AMIA ... Annual Symposium proceedings. AMIA Symposium最新文献

Clinical tools for tracking functional decline in amyotrophic lateral sclerosis (ALS) rely on in-clinic guided assessments, such as the gold standard ALS Functional Rating Scale Revised (ALSFRS-R) instrument, thus limiting the frequency of collection and potentially delaying needed treatments. As such, ALS clinicians may miss subtle yet critical shifts inpatient health -pointing to the needfor objective and continuous capturing of day-to-day functional status. In-home health sensors could supplement clinical instruments with more frequent, quantitative measurements as early indicators of change. Using the XGBoost regressor in base learning, we explore interpolation techniques for aligning monthly ALSFRS-R assessment targets with high frequency sensor-based health features. We evaluated 9 interpolation models, which demonstrate superior prediction of ALSFRS-R scores compared to traditional clinical scale estimates based on linear slope. This pilot work provides a practical approach of modeling mixed-frequency data and shows the potential of using sensor-based health estimates as sensitive prognostic markers.

While there is a growing recognition of the association between depression and asthma, few studies have leveraged deep learning-based (DL-based) models in a retrospective cohort study with a large sample size. We analyzed the association between depression and asthma among 239,161 participants of the All of Us Research Program through DL-based, logistic regression, and Cox Proportional Hazards (CoxPH) models. We used SHAP values to help interpret DL-based models and c-index to evaluate model performance. Results suggest a significant odds ratio for depression in asthma. The c-indices for the CoxPH, DeepSurv, and DeepHit models were 0.619, 0.625, and 0.596, respectively. SHAP indicated a different set of important variables when compared with the CoxPH model. In conclusion, we provide strong evidence of a positive relationship between depression and asthma. Also, DL-based models did not outperform the CoxPH model on the c-index. Sex at birth and income may play important roles in occurrence of depression in asthma patients.

The advent of clinical language models integrated into electronic health records (EHR) for clinical decision support has marked a significant advancement, leveraging the depth of clinical notes for improved decision-making. Despite their success, the potential vulnerabilities of these models remain largely unexplored. This paper delves into the realm of backdoor attacks on clinical language models, introducing an innovative attention-based backdoor attack method, BadCLM (Bad Clinical Language Models). This technique clandestinely embeds a backdoor within the models, causing them to produce incorrect predictions when a pre-defined trigger is present in inputs, while functioning accurately otherwise. We demonstrate the efficacy of BadCLM through an in-hospital mortality prediction task with MIMIC III dataset, showcasing its potential to compromise model integrity. Our findings illuminate a significant security risk in clinical decision support systems and pave the way for future endeavors in fortifying clinical language models against such vulnerabilities.

RealRisks is a decision aid that integrates patient-generated and electronic health record (EHR) data using Fast Healthcare Interoperability Resources (FHIR). It offers modules to enhance understanding of breast cancer risk and a way for individuals to review and modify their EHR data before it is used in their personal risk assessment. RealRisks intends to encourage high-risk patients to take risk-reducing measures. To better understand how patients understand risk and barriers to action, we conducted in-depth interviews as part of a usability study to assess the clarity and interpretability of RealRisks. Overall, participants demonstrated an improved understanding of breast cancer risk after using RealRisks. However, challenges were noted for certain concepts, in particular, lifetime risk, how benign breast disease affects your risk, and the differences between hereditary, sporadic, and familial cancer. The EHR download feature was well-received, but some raised concerns about insurance and privacy/security.

Although digital health tools are increasingly common for managing health conditions, these applications are often developed without consideration of differences across user populations. A reproducible framework is needed to support tailoring applications to include cultural considerations, potentially leading to better adoption and more effective use. As a first step, this study captures a snapshot of Black women's barriers and facilitators in using digital health products for self-management of hypertensive disorders of pregnancy (HDP). One-on-one semi-structured interviews were conducted with 17 Black pregnant women with HDP. We established a unique model for cultural tailoring with these experiences using Black feminist theory and the CDC's Social-Ecological Model (SEM). 38 themes across the four levels of SEM were found through grounded theory. These themes can inform the feature development of a digital health intervention. Future work will instantiate and validate a framework that provides theoretical constructs for developing culturally tailored digital health interventions.

