Table 2.

Performances of the five models optimized for detecting radiographic hip osteoarthritis on the X-ray images within the validation and test datasets. The CT-AP images were created by sequentially summing the CT slices. Model-1 was trained with unprocessed X-ray images. Model-2 was trained similarly with downsampled X-rays to resemble CT-AP images. Model-3 and Model-4 were trained on a combination of the CT-AP and X-ray images where the X-ray images were similar to the ones used in Model-1 and Model-2, respectively. Model-5 was trained solely on the CT-AP images

Trained models	Accuracy	Balanced accuracy	Precision	Recall	F1-score	PR AUC [95% CI]	ROC AUC [95% CI]
X-ray images of the validation dataset
Model-1	93.3	92.1	0.92	0.93	0.93	0.96 [0.95–0.97]	0.98 [0.98–0.99]
Model-2	90.6	89.2	0.89	0.89	0.89	0.94 [0.92–0.95]	0.97 [0.96–0.97]
Model-3	92.7	92.3	0.92	0.92	0.92	0.95 [0.93–0.95]	0.98 [0.97–0.98]
Model-4	90.4	89.2	0.89	0.89	0.89	0.94 [0.93–0.95]	0.94 [0.97–0.98]
X-ray images of the test dataset
Model-1	92.3	88.5	0.91	0.92	0.92	0.96 [0.95–0.98]	0.98 [0.97–0.98]
Model-2	90.2	88.1	0.88	0.90	0.89	0.95 [0.93–0.96]	0.97 [0.96–0.98]
Model-3	91.9	91.1	0.91	0.91	0.91	0.95 [0.93–0.96]	0.98 [0.97–0.98]
Model-4	91.3	90.8	0.91	0.90	0.91	0.93 [0.91–0.95]	0.97 [0.96–0.98]
Model-5†	49.1	58.9	0.59	0.64	0.61	0.53 [0.50–0.56]	0.69 [0.66–0.72]
Model-5‡	60.5	55.2	0.55	0.56	0.55	0.57 [0.52–0.62]	0.58 [0.54–0.61]

^†Performance of Model-5 on the X-ray images used in Model-1

^‡Performance of Model-5 on the downsampled X-ray images used in Model-2

PR AUC, area under the precision recall curve; ROC AUC, area under the receiver operating characteristics curve; CI, confidence interval