Table 1 Summary of ML-based analyses, predicting clinical ratings of blunted vocal affect and alogia.
From: Using machine learning of computerized vocal expression to measure blunted vocal affect and alogia
Training set | Test set | |||||||
|---|---|---|---|---|---|---|---|---|
Criterion | Speaking task | K Neg/Pos cases | Hit rate | Correct rejection | Accuracy | Hit rate | Correct rejection | Accuracy |
Blunted vocal affect | All | 1204/404 | 0.74 | 0.95 | 0.90 | 0.65 | 0.92 | 0.85 |
Blunted vocal affect | Picture task | 915/317 | 0.84 | 0.97 | 0.94 | 0.70 | 0.93 | 0.87 |
Blunted vocal affect | Free speech | 289/87 | 0.88 | 0.99 | 0.96 | 0.60 | 0.93 | 0.85 |
Alogia | All | 1452/253 | 0.75 | 0.98 | 0.95 | 0.66 | 0.96 | 0.92 |
Alogia | Picture task | 1220/140 | 0.89 | 1.00 | 0.99 | 0.76 | 0.99 | 0.96 |
Alogia | Free speech | 232/113 | 0.96 | 0.98 | 0.97 | 0.82 | 0.92 | 0.89 |
