Division of Biomedical Informatics, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA.
Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH, USA.
Acta Neurol Scand. 2021 Jul;144(1):41-50. doi: 10.1111/ane.13418. Epub 2021 Mar 26.
Epilepsy surgery is underutilized. Automating the identification of potential surgical candidates may facilitate earlier intervention. Our objective was to develop site-specific machine learning (ML) algorithms to identify candidates before they undergo surgery.
MATERIALS & METHODS: In this multicenter, retrospective, longitudinal cohort study, ML algorithms were trained on n-grams extracted from free-text neurology notes, EEG and MRI reports, visit codes, medications, procedures, laboratories, and demographic information. Site-specific algorithms were developed at two epilepsy centers: one pediatric and one adult. Cases were defined as patients who underwent resective epilepsy surgery, and controls were patients with epilepsy with no history of surgery. The output of the ML algorithms was the estimated likelihood of candidacy for resective epilepsy surgery. Model performance was assessed using 10-fold cross-validation.
There were 5880 children (n = 137 had surgery [2.3%]) and 7604 adults with epilepsy (n = 56 had surgery [0.7%]) included in the study. Pediatric surgical patients could be identified 2.0 years (range: 0-8.6 years) before beginning their presurgical evaluation with AUC =0.76 (95% CI: 0.70-0.82) and PR-AUC =0.13 (95% CI: 0.07-0.18). Adult surgical patients could be identified 1.0 year (range: 0-5.4 years) before beginning their presurgical evaluation with AUC =0.85 (95% CI: 0.78-0.93) and PR-AUC =0.31 (95% CI: 0.14-0.48). By the time patients began their presurgical evaluation, the ML algorithms identified pediatric and adult surgical patients with AUC =0.93 and 0.95, respectively. The mean squared error of the predicted probability of surgical candidacy (Brier scores) was 0.018 in pediatrics and 0.006 in adults.
Site-specific machine learning algorithms can identify candidates for epilepsy surgery early in the disease course in diverse practice settings.
癫痫手术的应用不足。自动化识别潜在手术候选者可能有助于更早进行干预。我们的目标是开发特定于部位的机器学习（ML）算法，以便在患者接受手术之前识别出候选者。
在这项多中心、回顾性、纵向队列研究中，从自由文本神经学记录、脑电图和磁共振成像报告、就诊代码、药物、程序、实验室和人口统计学信息中提取 n-gram，用于训练特定于部位的 ML 算法。在两个癫痫中心（一个儿科中心和一个成人中心）开发部位特定算法。病例定义为接受切除术治疗的癫痫患者，对照组为有癫痫但无手术史的患者。ML 算法的输出是接受切除术治疗的癫痫候选者的估计可能性。使用 10 倍交叉验证评估模型性能。
共有 5880 名儿童（n=137 例接受手术[2.3%]）和 7604 名成人癫痫患者（n=56 例接受手术[0.7%]）纳入研究。儿科手术患者可以在开始术前评估前 2.0 年（范围：0-8.6 年）识别出来，AUC=0.76（95%CI：0.70-0.82），PR-AUC=0.13（95%CI：0.07-0.18）。成人手术患者可以在开始术前评估前 1.0 年（范围：0-5.4 年）识别出来，AUC=0.85（95%CI：0.78-0.93），PR-AUC=0.31（95%CI：0.14-0.48）。当患者开始术前评估时，ML 算法分别识别出儿科和成人手术患者，AUC 分别为 0.93 和 0.95。预测手术候选者概率的均方误差（Brier 分数）在儿科为 0.018，在成人中为 0.006。
特定于部位的机器学习算法可以在疾病早期在不同的实践环境中识别出癫痫手术的候选者。
