Dynamic evolution of readmission risk factors across short-, medium-, and long-term horizons in type 2 diabetes: a machine learning-based predictive modeling study with SHAP interpretability - Report - MDSpire
Advertisement
Dynamic evolution of readmission risk factors across short-, medium-, and long-term horizons in type 2 diabetes: a machine learning-based predictive modeling study with SHAP interpretability
Clinical Report: Evolving Readmission Risk Factors in Type 2 Diabetes
Overview
This study identifies dynamic readmission risk factors in Type 2 Diabetes Mellitus (T2DM) patients using machine learning models. Key predictors vary across different time windows.
Background
Type 2 diabetes mellitus (T2DM) is a prevalent chronic disease with significant healthcare implications, including high readmission rates. Understanding the evolving risk factors for readmission is crucial. Current clinical guidelines lack specific predictive tools for assessing readmission risk over varying time frames.
Data Highlights
Model
AUROC
Time Window
ANN
Highest
30-day, 60-day
Random Forest
Competitive
365-day
Key Findings
Age is the dominant predictor for 30-day readmission.
Length of hospital stay and inflammatory markers (SII, SIRI) are key for 60-day readmission.
Chronic complications of diabetes are the primary risk factors for 365-day readmission.
Machine learning models outperformed traditional logistic regression in predicting readmission.
SHAP analysis provided insights into the evolution of risk factors over time.
Clinical Implications
Healthcare providers should consider using machine learning models tailored to specific time windows for predicting readmission risk in T2DM patients. Continuous monitoring of evolving risk factors can enhance patient management and reduce readmission rates.
Conclusion
The study highlights the importance of time-specific risk assessment in T2DM readmission.
