Abstract:
Amyotrophic lateral sclerosis (ALS) is a debilitating neurodegenerative condition leading to progressive muscle weakness, atrophy, and ultimately death. Traditional ALS clinical evaluations often depend on subjective metrics, making accurate disease detection and monitoring disease trajectory challenging. To address these limitations, we developed the nQiALS toolkit, a machine learning-powered system that leverages smartphone typing dynamics to detect and track motor impairment in people with ALS. The study included 63 ALS patients and 30 age- and sex-matched healthy controls. We introduce the three core components of this toolkit: the nQiALS-Detection, which differentiated ALS from healthy typing patterns with an AUC of 0.89; the nQiALS-Progression, which separated slow and fast progression at specific thresholds with AUCs ranging between 0.65 and 0.8; and the nQiALS-Fine Motor, which identified subtle progression in fine motor dysfunction, suggesting earlier prediction than the state-of-the-art assessment. Together, these tools represent an innovative approach to ALS assessment, offering a complementary, objective metric to traditional clinical methods and which may reshape our understanding and monitoring of ALS progression.
摘要:
肌萎缩侧索硬化症(ALS)是一种使人衰弱的神经退行性疾病,导致进行性肌肉无力,萎缩,最终死亡。传统的ALS临床评估通常取决于主观指标,使准确的疾病检测和监测疾病轨迹具有挑战性。为了解决这些限制,我们开发了nQiALS工具包,一个机器学习驱动的系统,利用智能手机打字动力学来检测和跟踪ALS患者的运动障碍。该研究包括63名ALS患者和30名年龄和性别匹配的健康对照。我们介绍这个工具包的三个核心组件:nQiALS检测,将ALS与健康分型模式区分开来,AUC为0.89;nQiALS-Progression,在特定阈值下分离缓慢和快速进展,AUC范围在0.65和0.8之间;和nQiALS精细运动,确定了精细运动功能障碍的微妙进展,这表明预测比最先进的评估更早。一起,这些工具代表了ALS评估的创新方法,提供一个补充,对传统临床方法的客观度量,这可能会重塑我们对ALS进展的理解和监测。