Physio: An LLM-Based Physiotherapy Advisor
Rúben Almeida, Hugo Sousa, Luís F. Cunha, Nuno Guimarães, Ricardo Campos, Alípio Jorge
TL;DR
The paper tackles trustworthy deployment of large language models in healthcare by grounding outputs in a domain-specific knowledge base for physical rehabilitation. It introduces Physio, a chat-based advisor that uses retrieval-augmented generation with BM25 document retrieval and an OpenAI GPT-4 backend to produce diagnoses, exercise plans, and OTC medication recommendations, all with verifiable references. The knowledge base is built from Rehab Hero, validated medical sources, and DrugBank, and a structured data pipeline handles input validation, condition identification, and retrieval-guided answer generation. Ethical safeguards include a clear research disclaimer and limiting medication suggestions to OTC options, with open-source code to foster transparency and further development toward a cautious, AI-assisted physiatrist.
Abstract
The capabilities of the most recent language models have increased the interest in integrating them into real-world applications. However, the fact that these models generate plausible, yet incorrect text poses a constraint when considering their use in several domains. Healthcare is a prime example of a domain where text-generative trustworthiness is a hard requirement to safeguard patient well-being. In this paper, we present Physio, a chat-based application for physical rehabilitation. Physio is capable of making an initial diagnosis while citing reliable health sources to support the information provided. Furthermore, drawing upon external knowledge databases, Physio can recommend rehabilitation exercises and over-the-counter medication for symptom relief. By combining these features, Physio can leverage the power of generative models for language processing while also conditioning its response on dependable and verifiable sources. A live demo of Physio is available at https://physio.inesctec.pt.