Centre for Health Services Research in Medicine, Department of Psychiatry and Psychotherapy, Uniklinikum Erlangen, Schwabachanlage 6, 91054, Erlangen, Germany.
BMC Med. 2024 Oct 15;22(1):472. doi: 10.1186/s12916-024-03647-x.
Computerised cognitive training (CCT) can improve the cognitive abilities of people with mild cognitive impairment (MCI), especially when the CCT contains a learning system, which is a type of machine learning (ML) that automatically selects exercises at a difficulty that corresponds to the person's peak performance and thus enables individualised training.
We developed one individualised CCT (iCCT) with ML and one basic CCT (bCCT) for an active control group (CG). The study aimed to determine whether iCCT in the intervention group (IG) resulted in significantly greater enhancements in overall cognitive functioning for individuals with MCI (age 60+) compared with bCCT in the CG across a 6-month period. This double-blind randomised controlled study was conducted entirely virtually. The 89 participants were community-dwelling people with a psychometric diagnosis of MCI living in Germany. The iCCT stimulates various cognitive functions, especially working memory, visuo-constructional reasoning, and decision-making. The bCCT includes fewer and simpler tasks. Both CCTs were used at home. At baseline and after 6 months, we assessed cognitive functioning with the Montreal Cognitive Assessment (MoCA). A mixed-model ANCOVA was conducted as the main analysis.
Both CCTs led to significant increases in average global cognition. The estimated marginal means of the MoCA score increased significantly in the CG by an average of 0.9 points (95% CI [0.2, 1.7]) from 22.3 (SE = 0.25) to 23.2 (SE = 0.41) points (p = 0.018); in the IG, the MoCA score increased by an average of 2.2 points (95% CI [1.4, 2.9]) from 21.9 (SE = 0.26) to 24.1 (SE = 0.42) points (p < 0.001). In a confound-adjusted multiple regression model, the interaction between time and group was statistically significant (F = 4.92; p = 0.029). The effect size was small to medium (partial η = 0.057). On average, the participants used the CCTs three times per week with an average duration of 34.9 min per application. The iCCT was evaluated as more attractive and more stimulating than the bCCT.
By using a multi-tasking CCT three times a week for 30 min, people with MCI living at home can significantly improve their cognitive abilities within 6 months. The use of ML significantly increases the effectiveness of cognitive training and improves user satisfaction.
ISRCTN14437015; registered February 27, 2020.
计算机认知训练（CCT）可以改善轻度认知障碍（MCI）患者的认知能力，尤其是当 CCT 包含学习系统时，学习系统是一种机器学习（ML），它可以自动选择与个人最佳表现相对应的难度的练习，从而实现个性化训练。
我们为实验组（IG）开发了一个具有 ML 的个体化 CCT（iCCT）和一个基本 CCT（bCCT）作为对照组（CG）。该研究旨在确定在 6 个月的时间内，IG 中的 iCCT 是否会导致 MCI 患者（年龄在 60 岁以上）的整体认知功能显著增强，而 CG 中的 bCCT 则没有。这是一项完全虚拟的双盲随机对照研究。89 名参与者是居住在德国的社区居民，通过心理测量诊断为 MCI。iCCT 刺激各种认知功能，特别是工作记忆、视空间建构推理和决策。bCCT 包括较少和较简单的任务。两种 CCT 都在家中使用。在基线和 6 个月后，我们使用蒙特利尔认知评估（MoCA）评估认知功能。主要分析采用混合模型方差分析。
两种 CCT 均导致平均整体认知显著增加。CG 的 MoCA 评分的估计边缘均值显著增加，平均增加 0.9 分（95%CI [0.2, 1.7]），从 22.3（SE=0.25）到 23.2（SE=0.41）（p=0.018）；IG 的 MoCA 评分平均增加 2.2 分（95%CI [1.4, 2.9]），从 21.9（SE=0.26）到 24.1（SE=0.42）（p<0.001）。在经过混杂因素调整的多元回归模型中，时间和组之间的交互作用具有统计学意义（F=4.92；p=0.029）。效应大小为小到中等（部分 η=0.057）。平均而言，参与者每周使用 CCT 三次，每次使用时间约为 34.9 分钟。iCCT 被评估为比 bCCT 更具吸引力和刺激性。
通过每周使用多任务 CCT 三次，每次 30 分钟，在家中的 MCI 患者可以在 6 个月内显著提高认知能力。使用 ML 可以显著提高认知训练的效果，并提高用户满意度。
ISRCTN83405147；注册于 2020 年 2 月 27 日。
