Multi Agent based Medical Assistant for Edge Devices
Sakharam Gawade, Shivam Akhouri, Chinmay Kulkarni, Jagdish Samant, Pragya Sahu, Aastik, Jai Pahal, Saswat Meher
TL;DR
The paper tackles privacy, latency, and offline operation challenges of cloud-based large action models in healthcare by proposing an on-device, multi-agent healthcare assistant. It introduces a three-component architecture—Health Manager, Memory, and Action Manager—with Planner and Caller agents fine-tuned via LoRA on a Qwen-based backbone to run entirely on edge devices. A synthetic data generation pipeline, including data formation, enhancement, verification, and interleaving, yields high performance (RougeL scores of 85.5 for Planner and 96.5 for Caller on appointment tasks) and full success on SOS scenarios. The work demonstrates a scalable, privacy-preserving, user-centric solution for health management with practical deployment potential on wearables and mobile devices.
Abstract
Large Action Models (LAMs) have revolutionized intelligent automation, but their application in healthcare faces challenges due to privacy concerns, latency, and dependency on internet access. This report introduces an ondevice, multi-agent healthcare assistant that overcomes these limitations. The system utilizes smaller, task-specific agents to optimize resources, ensure scalability and high performance. Our proposed system acts as a one-stop solution for health care needs with features like appointment booking, health monitoring, medication reminders, and daily health reporting. Powered by the Qwen Code Instruct 2.5 7B model, the Planner and Caller Agents achieve an average RougeL score of 85.5 for planning and 96.5 for calling for our tasks while being lightweight for on-device deployment. This innovative approach combines the benefits of ondevice systems with multi-agent architectures, paving the way for user-centric healthcare solutions.