Construction of a multi-source data integration learning decision framework for college students’ mental health

There is the need to have a multi-source data integration learning framework of mental health among college students so as to identify and address the psychological disorders within the academic setting. Mental health problems can be diagnosed, managed, and supported at an early stage, which enhances the well-being of students and their academic success. Mental health of the students is regularly assessed using surveys and counseling records, but they may be biased. These disjointed approaches do not reflect the complexity of mental health that is affected by behavioral, intellectual and social factors. The Multi-Source Data Integration Learning of Mental Health (MSDIL-MH) methodology resolves such problems. The method involves the use of strong machine learning to determine the mental health of the students based on academic performance, attendance, internet use, human contacts and the result of the psychological tests. With the help of MSDIL-MH, it is possible to monitor the well-being of the students, predict the risk in advance, and offer specific advice and interventions. The offered method also uses Deep Reinforcement Learning (DRL) to improve the process of early risk detection and specific guidance by drawing the best decision rules based on the multi-source student data. It is then possible to have flexible and individually-specific solutions. The system aids in scalable privacy-preserving university and mental health professional decision-making. The results of the experiments indicate that MSDIL-MH is more accurate and reliable in prediction than single-source methods and provides more detailed information about the mental health of students to intervene in time and in a practical way. The proposed method has the prediction accuracy of 95, F1 score of more than 0.8, 95% recall and precision of 0.9. Scalability and performance is more than 80%.