AI agents enable natural language on complex healthcare data models
Can AI Agents Interface Clinicians and Complex Models?
Multiagent SystemsArtificial Intelligence
Summary
Hospitals create huge amounts of data every day, but doctors and administrators often find it hard to use complex models to understand this data. The authors built EpidemIA, an AI system that lets users ask questions in plain language to get predictions and analyses from sophisticated health models. This system runs locally on a regular computer and protects sensitive data by not sending it outside. Their tests show it can answer simple and complex questions about emergency visits, helping bring advanced modeling to healthcare professionals more easily.
What this means in practice
- •For hospital data teams: Allow hospital analysts and administrators to query complex epidemiological models using everyday language without needing coding skills.
- •For medical software developers: Integrate local AI agents with existing health data platforms to provide clinicians quick access to predictive and statistical insights via natural language queries.
Tested on one dataset.
Authors
Giulio Colombini, Matteo Falcioni, Mario Massimo, Francesco Durazzi, Marianna Santoro, Paolo Tubertini, Mirko Degli Esposti, Armando Bazzani
Abstract
Background: Contemporary healthcare institutions generate daily very large amounts of data, which contain highly useful information for the improvement of work organization, planning and overall efficiency. While models and techniques to extract these insights are well known and constantly developed in mathematical biology, complex systems physics and data science, the adoption of these techniques among Clinicians and Hospital Administrators remains limited, mainly due to language and methodological divides. Objective: In this article we implement an AI Multi Agent System (MAS) that allows Users to use predictive models and make statistical analyses via natural language conversations. Methods: We developed EpidemIA, a training-free MAS endowed with a modular suite of epidemiological models written by human experts, executable on a single-GPU workstation. To avoid relaying sensitive data to third parties, the system uses locally an open-weights LLM as backbone for its two subagents. Results: Using a dataset of Emergency Department visits provided to us by IRCCS Sant'Orsola (Bologna) we test the model on a few typical use cases, demonstrating the possibility of performing both simpler statistical analyses and visualizations as well as more complicated epidemiological predictions based on dynamical models, through simple queries made in natural language. Conclusions: While still being a proof-of-concept implementation, EpidemIA showcases the transformative capabilities of LLM based technologies as an enabling factor in the widespread adoption of complex dynamical models, which is at present hindered by their steep learning curve.