Correction to: Scientific Reports 10.1038/s41598-024-51567-w, published online 05 February 2024
The original version of this Article contained errors in Table 4, results for the receiver operating characteristic curve analyses.
Table 4, “Receiver operating characteristic (ROC) curve analyses for each logistic regression model predicting PROMIS depression symptom T-score categories (mild, moderate, and severe) versus PROMIS depression symptom T-scores within normal limits (WNL) from self-reported body temperature.”
| Model | Outcome | Temperature Cut point (°C) | Sensitivity | Specificity | ROC AUC |
|---|---|---|---|---|---|
| Unadjusted | Severe vs WNL | 36.613 | 60.87% | 62.34% | 0.635 |
| Adjusted | Severe vs WNL | 36.375 | 85.87% | 34.05% | 0.762 |
| Unadjusted | Moderate vs WNL | 36.624 | 96.91% | 63.40% | 0.557 |
| Adjusted | Moderate vs WNL | 36.350 | 75.17% | 31.55% | 0.672 |
| Unadjusted | Mild vs WNL | 36.624 | 42.72% | 63.46% | 0.537 |
| Adjusted | Mild vs WNL | 36.500 | 58.54% | 47.47% | 0.612 |
now reads,
| Model | Outcome | Temperature Cut point (°C) | Sensitivity | Specificity | ROC AUC |
|---|---|---|---|---|---|
| Unadjusted | Severe vs WNL | 36.612 | 60.87% | 62.34% | 0.635 |
| Adjusted | Severe vs WNL | 36.375 | 85.87% | 34.05% | 0.762 |
| Unadjusted | Moderate vs WNL | 36.624 | 45.87% | 63.40% | 0.557 |
| Adjusted | Moderate vs WNL | 36.350 | 75.17% | 31.55% | 0.672 |
| Unadjusted | Mild vs WNL | 36.624 | 42.72% | 63.46% | 0.537 |
| Adjusted | Mild vs WNL | 36.460 | 61.76% | 44.13% | 0.612 |
Additionally, in the Results section, ‘Receiver operating characteristics (ROC) curve analyses’
“Using Youden’s Index, which locates the threshold value that maximizes the distance between the ROC curve and the line of chance, to identify optimally performing threshold values from each ROC curve resulted in 85.87% sensitivity to detect PROMIS depression T-scores within the severe range based on the adjusted model, but with specificity of 34.05%; the best performance was 96.91% sensitivity to detect PROMIS depression T-scores within the moderate range (with 63.40% specificity) from the unadjusted model. Sensitivity was lowest (42.72%) for detection of PROMIS depression T-scores within the mild range based on an unadjusted analysis. Specificity was lowest (31.55%) for detection of PROMIS depression T-scores within the moderate range based on the adjusted model (Table 4).”
now reads,
“Using Youden’s Index, which locates the threshold value that maximizes the distance between the ROC curve and the line of chance, to identify optimally performing threshold values from each ROC curve resulted in 85.87% sensitivity to detect PROMIS depression T-scores within the severe range based on the adjusted model, but with specificity of 34.05%; this was the best-performing model in terms of sensitivity. Sensitivity was lowest (42.72%) for detection of PROMIS depression T-scores within the mild range based on an unadjusted analysis. Specificity was lowest (31.55%) for detection of PROMIS depression T-scores within the moderate range based on the adjusted model (Table 4).”
The original version of this Article has been corrected.