sleep apnea in pregnant women. METHODS: Ninety-two pregnant women (34.5 ± 4.3 years; gestational age 25.4 ± 8.9 weeks; body mass index 29.9 ± 4.7 kg/m2) with suspected obstructive sleep apnea underwent HSAT with the Nox-T3 PM followed by overnight polysomnography (PSG) and PM recording simultaneously in the laboratory within 1 week. PMs were scored automatically and manually using a 3% criteria and compared with PSGs scored by following guidelines. RESULTS: Apnea-hypopnea indexes were 8.56 ± 10.42, 8.19 ± 13.79, and 8.71 ± 14.19 events/h on HSAT, in-laboratory PM recording, and PSG (P = .955), respectively. Bland-Altman analysis of the apnea-hypopnea index on PSG vs HSAT showed a mean difference (95% confidence interval) of -0.15 (-1.83, 1.53); limits of agreement (± 2 SD) were -16.26 to 16.56 events/h. Based on a threshold apnea-hypopnea index ≥ 5 events/h, HSAT had 91% sensitivity, 85% specificity, 84% positive-predictive value, and 92% negative-predictive value compared with PSG. When comparing the simultaneous recordings, closer agreement was observed. Automated vs manual analysis of PM showed no significant difference. CONCLUSIONS: A type III PM had an acceptable failure rate and high diagnostic performance operating as a reasonable alternative for in-laboratory PSG in pregnant women. CITATION: Wang J, Zhang C, Xu L, et al. Home monitoring for clinically suspected obstructive sleep apnea in pregnancy. J Clin Sleep Med. 2023;19(11):1951-1960.

Self-Supervised Time Series Representation Learning via Cross Reconstruction Transformer.

Since labeled samples are typically scarce in real-world scenarios, self-supervised representation learning in time series is critical. Existing approaches mainly employ the contrastive learning framework, which automatically learns to understand similar and dissimilar data pairs. However, they are constrained by the request for cumbersome sampling policies and prior knowledge of constructing pairs. Also, few works have focused on effectively modeling temporal-spectral correlations to improve the capacity of representations. In this article, we propose the cross reconstruction transformer (CRT) to solve the aforementioned issues. CRT achieves time series representation learning through a cross-domain dropping-reconstruction task. Specifically, we obtain the frequency domain of the time series via the fast Fourier transform (FFT) and randomly drop certain patches in both time and frequency domains. Dropping is employed to maximally preserve the global context while masking leads to the distribution shift. Then a Transformer architecture is utilized to adequately discover the cross-domain correlations between temporal and spectral information through reconstructing data in both domains, which is called Dropped Temporal-Spectral Modeling. To discriminate the representations in global latent space, we propose instance discrimination constraint (IDC) to reduce the mutual information between different time series samples and sharpen the decision boundaries. Additionally, a specified curriculum learning (CL) strategy is employed to improve the robustness during the pretraining phase, which progressively increases the dropping ratio in the training process. We conduct extensive experiments to evaluate the effectiveness of the proposed method on multiple real-world datasets. Results show that CRT consistently achieves the best performance over existing methods by 2%-9%. The code is publicly available at https://github.com/BobZwr/Cross-Reconstruction-Transformer.

A Ranking-Based Cross-Entropy Loss for Early Classification of Time Series.

Early classification tasks aim to classify time series before observing full data. It is critical in time-sensitive applications such as early sepsis diagnosis in the intensive care unit (ICU). Early diagnosis can provide more opportunities for doctors to rescue lives. However, there are two conflicting goals in the early classification task-accuracy and earliness. Most existing methods try to find a balance between them by weighing one goal against the other. But we argue that a powerful early classifier should always make highly accurate predictions at any moment. The main obstacle is that the key features suitable for classification are not obvious in the early stage, resulting in the excessive overlap of time series distributions in different time stages. The indistinguishable distributions make it difficult for classifiers to recognize. To solve this problem, this article proposes a novel ranking-based cross-entropy () loss to jointly learn the feature of classes and the order of earliness from time series data. In this way, can help classifier to generate probability distributions of time series in different stages with more distinguishable boundary. Thus, the classification accuracy at each time step is finally improved. Besides, for the applicability of the method, we also accelerate the training process by focusing the learning process on high-ranking samples. Experiments on three real-world datasets show that our method can perform classification more accurately than all baselines at all moments.

Continuous diagnosis and prognosis by controlling the update process of deep neural networks.

Continuous diagnosis and prognosis are essential for critical patients. They can provide more opportunities for timely treatment and rational allocation. Although deep-learning techniques have demonstrated superiority in many medical tasks, they frequently forget, overfit, and produce results too late when performing continuous diagnosis and prognosis. In this work, we summarize the four requirements; propose a concept, continuous classification of time series (CCTS); and design a training method for deep learning, restricted update strategy (RU). The RU outperforms all baselines and achieves average accuracies of 90%, 97%, and 85% on continuous sepsis prognosis, COVID-19 mortality prediction, and eight disease classifications, respectively. The RU can also endow deep learning with interpretability, exploring disease mechanisms through staging and biomarker discovery. We find four sepsis stages, three COVID-19 stages, and their respective biomarkers. Further, our approach is data and model agnostic. It can be applied to other diseases and even in other fields.

Trajectory of worst pain within the first two weeks following pelvic and sacral tumor surgery and long-term outcome: a pilot observational prospective cohort study.