Characterizing differential responses to Alzheimer's disease (AD) drugs will provide better insights into personalized treatment strategies. Our study aims to identify heterogeneous treatment effects and pre-treatment features that moderate the treatment effect of Galantamine, Bapineuzumab, and Semagacestat from completed trial data. The causal forest method can capture heterogeneity in treatment responses. We applied causal forest modeling to estimate the treatment effect and identify efficacy moderators in each trial. We found several patient's pretreatment conditions that determined treatment efficacy. For example, in Galantamine trials, whole brain volume (1092.54 vs. 1060.67 ml, P < .001) and right hippocampal volume (2.43e-3 vs. 2.79e-3, P < .001) are significantly different between responsive and non-responsive subgroups. Overall, our implementation of causal forests in AD clinical trials reveals the heterogeneous treatment effects and different moderators for AD drug responses, highlighting promising personalized treatment based on patient-specific characteristics in AD research and drug development.

Alzheimer's disease (AD) manifests with varying progression rates across individuals, necessitating the understanding of their intricate patterns of cognition decline that could contribute to effective strategies for risk monitoring. In this study, we propose an innovative interpretable population graph network framework for identifying rapid progressors of AD by utilizing patient information from electronic health-related records in the UK Biobank. To achieve this, we first created a patient similarity graph, in which each AD patient is represented as a node; and an edge is established by patient clinical characteristics distance. We used graph neural networks (GNNs) to predict rapid progressors of AD and created a GNN Explainer with SHAP analysis for interpretability. The proposed model demonstrates superior predictive performance over the existing benchmark approaches. We also revealed several clinical features significantly associated with the prediction, which can be used to aid in effective interventions for the progression of AD patients.

This study investigates the use of ChatGPT to support clinical teams with limited expertise in generating synthetic data for breast cancer research. It assesses ChatGPT's application, focusing on effective prompting and best practices for creating high-fidelity synthetic data. The research compares the generated synthetic data to the Wisconsin Breast Cancer Dataset through statistical analysis, structural similarity metrics, and machine learning performance. Results indicate that the quality of prompts and generation techniques significantly affects the data's fidelity. The study highlights the critical role of prompt engineering and data synthesis techniques in producing accurate synthetic data for healthcare research, underscoring the need for precise prompts and generation methods to maintain data integrity in sensitive areas like cancer research.

Remote interventionists in the novel ASSETs for Aging in Place demonstration project rely on smart home and wearable sensor data to understand daily behavior patterns of older adult clients discharged from nursing homes to the community and to inform coaching during telehealth visits to help clients self-manage personal goals for health and independence. We employed contextual inquiry during design of the interventionist dashboard to support the new ASSETs program. Focus groups with interventionists and leadership characterized themes for primary dashboard goals, interface and technology needs, and data collection expectations. Four contextual inquiry sessions with interventionists characterized user goals, barriers, and standardized user workflows. We articulated a sequential discovery process for user requirements that can be replicated in dashboard design for future remote service delivery programs that will rely on sensor data and telehealth visits.

To assist physicians in predicting diseases, most natural language processing (NLP) models have focused on progress notes in electronic medical records with full descriptions from the initial stage of patient diagnosis to the final stage of discharge. However, accurately predicting diseases in the early stage using initial notes is challenging due to limited information. To address this, a text-numerical hybrid method is developed to improve disease prediction accuracy. The method identifies "Reliably predicted diseases (RPD)" that can be robustly predicted in the NLP and Random Forest models even if there are missing values in the numerical data or the amount of text data is small. Results show that, among the predicted disease groups of the two models, diseases matching the RPD are preferentially adopted and integrated. Precision@10 reveals that our developed method has a relatively higher accuracy of 67.0% than the traditional NLP model.