Long-term care facilities face a critical shortage of nursing staff and an increasing administrative burden, reducing time for direct patient care. Generative artificial intelligence offers a promising solution to automate administrative tasks and support caregivers. This paper evaluates the relevance of using a fine-tuned large language model (LLM) to address these challenges. Interviews with healthcare professionals identified key needs, leading to the selection of two use cases: caregiver-patient communication assistance and medical record summarization. To comply with privacy and security constraints, the model was deployed in an embedded scenario. Performance evaluations showed significant improvements in BLEU and ROUGE metrics for both use cases, demonstrating enhanced accuracy. This study demonstrates the feasibility of leveraging LLMs to streamline workflows, reduce administrative strain, and improve operational efficiency. This work highlights the potential for broader AI applications in long-term care, paving the way for better working conditions for caregivers and improved patient care quality.
Large language models (LLMs) excel in natural language processing (NLP) but struggle with domain-specific complexities in electronic health records (EHRs). We demonstrate that retrieval-augmented generation (RAG) enhances LLMs for dietary supplement (DS) information extraction. By testing models like Llama-3 with diverse retrievers on tasks including entity recognition and usage classification, task-aligned retrieval outperforms reliance on model size or specialization. Smaller general models paired with optimized retrievers match or exceed specialized counterparts-structured retrieval aids complex tasks (e.g., triple extraction), while semantic retrieval improves classification. Results challenge assumptions that larger or domain-specific models are superior, emphasizing dynamic knowledge integration over brute-force scaling. This approach offers practical strategies for clinical NLP, enabling efficient EHR analysis without massive resources. Prioritizing retrieval strategies over model size advances tools for evidence-based healthcare, highlighting adaptability and cost-effectiveness in real-world medical applications.
Researchers have developed pharmacogenomics datasets for various purposes, such as biomarker identification, yet drug response prediction models often underperform due to dataset inconsistencies. These variations arise from inter-tumoral heterogeneity, experimental conditions, and cell subtype complexity, limiting model generalizability. To address this, we propose a computational model based on Aggregated Learning (AL) to enhance drug response prediction by learning from inconsistencies across multiple datasets. Our model minimizes discrepancies by training on overlapping inconsistent data points from three pharmacogenomic datasets-CCLE, GDSC2, and gCSI. Compared to four baseline methods-Selecting Better (SB), Result Average (RA), Combining Data (CD), and Model Average (MA)-our approach achieved superior performance with lower Mean Absolute Error (MAE) scores: 0.090 (CCLE-GDSC), 0.096 (CCLE-gCSI), and 0.081 (GDSC-gCSI). These results demonstrate that addressing inconsistencies enhances prediction accuracy and generalizability, making our model a promising solution for robust drug response predictions.
Semantic similarity measures (SSMs) are widely used in biomedical research but remain underutilized in pharmacovigilance. This study evaluates six ontology-based SSMs for clustering MedDRA Preferred Terms (PTs) in drug safety data. Using the Unified Medical Language System (UMLS), we assess each method's ability to group PTs around medically meaningful centroids. A high-throughput framework was developed with a Java API and Python/R interfaces support large-scale similarity computations. Results show that while path-based methods perform moderately with F1 scores of 0.36 for WUPALMER and 0.28 for LCH, intrinsic information content (IC)-based measures, especially INTRINSIC_LIN and SOKAL, consistently yield better clustering accuracy (F1 Score of 0.403). Validated against expert review and standard MedDRA queries (SMQs), our findings highlight the promise of IC-based SSMs in enhancing pharmacovigilance workflows by improving early signal detection and reducing manual review.
Clinical decision-making is inherently complex and fast-paced, particularly in emergency departments (EDs) where rapid and high-stakes decisions are made. Clinical Decision Rules (CDRs) are standardized evidence-based tools that combine signs, symptoms, and clinical variables into decision trees to make consistent and accurate diagnoses. CDR usage is often hindered by the clinician's cognitive load, limiting their ability to quickly recall and apply the appropriate rules. We introduce CDR-Agent, a novel LLM-based system designed to enhance ED decision-making by autonomously identifying and applying the most appropriate CDRs based on unstructured clinical notes. To validate CDR-Agent, we curated two novel ED datasets: synthetic and CDR-Bench, although CDR-Agent is applicable to non ED clinics. CDR-Agent achieves a 56.3% (synthetic) and 8.7% (CDR-Bench) accuracy gain relative to the standalone LLM baseline in CDR selection, with overall prediction accuracy improvements of 134.0% (synthetic) and 20.4% (CDR-Bench). Moreover, CDR-Agent significantly reduces computational overhead.
