Related Experiment Video
Updated: May 9, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Construction of a Machine Learning-Based Risk Prediction Model for Cognitive Frailty in Older Adults With Type 2
Li Li1, Yanping Ying2, Zhiling Ren1
1Department of Thoracic and Cardiovascular Surgery, The Second Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi Zhuang Autonomous Region, China.
Aims:
This study aimed to identify factors associated with cognitive frailty (CF) in older adults with type 2 diabetes mellitus (T2DM) and to develop a machine learning-based risk prediction model.
Methods:
Between December 2023 and December 2024, 349 participants were recruited through convenience sampling from the Department of Endocrinology, First Affiliated Hospital of Guangxi Medical University. The participants were randomly divided into a training set (n = 244) and a test set (n = 105) at a ratio of 7:3. Participants completed a structured questionnaire and were classified into CF and non-CF groups. Univariate and binary logistic regression analyses identified significant predictors, which were used as input features for six Machine Learning (ML) algorithms. Shapley additive explanations (SHAP) ranked feature importance and provided interpretability.
Results:
Of 349 older adult patients with T2DM, 87 (23.5%) had CF. Six significant predictors were identified: advanced age, lower educational attainment, insufficient physical activity, depression, malnutrition and a higher number of chronic diabetes-related complications. All models achieved satisfactory performance (AUC > 0.750). The Support Vector Machine (SVM) model performed best (AUC 0.836, accuracy 0.759, precision 0.495, recall 0.699, F1-score 0.575). A web-based application (https://webpredict1.streamlit.app/) was developed from the SVM model to enable individualised CF risk estimation.
Conclusion:
ML models effectively identified CF in older adult patients with T2DM, with the SVM model achieving the highest accuracy. Addressing the identified risk factors may help reduce CF risk and improve outcomes in this population.
Impact:
This study provides nurses with a risk prediction tool for identifying older adults with T2DM who are at high risk of CF and may facilitate the development of effective interventions for CF risk management.
Implications For The Profession And/Or Patient Care:
High-risk older adults with T2DM can be identified early through this model, enabling nurses to implement tailored interventions that may reduce CF and improve outcomes.
Reporting Method:
The study has adhered to STROBE guidelines.
Patient Or Public Contribution:
No patient or public contribution.
