Eye (London, England) · 2026
Optimised multivariable prediction model to predict treatment requirement in preterm infants with retinopathy of prematurity.
СтатьяПоддержка исследования: вне правительства СШАНазвание и аннотация: Перевод пока не подготовлен
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.
Toyama Taku, Zhou Han Peng, Sugimoto Sao, Hayashi Kentaro, Nagahara Masako, Kuriyama Takuya, Mihara Gen, Nakajima Kosuke, Ueta Takashi
Аннотация
BACKGROUND: Retinopathy of prematurity (ROP) is a leading cause of blindness in preterm infants. However, frequent fundus examinations place burdens on neonates, families, and healthcare staff. We aimed to develop a model to identify low-risk infants and reduce the number of patients requiring screening.
METHODS: We conducted a single-centre retrospective cohort study at the University of Tokyo Hospital (October 2019-December 2024), including infants with birth weight ≤1800 g or gestational age <34 weeks (n = 298). Fourteen variables-birth weight, gestational age (GD), sex, growth velocity (GV), SpO₂/FiO₂ ratio, CRP, platelet count (Plt), PLR, NLR, LMR, SII, haemoglobin, albumin, and infection status-were evaluated at 28, 30, 32, and 34 postmenstrual weeks. We performed logistic regression on all variable combinations and selected the optimal model using the Akaike Information Criterion. ROC analysis was conducted by fixing sensitivity at 1.0 and selecting the threshold that maximised specificity. This threshold was then simulated in a hypothetical cohort of 100 infants to estimate the reduction in exam need.
RESULTS: The optimal model included GD, sex, GV, NLR_max, Plt_min, and PLR_max. Discriminative performance (AUC) improved with age: 0.70 (28 weeks), 0.78 (30), 0.82 (32), and 0.85 (34). Evaluations at 32 and 34 weeks maintained sensitivity while reducing the number of infants requiring examination by ~50%.
CONCLUSIONS: This model can identify low-risk infants from 32 weeks postmenstrual age, potentially halving the number of patients needing fundus exams. All predictors are routinely available in NICUs. External validation in independent multicenter, multi-ethnic, and international prospective cohorts is warranted before clinical implementation.
METHODS: We conducted a single-centre retrospective cohort study at the University of Tokyo Hospital (October 2019-December 2024), including infants with birth weight ≤1800 g or gestational age <34 weeks (n = 298). Fourteen variables-birth weight, gestational age (GD), sex, growth velocity (GV), SpO₂/FiO₂ ratio, CRP, platelet count (Plt), PLR, NLR, LMR, SII, haemoglobin, albumin, and infection status-were evaluated at 28, 30, 32, and 34 postmenstrual weeks. We performed logistic regression on all variable combinations and selected the optimal model using the Akaike Information Criterion. ROC analysis was conducted by fixing sensitivity at 1.0 and selecting the threshold that maximised specificity. This threshold was then simulated in a hypothetical cohort of 100 infants to estimate the reduction in exam need.
RESULTS: The optimal model included GD, sex, GV, NLR_max, Plt_min, and PLR_max. Discriminative performance (AUC) improved with age: 0.70 (28 weeks), 0.78 (30), 0.82 (32), and 0.85 (34). Evaluations at 32 and 34 weeks maintained sensitivity while reducing the number of infants requiring examination by ~50%.
CONCLUSIONS: This model can identify low-risk infants from 32 weeks postmenstrual age, potentially halving the number of patients needing fundus exams. All predictors are routinely available in NICUs. External validation in independent multicenter, multi-ethnic, and international prospective cohorts is warranted before clinical implementation.