HEALTHCARE researchers have developed an online interpretable machine learning system capable of predicting cardiometabolic multimorbidity (CMM) risk in patients with type 2 diabetes mellitus (T2DM), potentially transforming early intervention strategies.
Machine Learning Identifies High-Risk Patients with Type 2 Diabetes
CMM, defined as the co-occurrence of cardiovascular disease, diabetes complications, and metabolic disorders, significantly increases mortality and healthcare burden among T2DM populations. Early detection of individuals at elevated risk is critical for targeted prevention and management. A team led by Xiaohan Liu and colleagues designed an AI-driven prediction model to address this challenge.
Using data from 793 patients at a tertiary hospital, the researchers split participants into training (80%) and internal validation (20%) sets, with an additional 360 patients from an independent centre serving for external validation. Nine key predictors were identified through recursive feature elimination using a random forest algorithm. Six machine learning algorithms were trained, with a Stacking model emerging as the top performer. In internal validation, it achieved an area under the curve of 0.868, maintaining robust performance in external validation with an area under the curve of 0.822.
The model’s interpretability was ensured through SHapley Additive exPlanations and Local Interpretable Model-Agnostic Explanations, allowing clinicians to understand the contribution of individual risk factors.
Online Tool Supports Clinical Decision-Making
The system has been made available online, potentially enabling healthcare providers to rapidly assess CMM risk in patients with T2DM and implement timely interventions to slow disease progression. This online tool bridges the gap between complex AI models and practical clinical use, supporting personalised care decisions.
While machine learning may enhance risk stratification, the study’s limitations include its reliance on data from specific hospital populations, potentially limiting generalisability. Further multi-centre trials are needed to validate model performance across diverse ethnic and demographic groups.
With diabetes prevalence rising globally, integrating AI-based risk prediction tools could improve outcomes for millions at risk of cardiometabolic complications. The study demonstrates how artificial intelligence in healthcare can be leveraged for precision medicine, supporting clinicians in delivering data-driven interventions.
Reference
Liu X et al. An online interpretable machine learning model for predicting cardiometabolic multimorbidity risk in patients with type 2 diabetes mellitus. Sci Rep. 2026; DOI:10.1038/s41598-026-36923-2.
