International journal of pediatric otorhinolaryngology · 2026
Construction and validation of a nomogram prediction model for postoperative residual risk of pediatric obstructive sleep apnea based on OCTA and LASSO machine learning algorithm.
СтатьяВалидационное исследованиеНазвание 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.
Liu Yin, Tang Yu, Liu Mingna, Yuan Xueshuang, Wu Qingsong, Su Kangjin, Zhu Li, Huang Shanshan, Zhou Fengping, Chen Sijin, Xu Chong
Аннотация
OBJECTIVE: To evaluate the value of retinal optical coherence tomography angiography (OCTA) microcirculatory parameters combined with clinical, biochemical, and polysomnography (PSG) indicators for predicting residual obstructive sleep apnea syndrome (OSAS) in children without clear anatomical or functional anomalies after adenotonsillectomy (AT), and to construct and validate a nomogram prediction model.
METHODS: A total of 157 school-aged children with PSG-diagnosed OSAS who underwent AT were retrospectively enrolled. Baseline demographic, anatomical (adenoid A/N ratio, tonsillar grading), biochemical (high-sensitivity C-reactive protein [hs-CRP], total antioxidant status [TAS]), PSG (apnea-hypopnea index [AHI], lowest oxygen saturation [LSaO2], oxygen desaturation index [ODI]), and OCTA parameters (vessel length density of deep capillary plexus [DCP-VLD]) were collected. Residual OSAS was defined as postoperative AHI >1 event/h with persistent symptoms at 6 months. LASSO-penalized logistic regression was used for feature selection, followed by multivariate logistic regression to build a nomogram. Model performance was assessed by C-index, calibration curve, Hosmer-Lemeshow test, and decision curve analysis (DCA). Longitudinal OCTA changes were analyzed using generalized estimating equations (GEE).
RESULTS: Residual OSAS occurred in 39 cases (24.84%). LASSO identified four core predictors: adenoid A/N ratio, AHI, hs-CRP, and DCP-VLD. Multivariable analysis showed DCP-VLD as an independent protective factor (OR = 0.64, 95%CI: 0.48-0.85, P = 0.0021). The nomogram achieved an AUC of 0.845 (95%CI:0.728-0.962) with good calibration (Hosmer-Lemeshow P = 0.5840). DCA demonstrated superior net clinical benefit. GEE revealed limited DCP-VLD recovery in the residual group (time×group interaction P = 0.0084).
CONCLUSION: Reduced macular DCP perfusion quantified by OCTA is an independent predictor of residual OSAS after AT. The LASSO-logistic nomogram integrating anatomical-metabolic-microcirculatory information shows good discrimination and clinical utility for preoperative risk stratification.
METHODS: A total of 157 school-aged children with PSG-diagnosed OSAS who underwent AT were retrospectively enrolled. Baseline demographic, anatomical (adenoid A/N ratio, tonsillar grading), biochemical (high-sensitivity C-reactive protein [hs-CRP], total antioxidant status [TAS]), PSG (apnea-hypopnea index [AHI], lowest oxygen saturation [LSaO2], oxygen desaturation index [ODI]), and OCTA parameters (vessel length density of deep capillary plexus [DCP-VLD]) were collected. Residual OSAS was defined as postoperative AHI >1 event/h with persistent symptoms at 6 months. LASSO-penalized logistic regression was used for feature selection, followed by multivariate logistic regression to build a nomogram. Model performance was assessed by C-index, calibration curve, Hosmer-Lemeshow test, and decision curve analysis (DCA). Longitudinal OCTA changes were analyzed using generalized estimating equations (GEE).
RESULTS: Residual OSAS occurred in 39 cases (24.84%). LASSO identified four core predictors: adenoid A/N ratio, AHI, hs-CRP, and DCP-VLD. Multivariable analysis showed DCP-VLD as an independent protective factor (OR = 0.64, 95%CI: 0.48-0.85, P = 0.0021). The nomogram achieved an AUC of 0.845 (95%CI:0.728-0.962) with good calibration (Hosmer-Lemeshow P = 0.5840). DCA demonstrated superior net clinical benefit. GEE revealed limited DCP-VLD recovery in the residual group (time×group interaction P = 0.0084).
CONCLUSION: Reduced macular DCP perfusion quantified by OCTA is an independent predictor of residual OSAS after AT. The LASSO-logistic nomogram integrating anatomical-metabolic-microcirculatory information shows good discrimination and clinical utility for preoperative risk stratification.