Department of Occupational Therapy, College of Medical Science, Soonchunhyang University, Asan, Republic of Korea.
J Med Internet Res. 2024 Oct 30;26:e59247. doi: 10.2196/59247.
Conventional neuropsychological screening tools for mild cognitive impairment (MCI) face challenges in terms of accuracy and practicality. Digital health solutions, such as unobtrusively capturing smartphone interaction data, offer a promising alternative. However, the potential of digital biomarkers as a surrogate for MCI screening remains unclear, with few comparisons between smartphone interactions and existing screening tools.
This study aimed to investigate the effectiveness of smartphone-derived keystroke dynamics, captured via the Neurokeys keyboard app, in distinguishing patients with MCI from healthy controls (HCs). This study also compared the discriminant performance of these digital biomarkers against the Korean version of the Montreal Cognitive Assessment (MoCA-K), which is widely used for MCI detection in clinical settings.
A total of 64 HCs and 47 patients with MCI were recruited. Over a 1-month period, participants generated 3530 typing sessions, with 2740 (77.6%) analyzed for this study. Keystroke metrics, including hold time and flight time, were extracted. Receiver operating characteristics analysis was used to assess the sensitivity and specificity of keystroke dynamics in discriminating between HCs and patients with MCI. This study also explored the correlation between keystroke dynamics and MoCA-K scores.
Patients with MCI had significantly higher keystroke latency than HCs (P<.001). In particular, latency between key presses resulted in the highest sensitivity (97.9%) and specificity (96.9%). In addition, keystroke dynamics were significantly correlated with the MoCA-K (hold time: r=-.468; P<.001; flight time: r=-.497; P<.001), further supporting the validity of these digital biomarkers.
These findings highlight the potential of smartphone-derived keystroke dynamics as an effective and ecologically valid tool for screening MCI. With higher sensitivity and specificity than the MoCA-K, particularly in measuring flight time, keystroke dynamics can serve as a noninvasive, scalable, and continuous method for early cognitive impairment detection. This novel approach could revolutionize MCI screening, offering a practical alternative to traditional tools in everyday settings.
Thai Clinical Trials Registry TCTR20220415002; https://www.thaiclinicaltrials.org/show/TCTR20220415002.
传统的轻度认知障碍（MCI）神经心理学筛查工具在准确性和实用性方面面临挑战。数字健康解决方案，如不引人注目的智能手机交互数据采集，提供了一种很有前途的替代方案。然而，智能手机交互作为 MCI 筛查替代指标的潜力尚不清楚，智能手机交互与现有筛查工具之间的比较也很少。
本研究旨在探讨通过 Neurokeys 键盘应用程序采集的智能手机衍生击键动力学在区分 MCI 患者和健康对照者（HC）中的有效性。本研究还比较了这些数字生物标志物与在临床环境中广泛用于 MCI 检测的韩国版蒙特利尔认知评估（MoCA-K）的判别性能。
共招募了 64 名 HCs 和 47 名 MCI 患者。在 1 个月的时间内，参与者生成了 3530 次打字会话，其中 2740 次（77.6%）用于本研究。提取击键指标，包括保持时间和飞行时间。使用受试者工作特征分析评估击键动力学区分 HCs 和 MCI 患者的敏感性和特异性。本研究还探讨了击键动力学与 MoCA-K 评分之间的相关性。
MCI 患者的击键潜伏期明显高于 HCs（P<.001）。特别是，键按下之间的潜伏期产生了最高的敏感性（97.9%）和特异性（96.9%）。此外，击键动力学与 MoCA-K 显著相关（保持时间：r=-.468；P<.001；飞行时间：r=-.497；P<.001），进一步支持了这些数字生物标志物的有效性。
这些发现强调了智能手机衍生击键动力学作为 MCI 筛查的有效和生态有效工具的潜力。与 MoCA-K 相比，特别是在测量飞行时间方面，击键动力学具有更高的敏感性和特异性，可作为一种非侵入性、可扩展和连续的早期认知障碍检测方法。这种新方法可能会彻底改变 MCI 筛查，为传统工具在日常环境中的应用提供一种实用的替代方案。
泰国临床试验注册中心 TCTR20220415002；https://www.thaiclinicaltrials.org/show/TCTR20220415002。