BACKGROUND: Pain management after pelvic and sacral tumor surgery is challenging and requires a multidisciplinary and multimodal approach. Few data on postoperative pain trajectories have been reported after pelvic and sacral tumor surgery. The aim of this pilot study was to determine pain trajectories within the first 2 weeks after surgery and explore the impact on long-term pain outcomes. METHODS: Patients scheduled for pelvic and sacral tumor surgery were prospectively recruited. Worst/average pain scores were evaluated postoperatively using questions adapted from the Revised American Pain Society Patient Outcome Questionnaire (APS-POQ-R) until pain resolution was reached or up to 6 months after surgery. Pain trajectories over the first 2 weeks were compared using the k-means clustering algorithm. Whether pain trajectories were associated with long-term pain resolution and opioid cessation was assessed using Cox regression analysis. RESULTS: A total of 59 patients were included. Two distinct groups of trajectories for worst and average pain scores over the first 2 weeks were generated. The median pain duration in the high vs low pain group was 120.0 (95% CI [25.0, 215.0]) days vs 60.0 (95% CI [38.6, 81.4]) days (log rank p = 0.037). The median time to opioid cessation in the high vs low pain group was 60.0 (95% CI [30.0, 90.0]) days vs 7.0 (95% CI [4.7, 9.3]) days (log rank p < 0.001). After adjusting for patient and surgical factors, the high pain group was independently associated with prolonged opioid cessation (hazard ratio [HR] 2.423, 95% CI [1.254, 4.681], p = 0.008) but not pain resolution (HR 1.557, 95% CI [0.748, 3.243], p = 0.237). CONCLUSIONS: Postoperative pain is a significant problem among patients undergoing pelvic and sacral tumor surgery. High pain trajectories during the first 2 weeks after surgery were associated with delayed opioid cessation. Research is needed to explore interventions targeting pain trajectories and long-term pain outcomes. TRIAL REGISTRATION: The trial was registered at ClinicalTrials.gov ( NCT03926858 , 25/04/2019).

Development of Expert-Level Classification of Seizures and Rhythmic and Periodic Patterns During EEG Interpretation.

BACKGROUND AND OBJECTIVES: Seizures (SZs) and other SZ-like patterns of brain activity can harm the brain and contribute to in-hospital death, particularly when prolonged. However, experts qualified to interpret EEG data are scarce. Prior attempts to automate this task have been limited by small or inadequately labeled samples and have not convincingly demonstrated generalizable expert-level performance. There exists a critical unmet need for an automated method to classify SZs and other SZ-like events with expert-level reliability. This study was conducted to develop and validate a computer algorithm that matches the reliability and accuracy of experts in identifying SZs and SZ-like events, known as "ictal-interictal-injury continuum" (IIIC) patterns on EEG, including SZs, lateralized and generalized periodic discharges (LPD, GPD), and lateralized and generalized rhythmic delta activity (LRDA, GRDA), and in differentiating these patterns from non-IIIC patterns. METHODS: We used 6,095 scalp EEGs from 2,711 patients with and without IIIC events to train a deep neural network, SPaRCNet, to perform IIIC event classification. Independent training and test data sets were generated from 50,697 EEG segments, independently annotated by 20 fellowship-trained neurophysiologists. We assessed whether SPaRCNet performs at or above the sensitivity, specificity, precision, and calibration of fellowship-trained neurophysiologists for identifying IIIC events. Statistical performance was assessed by the calibration index and by the percentage of experts whose operating points were below the model's receiver operating characteristic curves (ROCs) and precision recall curves (PRCs) for the 6 pattern classes. RESULTS: SPaRCNet matches or exceeds most experts in classifying IIIC events based on both calibration and discrimination metrics. For SZ, LPD, GPD, LRDA, GRDA, and "other" classes, SPaRCNet exceeds the following percentages of 20 experts-ROC: 45%, 20%, 50%, 75%, 55%, and 40%; PRC: 50%, 35%, 50%, 90%, 70%, and 45%; and calibration: 95%, 100%, 95%, 100%, 100%, and 80%, respectively. DISCUSSION: SPaRCNet is the first algorithm to match expert performance in detecting SZs and other SZ-like events in a representative sample of EEGs. With further development, SPaRCNet may thus be a valuable tool for an expedited review of EEGs. CLASSIFICATION OF EVIDENCE: This study provides Class II evidence that among patients with epilepsy or critical illness undergoing EEG monitoring, SPaRCNet can differentiate (IIIC) patterns from non-IIIC events and expert neurophysiologists.

Self-sovereign identity empowered non-fungible patient tokenization for health information exchange using blockchain technology.

BACKGROUND: Patient tokenization is a novel approach that allows anonymous patient-level linkage across healthcare facilities, minimizing the risk of breaching protected health information in health information exchange (HIE). Most patient tokenization is the centralized approach that is unable to address data security concerns fundamentally. Non-Fungible Tokens (NFT), which are non-transferable cryptographic assets on the blockchain, have the potential to provide secure, decentralized, and trustworthy patient tokenization. Self-Sovereign Identity (SSI) is a user-centric approach to verify the ownership of NFTs in a decentralized manner. METHODS: We have developed a blockchain architecture that contains four modules: (1) Creation module for NFTs creation, (2) Linkage module to link the local patients' accounts to their NFTs, (3) Authentication module that allows patients to permit healthcare providers to access their token, and (4) Exchange module, which involves the HIE process and the validation of the legitimacy of the token through SSI. RESULTS: A case study has been conducted on the proposed architecture. Over 3 million transactions have been completed successfully with a blockchain validation and written time of 1.17 s on average. A stability test has also been conducted with a higher throughput of 200 transactions per second running for an hour with an average transaction processing time of 1.42 s. CONCLUSIONS: This study proposed a blockchain architecture that achieves SSI-enabled NFT-based patient tokenization. Our architecture design, implementation, and case studies have demonstrated the feasibility and potential of NFT with SSI to establish a secure, transparent, and patient-centric identity management and HIE.