A Decision Support System for Diabetes Chronic Care Models Based on General Practitioner Engagement and EHR Data...

Area of Science:

• Health Informatics
• Artificial Intelligence in Medicine
• Chronic Disease Management

Background:

• Decision support systems (DSS) aim to enhance healthcare quality but face challenges in clinical strategy, resource management, and preventive medicine implementation.
• A lack of standardized clinical strategies and resource management hinders effective preventive medicine, impacting chronic disease care.

Purpose of the Study:

• To propose an integrated DSS leveraging shared Electronic Health Records (EHRs) for improved chronic Type 2 Diabetes (T2D) care.
• To develop a Machine Learning (ML) model for analyzing EHR data, addressing variability and enhancing T2D patient management.
• To establish novel T2D care-quality indicators for economic incentives and ML model prediction.

Main Methods:

• Utilized a cloud-based infrastructure (Netmedica Italian) for secure sharing of General Practitioners' (GPs) EHRs.
• Implemented a Machine Learning (ML) approach to analyze EHR data from 41,237 T2D patients.
• Developed and applied a set of T2D care-quality indicators to assess performance and determine economic incentives.

Main Results:

• The DSS achieved high classification performance for chronic care quality (98%±2% accuracy, 96%±1% macro-recall).
• The integrated system led to a significant increase (up to 12%) in T2D patients remaining without complications.
• Economic incentives were provided to GPs for system adoption and achieving performance targets.

Conclusions:

• Empowering General Practitioners (GPs) with an integrated DSS platform and economic incentives can accelerate improvements in chronic disease care quality.
• The proposed system demonstrates a viable model for enhancing T2D management and potentially other chronic conditions.
• The study highlights the effectiveness of ML-driven analysis of shared EHRs in improving patient outcomes and care quality.