A measurement method for mental health based on dynamic multimodal feature recognition

Area of Science:

• Psychiatry
• Computer Science
• Education

Background:

• College student mental health issues have risen, exacerbated by the COVID-19 pandemic.
• Early-stage psychological problems are often subclinical, leading to missed intervention opportunities.
• Large-scale screening is crucial for identifying at-risk students.

Purpose of the Study:

• To propose an integrated mental health assessment method for college students.
• To leverage artificial intelligence (AI) and multimodal intelligent recognition for enhanced screening.
• To support large-scale and normalized mental health problem detection in universities.

Main Methods:

• Utilized human-computer interaction-based psychological assessment scales for questionnaires.
• Integrated machine learning technology for identifying student mental states and problem severity.
• Recruited 1,500 students for the mental health assessment study.

Main Results:

• The proposed multimodal intelligent recognition method demonstrated high accuracy.
• The method effectively complements traditional scale results for mental health assessment.
• Identified incidence rates: 36.3% moderate/higher stress, 48.1% anxiety, 23.0% depression.

Conclusions:

• The interactive multimodality emotion recognition method offers an effective solution for large-scale mental health screening.
• This approach facilitates monitoring and intervention for college student mental health.
• Supports normalized and efficient mental health problem identification in educational institutions.