Table 1 Classification report for base GPT-4o-mini model.

From: Comparing traditional natural language processing and large language models for mental health status classification: a multi-model evaluation

Class

Precision

Recall

F1-score

Depression

0.72

0.49

0.58

Normal

0.88

0.73

0.80

Suicidal

0.60

0.80

0.69

Bipolar

0.83

0.57

0.68

Stress

0.25

0.48

0.33

Anxiety

0.49

0.77

0.60

Personality disorder

0.28

0.34

0.31

  1. Overall accuracy: 65%.