Type 2 Diabetes Prediction Model in China: A Five-Year Systematic Review

Area of Science:

• Public Health
• Epidemiology
• Biostatistics

Background:

• China faces a significant burden of type 2 diabetes (T2D) and its complications.
• Accurate T2D risk prediction models are crucial for early intervention in high-risk populations.
• Current Chinese T2D risk models exhibit methodological and clinical application limitations.

Purpose of the Study:

• To systematically review and synthesize existing T2D risk prediction models developed in China.
• To identify shortcomings in current models regarding methodology and clinical utility.
• To provide recommendations for developing higher-quality T2D risk prediction models in China.

Main Methods:

• Systematic literature review following PRISMA guidelines (October 2019 - October 2024).
• Searched PubMed, CNKI, and Web of Science for T2D risk prediction models in non-diabetic Chinese populations.
• Analyzed 20 selected studies based on model type, variable selection, validation, and performance.

Main Results:

• Twenty studies utilized diverse statistical and machine learning methods, with AUCs ranging from 0.728 to 0.977.
• Limited external validation (1 study) and variable binning (45%) may hinder model generalization.
• Key variables like lifestyle, socioeconomic factors, and cultural background were largely omitted, affecting representativeness.

Conclusions:

• Chinese T2D risk prediction models are in early development, suffering from insufficient validation and incomplete factor coverage.
• Future research must prioritize multicenter external validation and standardize modeling processes.
• Incorporating multidimensional social and behavioral variables is essential to enhance clinical utility and applicability.