MedIntelliCare: neurodynamic-inspired AI for medical decision support by integrating retrieval-augmented generation...

Area of Science:

• Computational Neuroscience
• Artificial Intelligence in Medicine
• Medical Diagnostics

Background:

• Current AI medical systems primarily focus on information retrieval and response generation.
• There is a need for AI systems that reduce cognitive load and integrate real-time data for healthcare professionals.
• MedIntelliCare aims to bridge this gap by incorporating advanced computational principles.

Purpose of the Study:

• To explore MedIntelliCare's capability to simulate biologically inspired information processing.
• To integrate brain-like computing, predictive modeling, and multimodal analysis (EEG, neuroimaging) into an AI medical assistant.
• To enhance clinical decision support through adaptive information retrieval and neuro-inspired frameworks.

Main Methods:

• Utilizing Retrieval-Augmented Generation (RAG) combined with neural computation and decision-making principles.
• Integrating multimodal data analysis, including electroencephalography (EEG) and neuroimaging.
• Applying cosine similarity metrics for experimental validation against expert-generated reports.

Main Results:

• MedIntelliCare demonstrated a 73% alignment with expert-generated medical reports.
• The system successfully simulated aspects of biologically inspired information processing.
• Validation confirmed the system's potential in neuro-inspired medical intelligence.

Conclusions:

• MedIntelliCare shows significant potential as an AI-powered medical assistant for enhancing diagnostic accuracy.
• The system's neuro-inspired approach offers a novel framework for clinical decision support.
• Future implications include advancements in cognitive disorder modeling and brain-computer collaboration.