Frontiers in oncology · 2026
Comparative study of four sarcopenia-related risk prediction models for identifying SARC-F screening positivity in patients with colorectal cancer.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Li Zhuoyun, Wang Xiaoying, Liu Fang, Hu Yali
Аннотация
OBJECTIVE: To compare the performance of four sarcopenia-related risk prediction models in identifying SARC-F screening positivity among patients with colorectal cancer, and to provide evidence for early inpatient screening.
METHODS: A total of 340 hospitalized patients with colorectal cancer were included. SARC-F screening positivity was defined as a Chinese version SARC-F score of ≥4. The Ren LSMI score, Lim LP-SMG score, Zhang Xuejing score, and Zhang Ying linear predictor were reconstructed according to their original formulas. Model performance was evaluated using receiver operating characteristic (ROC) curves, the DeLong test, 10-fold cross-validated recalibration, Brier score, calibration indices, and decision curve analysis.
RESULTS: Among the 340 patients, 52 were SARC-F screening positive, with a positivity rate of 15.29%. The AUC of Model 1 was significantly lower than those of the other three models, while no significant differences were observed among Models 2, 3, and 4. After recalibration, Model 3 had the lowest Brier score of 0.088, with a calibration intercept of -0.014 and a calibration slope of 0.909, and showed clinical net benefit across the threshold probability range of 1%-50%. Model 4 had the highest sensitivity, whereas Model 1 had relatively high specificity. Although Model 2 showed a high AUC, its recalibration slope was negative, suggesting that it was not suitable for direct prediction of this outcome.
CONCLUSION: All four models showed some ability to identify SARC-F screening positivity. Among them, the Zhang Xuejing score demonstrated relatively stable overall performance and may serve as a preferred candidate tool for early inpatient screening. The Zhang Ying model may be useful for reducing missed cases, while the Ren LSMI score may help reduce false positives. Further validation in independent multicenter cohorts is warranted.
METHODS: A total of 340 hospitalized patients with colorectal cancer were included. SARC-F screening positivity was defined as a Chinese version SARC-F score of ≥4. The Ren LSMI score, Lim LP-SMG score, Zhang Xuejing score, and Zhang Ying linear predictor were reconstructed according to their original formulas. Model performance was evaluated using receiver operating characteristic (ROC) curves, the DeLong test, 10-fold cross-validated recalibration, Brier score, calibration indices, and decision curve analysis.
RESULTS: Among the 340 patients, 52 were SARC-F screening positive, with a positivity rate of 15.29%. The AUC of Model 1 was significantly lower than those of the other three models, while no significant differences were observed among Models 2, 3, and 4. After recalibration, Model 3 had the lowest Brier score of 0.088, with a calibration intercept of -0.014 and a calibration slope of 0.909, and showed clinical net benefit across the threshold probability range of 1%-50%. Model 4 had the highest sensitivity, whereas Model 1 had relatively high specificity. Although Model 2 showed a high AUC, its recalibration slope was negative, suggesting that it was not suitable for direct prediction of this outcome.
CONCLUSION: All four models showed some ability to identify SARC-F screening positivity. Among them, the Zhang Xuejing score demonstrated relatively stable overall performance and may serve as a preferred candidate tool for early inpatient screening. The Zhang Ying model may be useful for reducing missed cases, while the Ren LSMI score may help reduce false positives. Further validation in independent multicenter cohorts is warranted.