MedDecXtract-XAI: Explainable Medical Decision Extraction with Grounded LLM Rationales

Presentation Type: Systems Demonstration
Presentation Time: 08:00 AM - 08:18 AM

Abstract Keywords: Natural Language Processing, Information Extraction, Large Language Models (LLMs), Human-computer Interaction, Clinical Decision Support
Programmatic Theme: Clinical Research Informatics

Understanding clinical reasoning in unstructured notes is challenging.
To address this, we present MedDecXtract-XAI, an interactive system that extracts medical decisions and their supporting rationales from clinical narratives.
Our system combines a fine-tuned RoBERTa token classifier for identifying decision spans with a localized large language model (Qwen3.5 2B) that generates explanatory rationales grounded in specific support sentences.
Deployed on Hugging Face Spaces, MedDecXtract-XAI provides explainable AI to support clinical decision-making analysis and research.

Speaker(s):
Mohamed Elgaar, PhD Student
University of Massachusetts Lowell

Hadi Amiri, PhD
UMass Lowell

Author(s):
Mohamed Elgaar, PhD Student - University of Massachusetts Lowell; Hadi Amiri, PhD - UMass Lowell; Leo Anthony Celi, MD;

EngageRx: A Remote Monitoring-Enabled, EHR-Integrated Clinical Decision Support System using Patient-Generated Data for Team-Based Hypertension Care

Presentation Type: Systems Demonstration
Presentation Time: 08:18 AM - 08:36 AM

Abstract Keywords: Clinical Decision Support, Patient-/Person-Generated Health Data, Workflow, Chronic Care Management, Mobile Health
Programmatic Theme: Clinical Informatics

EngageRx is an EHR-integrated clinical decision support system designed to synchronize remote blood pressure (BP) monitoring with actionable treatment recommendations. Built as a SMART-on-FHIR application, the platform integrates patient-generated blood pressure data with clinical information to generate guideline-based recommendations for hypertension management within routine workflows. By transforming remote monitoring data into executable decision support, EngageRx addresses therapeutic inertia in hypertension care.

Speaker(s):
Valy Fontil, Medical Doctor
NYU Langone Health

Author(s):
Valy Fontil, MD MAS - NYU Grossman School of Medicine,; Jory Purvis, BSc - University of California San Francisco; Nicole Redfern, MPH - NYU Grossman School of Medicine; Deborah Onakomaiya, PhD MPH - NYU Grossman School of Medicine; Devin Mann, MD MS - NYU Grossman School of Medicine,; Madelaine Modrow, MPH - University of California San Francisco; Mark Pletcher, MD MPH - University of California San Francisco (;

PWIN: Prospective Evaluation of Realtime Deployment of a Machine Learning Model to Predict Critical Deterioration in the Pediatric ICU

Presentation Type: Systems Demonstration
Presentation Time: 08:36 AM - 08:54 AM

Abstract Keywords: Critical Care, Clinical Decision Support, Machine Learning
Programmatic Theme: Clinical Informatics

This study prospectively evaluated the real-time deployment of the PICU Warning INdex (PWIN), a machine learning system predicting critical deterioration events (CDEs), in a 100-bed pediatric intensive care unit. Over six weeks, PWIN demonstrated higher sensitivity and lower alert burden than an existing automatic PICU Warning Tool system, detecting 69% of CDEs 1-24 hours in advance. Our findings support the feasibility and clinical value of integrating machine learning-based deterioration prediction into critical care workflows.

Speaker(s):
Eamonn Tweedy, PhD
Children's Hospital of Philadelphia

Author(s):
Eamonn Tweedy, PhD - Children's Hospital of Philadelphia; Sanjiv Mehta, MD, MSCE - Children's Hospital of Philadelphia and University of Pennsylvania School of Medicine; Sachin Grover, Data Scientist - The Children's Hospital of Philadelphia; Hannah Stinson, MD, MHQS - Children's Hospital of Philadelphia and Perelman School of Medicine at the University of Pennsylvania; Victor Ruiz, PhD - Children's Hospital of Philadelphia; James Sannino, MS; Akira Nishisaki, MD, MSCE - Children's Hospital of Philadelphia and Perelman School of Medicine at the University of Pennsylvania; Fuchiang (Rich) Tsui, PhD, FAMIA, IEEE Senior Member - Children's Hospital of Philadelphia and University of Pennsylvania;

Alemana IA and Alemana Agéntica: An Agentic Clinical AI Platform Integrated into a Custom Electronic Health Record

Presentation Type: Systems Demonstration
Presentation Time: 08:54 AM - 09:15 AM

Abstract Keywords: Large Language Models (LLMs), Informatics Implementation, Workflow
Programmatic Theme: Clinical Informatics

Alemana IA is an agentic clinical AI platform embedded directly within a production electronic health record, enabling clinicians to retrieve, synthesize, and interact with patient data in context. This live demo showcases real-world EHR integration, multi-tool agent orchestration, governance, and iterative evaluation through Alemana Agéntica, a purpose-built platform for deploying safe, workflow-integrated clinical AI copilots at institutional scale.

Speaker(s):
Fernando Eimbcke, MD
Clinica Alemana Santiago

Alejandro Mauro, MD
ClÌnica Alemana de Santiago

Author(s):
Fernando Eimbcke, MD - Clinica Alemana Santiago; Alejandro Mauro, MD - ClÌnica Alemana de Santiago; Jaime de los Hoyos, Dr. - Clínica Alemana de Santiago; Cristian Carmona, Eng - NTT data; Giorgio Cabrera, Bs - HICAPPS; Sebastian Gutierrez, Bs - HICAPPS; Emilse Bover, Eng - HICAPPS; Marcelo Lopetegui, MD, MS - HICAPPS;