European journal of obstetrics, gynecology, and reproductive biology · 2024
Constructing small for gestational age prediction models: A retrospective machine learning study.
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.
Chen Xinyu, Wu Siqing, Chen Xinqing, Hu Linmin, Li Wenjing, Mi Ningning, Xie Peng, Huang Yujun, Yuan Kun, Sui Yajuan, Li Renjie, Wang Kangting, Sun Nan, Yao Yuyang, Xu Zuofeng, Yuan Jinqiu, Zhu Yunxiao
Аннотация
OBJECTIVE: To develop machine learning prediction models for small for gestational age with baseline characteristics and biochemical tests of various pregnancy stages individually and collectively and compare predictive performance.
STUDY DESIGN: This retrospective study included singleton pregnancies with infants born between May 2018 and March 2023. Small for gestational age was defined as a birth weight below the 10th percentile according to the Intergrowth-21st fetal growth standards. The pregnancy data were categorized into four datasets at different gestational time points (14 and 28 weeks and admission). The LightGBM framework was utilized to assess the variable importance by employing a five-fold cross-validation. RandomizedSearchCV and sequential feature selection were applied to estimate the optimal number of features. Seven machine learning algorithms were used to develop prediction models, with an 8:2 ratio for training and testing. The model performance was evaluated using receiver operating characteristic curve analysis and sensitivity at a false positive rate of 10 %.
RESULTS: We included data of 4,394 women with singleton pregnancies, including 148 (3.4%) small for gestational age infants. Women delivering small for gestational age infants exhibited significantly shorter stature and lower fundal height and abdominal circumference at admission. Maternal height, age, and pre-pregnancy weight consistently ranked among the top 20 features in prediction models with any dataset. The models incorporated variables of admission stage have strong predictive performance with the area under the curves exceeding 0.8. The prediction model developed with variables of admission stage yielded the best performance, achieving an area under the curve of 0.85 and a sensitivity of 73% at the false positive rate of 10%.
CONCLUSIONS: By machine learning, various pregnancy stages' prediction models for small for gestational age showed good predictive performance, and the predictive value of variables at each pregnancy stage was fully explored. The prediction model with the best performance was established with variables of admission stage and emphasized the significance of prenatal physical examinations.
STUDY DESIGN: This retrospective study included singleton pregnancies with infants born between May 2018 and March 2023. Small for gestational age was defined as a birth weight below the 10th percentile according to the Intergrowth-21st fetal growth standards. The pregnancy data were categorized into four datasets at different gestational time points (14 and 28 weeks and admission). The LightGBM framework was utilized to assess the variable importance by employing a five-fold cross-validation. RandomizedSearchCV and sequential feature selection were applied to estimate the optimal number of features. Seven machine learning algorithms were used to develop prediction models, with an 8:2 ratio for training and testing. The model performance was evaluated using receiver operating characteristic curve analysis and sensitivity at a false positive rate of 10 %.
RESULTS: We included data of 4,394 women with singleton pregnancies, including 148 (3.4%) small for gestational age infants. Women delivering small for gestational age infants exhibited significantly shorter stature and lower fundal height and abdominal circumference at admission. Maternal height, age, and pre-pregnancy weight consistently ranked among the top 20 features in prediction models with any dataset. The models incorporated variables of admission stage have strong predictive performance with the area under the curves exceeding 0.8. The prediction model developed with variables of admission stage yielded the best performance, achieving an area under the curve of 0.85 and a sensitivity of 73% at the false positive rate of 10%.
CONCLUSIONS: By machine learning, various pregnancy stages' prediction models for small for gestational age showed good predictive performance, and the predictive value of variables at each pregnancy stage was fully explored. The prediction model with the best performance was established with variables of admission stage and emphasized the significance of prenatal physical examinations.