Community screening for dementia among older adults in China: a machine learning-based strategy

Yan Zhang,Jian Xu,Chi Zhang,Xu Zhang,Xueli Yuan,Wenqing Ni,Hongmin Zhang,Yijin Zheng,Zhiguang Zhao

DOI: https://doi.org/10.1186/s12889-024-18692-7

IF: 4.5

2024-05-03

BMC Public Health

Abstract:Dementia is a leading cause of disability in people older than 65 years worldwide. However, diagnosing dementia in its earliest symptomatic stages remains challenging. This study combined specific questions from the AD8 scale with comprehensive health-related characteristics, and used machine learning (ML) to construct diagnostic models of cognitive impairment (CI).

public, environmental & occupational health

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve

This paper aims to address the difficulty in diagnosing early symptoms of dementia in the elderly population. Specifically, the study combines specific questions from the AD8 scale with comprehensive health-related features and utilizes machine learning (ML) to construct a cognitive impairment (CI) diagnostic model. The goal of the research is to provide a new approach for dementia screening in community settings while minimizing the use of medical resources.

### Main Research Content

1. **Background**:
- Dementia is one of the leading causes of disability among people aged 65 and older worldwide. However, diagnosing dementia at the earliest symptom stage remains challenging.
- Current screening tools like the MMSE, although widely used, are not efficient enough for large-scale community screening.

2. **Methods**:
- Based on the Shenzhen Healthy Aging Research (SHARE) project, 823 participants aged 65 and above were recruited for comprehensive health assessments and cognitive function evaluations.
- Five machine learning models (Support Vector Machine SVM, Multilayer Perceptron MLP, AdaBoost, Gradient Boosting Decision Tree GBDT, and Logistic Regression LR) were used, and the BalanceCascade method was applied to handle the imbalanced dataset to predict cognitive impairment.
- Feature selection was performed using permutation importance, and model results were interpreted using SHAP values.

3. **Results**:
- Features such as gender, age, waist circumference, and body mass index were selected as predictors.
- Compared to the traditional AD8 scale (AUC=0.65), the MLP model performed the best (AUC=0.83±0.04), followed by AdaBoost (AUC=0.80±0.04), SVM (AUC=0.78±0.04), and GBDT (AUC=0.76±0.04).
- Factors such as being female, older age, and lower waist circumference had the greatest impact on the model's prediction of CI.

4. **Conclusion**:
- The cognitive impairment diagnostic model constructed using machine learning, particularly the MLP model, significantly outperformed the traditional AD8 scale (score ≥2 points). These findings provide new insights for community dementia screening and help in intervening with high-risk individuals.