Current assessment protocols for attention-deficit/hyperactivity disorder (ADHD) focus heavily on a set of highly overlapping symptoms, with well-validated factors like cognitive disengagement syndrome (CDS), executive function (EF), age, sex, and race and ethnicity generally being ignored. Using machine learning techniques, the current study aimed to validate recent findings proposing a subset of ADHD symptoms that, together, predict ADHD diagnosis, severity, and impairment level better than the full symptom list, while also testing whether the inclusion of the factors listed above could further increase accuracy. Parents of 1,922 children (50.1% male) aged 6-17 years completed rating scales of ADHD, CDS, EF, and impairment. Results suggested nine symptoms as most important in predicting outcomes: (a) has difficulty sustaining attention in tasks or play activities; (b) does not follow through on instructions and fails to finish work; (c) avoids tasks (e.g., schoolwork, homework) that require sustained mental effort; (d) is often easily distracted; (e) has difficulty organizing tasks and activities; (f) is often forgetful in daily activities; (g) fidgets with hands or feet or squirms in seat; (h) interrupts/intrudes on others; and (i) shifts around excessively or feels restless or hemmed in. The abbreviated algorithm achieved accuracy rates that did not significantly differ compared to an algorithm comprising all 18 symptoms in predicting impairment, while also demonstrating excellent discriminative ability in predicting ADHD diagnosis. Adding CDS and EF to the abbreviated algorithm further improved the prediction of global impairment. Continued refinement of screening tools will be key to ensuring access to clinical services for youth at risk for ADHD. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

中文翻译：

---

在以白人为主的样本中预测注意力缺陷/多动障碍和损伤的算法的应用和扩展。


当前注意力缺陷/多动障碍 (ADHD) 的评估方案主要关注一组高度重叠的症状，以及经过充分验证的因素，如认知脱离综合征 (CDS)、执行功能 (EF)、年龄、性别以及种族和民族一般都会被忽视。使用机器学习技术，当前的研究旨在验证最近的发现，提出 ADHD 症状的一个子集，这些症状比完整的症状列表更好地预测 ADHD 的诊断、严重程度和损伤水平，同时还测试是否包含上述因素可以进一步提高准确性。 1,922 名 6-17 岁儿童（50.1% 为男性）的家长完成了 ADHD、CDS、EF 和障碍的评级量表。结果表明，九个症状对于预测结果最重要：(a) 在任务或游戏活动中难以保持注意力； (b) 不按照指示执行、未完成工作的； (c) 避免需要持续脑力劳动的任务（例如学业、家庭作业）； (d) 经常容易分心； (e) 难以组织任务和活动； (f) 在日常活动中经常健忘； (g) 手脚坐立不安或在座位上蠕动； (h) 打扰/侵入他人；在预测损伤方面，简化算法的准确率与包含所有 18 种症状的算法相比没有显着差异，同时在预测 ADHD 诊断方面也表现出出色的辨别能力。在简化算法中添加 CDS 和 EF 进一步改进了全局损伤的预测。不断完善筛查工具将是确保面临多动症风险的青少年获得临床服务的关键。 （PsycInfo 数据库记录 (c) 2024 APA，保留所有权利）。