This work introduces the Sequential Multiple Instance Learning (SMIL) framework, addressing the challenge of interpreting sequential, variable-length sequences of medical images with a single diagnostic label. Diverging from traditional MIL approaches that treat image sequences as unordered sets, SMIL systematically integrates the sequential nature of clinical imaging. We develop a bidirectional Transformer architecture, BiSMIL, that optimizes for both early and final prediction accuracies through a novel training procedure to balance diagnostic accuracy with operational efficiency. We evaluated BiSMIL on three medical image datasets to demonstrate that it simultaneously achieves state-of-the-art final accuracy and superior performance in early prediction accuracy, requiring 30-50% fewer images for a similar level of performance compared to existing models. Additionally, we introduce SMILU, an interpretable uncertainty metric that outperforms traditional metrics in identifying challenging instances.
De-identified health data are frequently used in research. As AI advances heighten the risk of re-identification, it is important to respond to concerns about transparency, data privacy, and patient preferences. However, few practical and user-friendly solutions exist. We developed iAGREE, a patient-centered electronic consent management portal that allows patients to set granular preferences for sharing electronic health records and biospecimens with researchers. To refine the iAGREE portal, we conducted a mixed-methods usability evaluation with 40 participants from three U.S. health systems. Our results show that the portal received highly positive usability feedback. Moreover, participants identified areas for improvement, suggested actionable enhancements, and proposed additional features to better support informed granular consent while reducing patient burden. Insights from this study may inform further improvements to iAGREE and provide practical guidance for designing patient-centered consent management tools.
Cyanosis is a discoloration of the skin arising from deoxygenated hemoglobin in the blood, caused by heart, lung, and blood diseases and treated with interventions including supplemental oxygen therapy. Cyanosis presents as a bluish discoloration in light-skinned patients, but as a gray or white discoloration in dark-skinned patients. While prior work hints at the under-identification of cyanosis for people with black and brown skin, in this study, we quantify differences in cyanosis identification rates and associated clinical treatments by race/ethnicity. Leveraging EHR datasets from two hospital systems, we extract cyanosis mentions from clinical notes and compare cyanosis documentation rates by documented race/ethnicity. Cyanosis documentation was significantly less frequent for Black patients than White patients after adjusting for confounders. We measure impacts of cyanosis identification on provision of oxygen, vasopressors, and fluids. Adjusting for severity of a patient's condition, documentation of cyanosis was associated with faster provision of oxygen.
Mild Cognitive Impairment (MCI) is an early stage of dementia characterized by cognitive decline and behavioral changes. Early detection is crucial for timely interventions, improved clinical trial cohort selection, and the development of targeted therapies. Linguistic markers have recently emerged as a non-invasive, cost-effective method for MCI detection. This study analyzes linguistic markers from conversations between participants and healthcare professionals to distinguish MCI from cognitively normal (NL) individuals. The dynamics of multiple conversations of a subject carry fine-granular linguistic change over time and expect to greatly enhance detection accuracy. However, individual variations in speaking styles pose challenges for learning cognitive characteristics from temporal sequences of conversations. To address this, we propose a temporal harmonization method to mitigate distributional differences in linguistic features across subjects, improving model generalization. Our results show that machine learning models leveraging subject-invariant harmonized temporal features greatly improve the prediction performance of MCI detection from multiple conversations.
In vivo confocal microscopy (IVCM) assesses corneal innervation in the sub-basal nerve plexus but is typically quantified manually from a single z-scan, limiting biomarker extrapolation for diagnosing limbal stem cell deficiency (LSCD). We developed an automated 3D reconstruction method of IVCM image volumes to improve sub-basal nerve density quantification. Our dataset comprised 99 IVCM stacks from 63 LSCD eyes (51 patients) and 23 stacks from 15 normal eyes. We designed an image registration algorithm combining phase correlation and homography transformation, which achieved a pairwise image correlation of 0.69 and mutual information of 0.60, significantly outperforming manual registration (0.60 and 0.43, respectively; p<0.001). Validation on an independent dataset of 325 volume scans from 24 eyes of 12 unilateral, severe LSCD patients yielded a correlation of 0.75 and MI of 0.76. This method enhances sequential IVCM scan alignment and supports more accurate, reproducible 3D evaluation of LSC biomarkers.
