Table 4 Integrated experiment results (Part 2)
From: Integrative diagnosis of psychiatric conditions using ChatGPT and fMRI data
Experiment | F1-score | Mean accuracy | Variance accuracy | P-value |
|---|---|---|---|---|
Baseline Model (without NLP) | 0.77 | 0.79 | 0.00015 | \(1.2 \times 10^{-6}\) |
Full Model (with NLP) | 0.84 | 0.87 | 0.00013 | \(4.5 \times 10^{-7}\) |
Ablation Study 1 (Removing NLP Features) | 0.79 | 0.81 | 0.00014 | \(9.3 \times 10^{-7}\) |
Ablation Study 2 (NLP Features Only) | 0.81 | 0.83 | 0.00012 | \(7.8 \times 10^{-7}\) |