High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

Care-Chat

Author(s):

S Rahul Rao , PES Modern College of Engineering, Pune; Dnyati Khaladkar, PES Modern College of Engineering, Pune; Anjali Jadhav, PES Modern College of Engineering, Pune; Tareeka Rane, PES Modern College of Engineering, Pune; Dr. Rama Bansode, PES Modern College of Engineering, Pune

Keywords:

Medical Document Understanding, Healthcare AI, Large Language Models, OCR, Privacy-Preserving AI, Local Inference, Health Literacy, Conversational AI

Abstract

Medical reports and prescriptions are essential for healthcare continuity, but they are commonly written for clinical audiences and may be difficult for patients to interpret independently. Care-Chat is a lightweight, privacy-preserving application designed to explain uploaded laboratory reports and prescriptions in plain language. The proposed system accepts scanned or photographed documents, extracts their content using Optical Character Recognition (OCR), and uses a locally hosted large language model (LLM) to generate explanations through a conversational interface. Unlike cloud-based medical AI systems, Care-Chat is designed so that the uploaded document, extracted text, conversation history, and memory remain on the user's device. The system applies explicit safety constraints that limit the model to explanation and prohibit diagnosis or treatment recommendations. A two-tier memory mechanism based on TF-IDF retrieval and a persistent patient-memory summary supports contextual conversations across recent interactions and multiple uploaded documents. The proposed architecture combines local processing, OCR, lightweight language-model inference, safety-constrained prompting, and conversational memory into a consumer-hardware-oriented solution.

Other Details

Paper ID: IJSRDV14I90036
Published in: Volume : 14, Issue : 9
Publication Date: 01/12/2026
Page(s): 59-64

Article Preview

Download Article