Table 1. Demographic data of patients.
| Characteristics | Luhe Hospital (training cohort) |
Luhe Hospital (testing cohort) |
Xuanwu Hospital (alternative hospital validation cohort) |
|---|---|---|---|
| Patient count | 819 | 232 | 46 |
| Age (year) | 73±8 | 72±8 | 76±9 |
| Sex, n (%) | |||
| Male | 192 (23.4) | 52 (22.4) | 13 (28.3) |
| Female | 627 (76.6) | 180 (77.6) | 33 (71.7) |
| DEXA T-score | −2.9±1.3 | −2.9±1.4 | −2.9±0.5 |
| Number of slices | 8,548 | 2,456 | 467 |
Data were presented as mean ± SD or n (%). The training cohort, testing cohort, and alternative hospital validation cohort showed no significant differences in gender distribution (Pearson χ2 test, χ2=0.732, P=0.693). These datasets also showed no significant differences in DEXA T-score (ANOVA Bartlett’s test with Tukey’s multiple comparisons; P=0.368). The patients used for alternative hospital validation were older than the patients in the training and testing cohorts from Luhe Hospital (ANOVA Bartlett’s test with Tukey’s multiple comparisons; P=0.004; Train vs. Test P=0.093; Train vs. Validation, P=0.003; Test vs. Validation, P=0.014). DEXA, dual-energy X-ray absorptiometry; SD, standard deviation; ANOVA, analysis of variance.