A risk model to predict the mental health of older people in Chinese communities based on machine learning.

Background: Anxiety, depression, and dementia are important issues affecting the mental health of the older population. Given the relationship between mental health and physical disorders, it is particularly important to diagnose and identify these psychological problems in older people. Methods: Psychological data of 15,173 older people living in various districts and counties of Shanxi province, China, were extracted from data collected through the '13th Five-Year Plan for Healthy Aging-Psychological Care for the Elderly Project' of the National Health Commission of China in 2019. Three different ensemble learning classifiers [random forest (RF), Extreme Gradient Boosting (XGBoost) and Light Gradient Boosting Machine (LightGBM)] were evaluated, and the best classifier with the selected feature set was selected. The ratio of training to testing cases was 8:2. The predictive performance of the three classifiers was evaluated by calculating the area under the receiver operating characteristic curve (AUC), accuracy, recall, and F measurement based on 10-fold cross-validation and ranked by AUC. Results: All the three classifiers have achieved good prediction results. In the test set, the AUC value range for the three classifiers was 0.79 to 0.85. The LightGBM algorithm showed higher accuracy than both the baseline and XGBoost. A new machine learning (ML)-based model to predict mental health problems in older people was constructed. The model was interpretative and could hierarchically predict psychological problems including anxiety, depression, and dementia in older people. Experimental results showed that the method could accurately identify those suffering from anxiety, depression, and dementia in different age groups. Conclusions: A simple method model was designed based on only eight problems, which had good accuracy and was widely applicable to the older of all ages. Overall, this research approach avoided the need to identify older people with poor mental health through the traditional standardized questionnaire approach.

Learning self-driven collective dynamics with graph networks.

Despite decades of theoretical research, the nature of the self-driven collective motion remains indigestible and controversial, while the phase transition process of its dynamic is a major research issue. Recent methods propose to infer the phase transition process from various artificially extracted features using machine learning. In this thesis, we propose a new order parameter by using machine learning to quantify the synchronization degree of the self-driven collective system from the perspective of the number of clusters. Furthermore, we construct a powerful model based on the graph network to determine the long-term evolution of the self-driven collective system from the initial position of the particles, without any manual features. Results show that this method has strong predictive power, and is suitable for various noises. Our method can provide reference for the research of other physical systems with local interactions.