AI-Powered Palliative Care Management System: Integrating Large Language Models for Clinical Translation and Automated Condition Assessment

Abstract

Abstract—Delivering effective palliative care in distributed home settings presents significant challenges, particularly regarding accurate clinical documentation and patient monitoring by field workers (e.g., Asha workers) who may lack formal medical training or proficiency in professional medical English. We present the design and implementation of an AI-powered Palliative Care Assistance System built on the Django web framework. The platform seamlessly integrates Large Language Models (a local MedGemma 4B instance and Google Gemini Vision) to automate clinical translation, converting layman terminology and regional languages (Malayalam) into standardized medical English. Furthermore, the system employs vision-capable LLMs to generate structured summaries of medical images (X-Rays, Lab Reports, Prescriptions) and evaluates patient visit histories to dynamically assess condition severity (Stable, Moderate, Severe). Supported by a robust backend managing bilingual visit logs, material allocations, and role-based portals, the system significantly reduces documentation overhead and enhances clinical decision-making for doctors and headnurses. Index Terms—palliative care, large language models, generative AI, medical translation, clinical assessment, Django, health informatics, bilingual systems

Ähnliche Arbeiten