JCO clinical cancer informatics · 2026
Development of a Digital Phenotype for Immune-Related Colitis.
СтатьяПоддержка исследования: вне правительства СШАНазвание 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.
Teng Jasmine C, Worth Leon J, Thursky Karin A, Trinh Andrew, Spain Lavinia, Tran Phillip K, Spelman Tim, Nkonde Chowa, Terreros-Martin Sofia, Roubos George, Shinkov Dimiter, Capurro Daniel
Аннотация
PURPOSE: Immune-related colitis (IR-colitis) is a significant side effect of immune checkpoint inhibitor (ICI) therapy for an expanding group of cancers. Timely identification is required to optimize outcomes, but existing approaches to retrospective case detection are resource intensive. We sought to develop an electronic medical record (EMR)-based digital phenotype to identify patients with IR-colitis.
METHODS: Adult patients commenced on ICIs from August 1, 2020, to June 30, 2023, at the Peter MacCallum Cancer Centre were studied. ICIs were prescribed as monotherapy or in combination with other treatment modalities for a variety of solid cancers. Deidentified EMR data were extracted. A rule-based digital phenotype was developed using expert-selected features, with manual chart review serving as the gold standard comparator.
RESULTS: The ICI-treated cohort comprised 1,464 adults, with a male predominance (62%). The most frequent cancer diagnoses were melanoma (28.5%) and non-small cell lung cancer (20.5%). Administered ICIs included nivolumab (41.4%) and ipilimumab (23.2%). On manual chart review, 147 (10.0%), 1,287 (87.9%), and 30 (2%) patients were defined as IR-colitis cases, controls, and equivocal cases, respectively. For the best-performing rule-based digital phenotype, sensitivity was 0.80 (95% CI, 0.73 to 0.86), specificity was 0.97 (95% CI, 0.96 to 0.98), positive predictive value was 0.74 (95% CI, 0.66 to 0.80), negative predictive value was 0.98 (95% CI, 0.97 to 0.98), accuracy was 0.89 (95% CI, 0.85 to 0.92), positive likelihood ratio (LR) was 24.6 (95% CI, 18.08 to 33.47), and negative LR was 0.2 (95% CI, 0.15 to 0.28). The F1 score was 0.77.
CONCLUSION: We successfully developed an accurate digital phenotype for IR-colitis using real-world EMR data. It allows rapid and accurate cohort identification, which may be adapted to support retrospective immune-related adverse event surveillance and translational biomarker research in the real world.
METHODS: Adult patients commenced on ICIs from August 1, 2020, to June 30, 2023, at the Peter MacCallum Cancer Centre were studied. ICIs were prescribed as monotherapy or in combination with other treatment modalities for a variety of solid cancers. Deidentified EMR data were extracted. A rule-based digital phenotype was developed using expert-selected features, with manual chart review serving as the gold standard comparator.
RESULTS: The ICI-treated cohort comprised 1,464 adults, with a male predominance (62%). The most frequent cancer diagnoses were melanoma (28.5%) and non-small cell lung cancer (20.5%). Administered ICIs included nivolumab (41.4%) and ipilimumab (23.2%). On manual chart review, 147 (10.0%), 1,287 (87.9%), and 30 (2%) patients were defined as IR-colitis cases, controls, and equivocal cases, respectively. For the best-performing rule-based digital phenotype, sensitivity was 0.80 (95% CI, 0.73 to 0.86), specificity was 0.97 (95% CI, 0.96 to 0.98), positive predictive value was 0.74 (95% CI, 0.66 to 0.80), negative predictive value was 0.98 (95% CI, 0.97 to 0.98), accuracy was 0.89 (95% CI, 0.85 to 0.92), positive likelihood ratio (LR) was 24.6 (95% CI, 18.08 to 33.47), and negative LR was 0.2 (95% CI, 0.15 to 0.28). The F1 score was 0.77.
CONCLUSION: We successfully developed an accurate digital phenotype for IR-colitis using real-world EMR data. It allows rapid and accurate cohort identification, which may be adapted to support retrospective immune-related adverse event surveillance and translational biomarker research in the real world.