Neuro-oncology advances · 2025
Sociodemographic factors predict outcomes and reveal spatial tumor patterns in glioblastoma.
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.
Luckett Patrick H, Olufawo Michael, Naddaff-Slocum Noah, Park Ki Yun, Lamichhane Bidhan, Dierker Donna, Verastegui Gabriel Trevino, Lee John J, Yang Peter, Kim Albert, Butt Omar H, Chheda Milan G, Snyder Abraham Z, Shimony Joshua S, Leuthardt Eric C
Аннотация
BACKGROUND: Glioblastoma (GBM) is the deadliest malignant glioma of the central nervous system. Postsurgical functional impairment correlates with survival and is estimated using metrics such as extent and location of resection. This study uses machine learning to evaluate the predictive ability of baseline sociodemographic and lifestyle factors for forecasting postoperative functional outcomes in GBM patients.
METHODS: Glioblastoma patients (N = 115) from the neurosurgery brain tumor service at Washington University were retrospectively identified. All patients underwent neuroimaging, surgical resection of the GBM, and at least 3 postoperative follow-up visits. Demographic, lifestyle, and socioeconomic factors (socioeconomic status [SES] and Social Vulnerability Index [SVI]) were used to train decision tree classifiers to predict postoperative Karnofsky Performance Status (KPS ≤ 70, KPS > 70), as well as classify the change in KPS (KPS slope) over multiple visits (decreased/improved or maintained constant).
RESULTS: Utilizing decision trees with age, SVI/SES, sex, tobacco use, alcohol use, obesity, and race as predictors, we achieved 88% accuracy in classifying median KPS and 85% accuracy in classifying KPS slope. Socioeconomic factors were the strongest predictors. Age, sex, and tobacco use were also strong predictors. Significant correlations in spatial tumor distributions were observed based on outcome measures and SVI/SES.
CONCLUSIONS: The current work demonstrates the utility of machine learning to predict functional outcomes in GBM patients prior to treatment using lifestyle and sociodemographic factors. Our results suggest that socioeconomic factors, age, tobacco use, and biological sex can be reliable predictors of functional outcomes. Incorporating these factors could improve therapeutic approaches tailored to individual patients.
METHODS: Glioblastoma patients (N = 115) from the neurosurgery brain tumor service at Washington University were retrospectively identified. All patients underwent neuroimaging, surgical resection of the GBM, and at least 3 postoperative follow-up visits. Demographic, lifestyle, and socioeconomic factors (socioeconomic status [SES] and Social Vulnerability Index [SVI]) were used to train decision tree classifiers to predict postoperative Karnofsky Performance Status (KPS ≤ 70, KPS > 70), as well as classify the change in KPS (KPS slope) over multiple visits (decreased/improved or maintained constant).
RESULTS: Utilizing decision trees with age, SVI/SES, sex, tobacco use, alcohol use, obesity, and race as predictors, we achieved 88% accuracy in classifying median KPS and 85% accuracy in classifying KPS slope. Socioeconomic factors were the strongest predictors. Age, sex, and tobacco use were also strong predictors. Significant correlations in spatial tumor distributions were observed based on outcome measures and SVI/SES.
CONCLUSIONS: The current work demonstrates the utility of machine learning to predict functional outcomes in GBM patients prior to treatment using lifestyle and sociodemographic factors. Our results suggest that socioeconomic factors, age, tobacco use, and biological sex can be reliable predictors of functional outcomes. Incorporating these factors could improve therapeutic approaches tailored to individual patients.