IMPACT: an interactive multi-disease prevention and counterfactual treatment system using explainable AI and a...

Area of Science:

• Computational health science
• Artificial intelligence in public health
• Preventive medicine

Background:

• Concurrent multi-disease conditions increase mortality risk and complicate recovery.
• Effective prevention strategies for multiple diseases are crucial for public well-being.
• Explainable AI (XAI) and large language models (LLMs) offer potential for accessible health guidance.

Purpose of the Study:

• To propose an interactive system for concurrent multi-disease prevention.
• To leverage XAI and multimodal LLMs for personalized risk reduction recommendations.
• To address the challenge of simultaneously managing risks for multiple diseases.

Main Methods:

• Utilized Google Gemini Pro, a multimodal LLM, for natural language interaction.
• Employed the Non-dominated Sorting Genetic Algorithm II (NSGA-II) for multi-objective optimization.
• Developed a system recommending feature value adjustments to minimize concurrent disease risks.

Main Results:

• The system provides personalized recommendations to concurrently minimize risks of diseases like heart attacks and diabetes.
• Achieved significant reduction in disease attack probabilities through personalized feature selection.
• Demonstrated the feasibility of interactive, multi-disease risk management for the general public.

Conclusions:

• The proposed system offers a novel approach to interactive multi-disease prevention.
• XAI and LLMs can empower the public in managing multiple health risks.
• This approach has the potential to significantly improve public health outcomes by addressing complex, concurrent disease prevention.