You Fu ^*

• Shanxi Vocational University of Engineering Science and Technology, Taiyuan, China

In response to the burgeoning trend of automation and intelligence in the domain of mental health services within higher education institutions, this study delves into the application of artificial intelligence technologies for the early warning analysis of mental health risks among students. Utilizing the Apriori algorithm, a predictive model for student mental health risks has been developed. The model initiates by uncovering the association rules between various mental health risk factors through the Apriori algorithm. Following this, a hybrid predictive model combining the Prophet time series model and Long Short-Term Memory (LSTM) networks, known as Prophet-LSTM, was constructed to leverage their combined strengths in addressing the mental health risk prediction challenge. Subsequently, the highly associated mental health risk factors, derived from the data mining process, were employed as inputs. The weight coefficients of the hybrid model were optimized using the Quantum Particle Swarm Optimization (QPSO) algorithm to forecast associated mental health risks. The efficacy of the proposed student mental health risk prediction model was validated using data from a mental health historical questionnaire survey conducted among college students from March 2020 to August 2023 at a university in China. Additionally, evaluation metrics for the detection rate of psychological issues and the detection rate of no psychological issues were introduced. Compared to other classical machine learning algorithms, the proposed model demonstrated the highest detection rates for both psychological issues and the absence thereof, exhibiting a robust predictive capability. This study underscores the practicality and effectiveness of the model, providing a scientific foundation for the enhancement of mental health services, which holds significant research value and societal importance.