External validation and application of a machine learning-based model for diabetes progression in prediabetes.

Area of Science:

• Endocrinology
• Computational Biology
• Preventive Medicine

Background:

• Prediabetes is a significant risk factor for type 2 diabetes (T2D) and its complications.
• Accurate risk prediction models are crucial for timely intervention in prediabetes.
• Existing models may not fully capture the complexity of T2D progression.

Purpose of the Study:

• To externally validate a machine learning-based model for type 2 diabetes progression (ML-PR).
• To evaluate the clinical utility of the ML-PR model in individuals with prediabetes.
• To compare the ML-PR model's performance with existing diabetes prediction models.

Main Methods:

• Utilized 3,081 participants from the Diabetes Prevention Program (DPP) and DPP Outcome Study (DPPOS).
• Assessed the ML-PR model using discrimination, calibration, and decision curve analyses.
• Employed Cox proportional hazards and logistic regression to analyze T2D incidence, microvascular complications, and cardiovascular events.

Main Results:

• The ML-PR model achieved an AUC of 0.74 for predicting 3-year T2D progression.
• High-risk individuals showed significantly higher T2D incidence and a 67% increased risk of microvascular complications.
• High-risk participants demonstrated greater benefits from lifestyle modification and metformin interventions (P < 0.05).

Conclusions:

• The ML-PR model is a validated tool for identifying high-risk prediabetes patients.
• The model aids in predicting T2D progression and microvascular complications.
• Findings support personalized treatment strategies, particularly for high-risk individuals receiving intensive interventions.