Identifying social drivers of health (SDoH) can help healthcare professionals connect patients with community resources that impact health outcomes. However, screening and referral require time and effort, highlighting the potential for electronic health record (EHR) support to improve efficacy. We surveyed primary care clinicians and staff (n=122) about current SDoH screening and referral practices, as well as perceived barriers and benefits for standardizing those practices with EHR support. Although the majority currently screen and refer patients for SDoH at least occasionally, most respondents document in the EHR using unstructured formats and rate current practices as only moderately feasible and acceptable. While the majority feel competent with SDoH screening, key barriers included time constraints, lack of dedicated staff, and a desire for access to more community resources. Findings underscore the need for an EHR-based standardized screening and referral implementation with flexibility to adapt workflows in different clinic settings and clinical positions.
Reliable drug safety reference databases are essential for pharmacovigilance, yet existing resources like SIDER are outdated and static. We introduce PVLens, an automated system that extracts labeled safety information from FDA Structured Product Labels (SPLs) and maps terms to MedDRA. PVLens integrates automation with expert oversight through a web-based review tool. In validation against 97 drug labels, PVLens achieved an F1 score of 0.882, with high recall (0.983) and moderate precision (0.799). By offering a scalable, more accurate and continuously updated alternative to SIDER, PVLens enhances real-time pharamcovigilance with improved accuracy and contemporaneous insights.
American Indian and Alaska Native (AI/AN) communities not only face significant health disparities but are often underrepresented in health research dissemination. Existing communication tools may fail to effectively reach these communities in culturally relevant and accessible ways, limiting their ability to benefit from critical health research. We co-designed and evaluated a prototype for health research results dissemination for AI/AN communities. We created and evaluated the prototype drawing from previous co-design workshops with AI/AN people. 38 participants completed an evaluation providing feedback for further iteration and highlighting key features such as search functionality, ease of use, visual and interactive elements, and content accessibility. Participants emphasized the importance of community connection, educational resources, and personalized experiences. We feature an alternative design approach we call Indigenous Community-Centered Design to create more accessible and engaging health research communication tools for AI/AN communities, fostering stronger connections and more accurate research representation.
Health literacy significantly impacts patient outcomes, yet many struggle to understand complex medication information. Medication non-adherence often results from poor comprehension of drug instructions and contributes to preventable hospitalizations and poor treatment outcomes. With the increasing use of digital health interventions, AI-powered chatbots present an opportunity to improve patient access to understandable and personalized medication information. This study evaluates the usability, accessibility, and effectiveness of EMPATHICA, an AI-powered chatbot designed to provide patient-centric medication information. The study assesses whether chatbot-generated responses improve patient comprehension and engagement. The research observed the interaction between participants and the web-based application, where participants asked the chatbot to measure the chatbot's usability and accuracy using various qualitative and quantitative measures and expert physician evaluation of the responses. AI-driven chatbots have the potential to bridge health literacy gaps by providing clear and accessible medication information. By evaluating EMPATHICA, this study contributes to the growing field of AI applications supporting patient-informed medication use.
Documenting clients, screenings and vaccinations administered is of particular importance during mass vaccination, since information regarding uptake is critical for monitoring adverse effects and vaccine efficacy. This is especially essential when a newly-developed vaccine is being dispensed or when multiple doses of vaccine are needed per person. Despite these needs, there is no uniform or integrated system for effective vaccine data collection. This work focuses on modernizing public health infrastructure through informatics. In this paper, we describe and analyze five types of electronic technologies for registration and screening in vaccination clinics. We contrast their functionalities, usability and operations performance based on time-motion studies and service data collected during actual influenza vaccination campaigns. We evaluate their dispensing performance under an optimal dispensing clinic design. Our analysis shows that each of these electronic technologies can improve overall throughput by 16% to 45%. Based on our findings, we design a prototypical registration and screening system with integrated information flow that can be used for dispensing, monitoring and assessing mass vaccination. The system connects to the local Immunization Information System and electronic medical record systems. The design is flexible and adaptable for different types of medical countermeasures, and is suitable for regional public health departments. Our approach bridges research and public health informatics in a practical way, demonstrating how data can guide both system design and public health response planning.
