Construction of machine learning based psychological crisis warning model for college students integrating biomechanics indicators

College students these days are under a great deal of psychological stress. This stress, whether originating from academic demands or personal life challenges, triggers a cascade of biological responses within the body. Physiologically, stress can disrupt the normal functioning of the endocrine system, leading to abnormal secretion of stress hormones like cortisol, affecting neurotransmitter levels and neural plasticity, thereby impacting mental health. If left unaddressed, this psychological stress can have severe and lasting negative effects on their well-being. In this context, it is crucial to quickly identify students experiencing mental health crises. However, the manual verification method has significant constraints and cannot effectively ascertain the mental state of the students. To address these challenges, this research proposes a Machine Learning-based Psychological Crisis Warning (ML-PCW) framework, integrating biomechanical indicators to provide unique insights into students’ psychological states. For instance, changes in gait patterns can be associated with different emotional states, where abnormalities in walking speed, stride length, or body sway may indicate increased stress or anxiety. In the digital age, college students are more inclined to express their emotions through online platforms. Big Data analytics has emerged as a powerful tool for analyzing this digital footprint, providing valuable insights into their psychological states. Additionally, statistical techniques are employed to establish an emotional assessment paradigm that considers not only traditional psychological factors but also biological and biomechanical cues. In this research, the honey badger search-joint adaptive kernelized support vector machine (HBS-AKSVM) technique is developed. This technique is designed to handle the labeling process of the initial data, which includes both psychological and biological data. By incorporating biomechanical indicators, the HBS-AKSVM can more accurately categorize and analyze the data while minimizing the computational load during the development of the PCW system. Research findings show that the suggested approach operates more effectively than current approaches in terms of supplying professionals with a psychological supplementary assessment that is dependable. By integrating biomechanical indicators, the ML-PCW framework offers a more comprehensive and accurate understanding of college students’ psychological states, enabling early detection and timely intervention in mental health crises.