Stress, depression, anxiety, and other mental health issues have a big influence on college students’ academic performance and general well-being. It is difficult to identify these problems because students frequently are not aware of their mental health. This study looks at how well Long Short-Term Memory (LSTM), Random Forest, and K-Nearest Neighbors classifiers work together to predict mental health problems in students. The “Entrepreneurial Competency in University Students” dataset was obtained from Kaggle and underwent extensive preprocessing, which included data transformation, cleaning, and Chi-Square feature selection. By optimizing its hyper parameters, LSTM model was able to attain 96% accuracy. The accuracy of the proposed model is superior to that of the cognitive model. Given the Chi-Square value of 914.19 and the p-value of 9.28981693913268e−20, there is a strong statistical correlation between features and mental health conditions. These findings underline the significance of early mental health issue detection and treatment and give educational institutions a strong foundation for enhancing students’ well-being.