We conducted formal analyses of a scoring system for a contraception decision aid to support transgender and gender-nonconforming (TGNC) assigned female at birth (AFAB) individuals. For this purpose, we developed a methodology framework to assess the weights for each decision factor, to conduct univariate and multivariate sensitivity analyses, and to provide data visualization, which led to successful identification of critical values in weight assignment that can impact the recommendations generated by the system. These analyses made critical contributions to development and validation of the system's knowledge base, providing explainable recommendations, and conducting additional research related to system users and functions. Future research is required to explore high-dimensional sensitivity analyses, to address technical issues identified, and to examine the generalizability of the methodology to other applications.
In the last decade, varied state-level policies on contraception access have highlighted the importance of large-scale public health datasets in assessing the impact of these policies on reproductive healthcare access. This study uses PRAMS Phase 8 (2016-2022) data to examine predictive factors of postpartum birth control use, hypothesizing that state policies impact contraception uptake and barriers, particularly regarding the expansion of immediate postpartum long-acting reversible contraception (LARC) reimbursement policies. Two distinct logistic regression models were constructed, and the inclusion of state as a covariate significantly reduced residual deviance (ΔDeviance = 13.696, p=0.0002). This finding indicates that state of residence is a statistically significant predictor of postpartum birth control usage. This study underscores the significant impact of state-level and institutional policies on birth control usage and LARC uptake, emphasizing the need for informed policy changes and patient-centered strategies to address disparities and improve postpartum reproductive health outcomes.
Recent advances in vision-language models (VLMs) have enabled automatic radiology report generation, yet current evaluation methods remain limited to general-purpose NLP metrics or coarse classification-based clinical scores. In this study, we propose a clinically informed evaluation framework for VLM-generated radiology reports that goes beyond traditional performance measures. We define a taxonomy of 12 radiology-specific error types, each annotated with clinical risk levels (low, medium, high) in collaboration with physicians. Using this framework, we conduct a comprehensive error analysis of three representative VLMs, i.e., DeepSeek VL2, CXR-LLaVA, and CheXagent, on 685 gold-standard, expert-annotated MIMIC-CXR cases. We further introduce a risk-aware evaluation metric, the Clinical Risk-weighted Error Score for Text-generation (CREST), to quantify safety impact. Our findings reveal critical model vulnerabilities, common error patterns, and condition-specific risk profiles, offering actionable insights for model development and deployment. This work establishes a safety-centric foundation for evaluating and improving medical report generation models. The source code of our evaluation framework, including CREST computation and error taxonomy analysis, is available at https://github.com/guanharry/VLM-CREST.
Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive cancer, with most cases diagnosed at stage IV and a five-year overall survival rate below 5%. Early detection and prognosis modeling are crucial for improving patient outcomes and guiding early intervention strategies. In this study, we implemented and evaluated deep learning fusion models that integrate radiology reports and CT imaging to predict PDAC risk. The DeepSurv model achieved a concordance index (C-index) of 0.6773 (95% CI: 0.6484, 0.7061) and 0.6596 (95% CI: 0.6260, 0.6937) on the internal and external dataset, respectively, for 5-year survival risk estimation. Kaplan-Meier analysis demonstrated significant separation (p<0.0001) between the low and high risk groups predicted by the fusion model. These findings highlight the potential of deep learning-based survival models in leveraging clinical and imaging data for pancreatic cancer.
Despite significant advancements in Generative Artificial Intelligence (GenAI), practical adoption in healthcare, particularly patient safety, remains challenging due to concerns regarding data privacy, model transparency, clinical relevance and user engagement. We present LLaMPS (Large Language Model for Patient Safety), a locally deployed GenAI platform designed to enhance patient safety event management and reporting. LLaMPS integrates automated incident classification, harm-level prediction, intelligent search, and an interactive chatbot. The system employs a Retrieval-Augmented Generation (RAG) approach, leveraging secure, institutionally hosted large language models (LLMs) and a vector database to ensure data privacy and regulatory compliance. Developed iteratively with direct input from clinicians and patient safety experts, LLaMPS demonstrates high classification accuracy and improved user satisfaction, underscoring the potential of locally controlled AI solutions to enhance patient safety workflows.