Weaver John K, Logan Joseph, Van Batavia Jason P, Weiss Dana A, Long Christopher J, Smith Ariana L, Zderic Stephen A, Gan Zoe, Godlewski Karl, Broms Reiley, Antony Maria, Overland Maya, Gaines Tyler, Head Dennis, Erdman Lauren, Viteri Bernarda, Martin-Olenski Madalyne, Huang Jing, Fan Yong, Tasian Gregory E
Division of Pediatric Urology, Cleveland Clinic Children's, Cleveland, Ohio.
Division of Pediatric Urology, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania.
J Urol. 2025 Jul;214(1):80-89. doi: 10.1097/JU.0000000000004547. Epub 2025 Mar 25.
Variability in the interpretation of videourodynamics studies limits reliable classification of kidney injury risk for patients with spina bifida. We developed machine learning models to predict incident hydronephrosis in patients with spina bifida using videourodynamics data.
We trained machine learning models using data from videourodynamics studies performed between 2016 and 2022 on patients with spina bifida aged 2 months to 42 years. We evaluated the performance of 4 models to predict incident hydronephrosis following an index videourodynamics study: (1) random survival forest model using data prospectively abstracted from videourodynamics studies by urologists, (2) random survival forest of bladder volume-pressure data, (3) random survival forest using deep learning features extracted from fluoroscopic images of the bladder, (4) ensemble model averaging the probabilities of the volume-pressure and fluoroscopic models.
We included 354 and 200 patients in the training and validation cohorts, respectively. Among the training and validation cohorts, 89 (25.1%) and 71 (35.5%) patients developed incident hydronephrosis at a median time of 1.6 (IQR, 0.5-3) and 2.49 (IQR, 1.72-3.03) years after the index videourodynamics study, respectively. The ensemble model that included data from studies during which ≥ 75% expected bladder capacity was reached had the best discrimination (C statistic 0.73; 95% CI, 0.68-0.76). The specificity of high-risk scores (top 10% in the ensemble model) was 97%.
Automated extraction of features from pressure/volume recordings and fluoroscopic images of the bladder predicted incident hydronephrosis in patients with spina bifida.
视频尿动力学研究解读的变异性限制了对脊柱裂患者肾损伤风险的可靠分类。我们开发了机器学习模型，以利用视频尿动力学数据预测脊柱裂患者发生肾盂积水的情况。
我们使用2016年至2022年期间对年龄在2个月至42岁的脊柱裂患者进行的视频尿动力学研究数据训练机器学习模型。我们评估了4种模型在索引视频尿动力学研究后预测肾盂积水发生情况的性能：（1）使用泌尿科医生从视频尿动力学研究中前瞻性提取的数据的随机生存森林模型，（2）膀胱容量-压力数据的随机生存森林模型，（3）使用从膀胱荧光透视图像中提取的深度学习特征的随机生存森林模型，（4）对容量-压力模型和荧光透视模型的概率进行平均的集成模型。
我们分别在训练队列和验证队列中纳入了354例和200例患者。在训练队列和验证队列中，分别有89例（25.1%）和71例（35.5%）患者在索引视频尿动力学研究后的中位时间1.6（四分位间距，0.5 - 3）年和2.49（四分位间距，1.72 - 3.03）年发生肾盂积水。包含达到≥75%预期膀胱容量的研究数据的集成模型具有最佳的辨别力（C统计量0.73；95%置信区间，0.68 - 0.76）。高风险评分（集成模型中前10%）的特异性为97%。
从膀胱压力/容量记录和荧光透视图像中自动提取特征可预测脊柱裂患者发生肾盂积水的情况。
