Department of Clinical Biostatistics, Graduate School of Medical and Dental Sciences, Tokyo Medical and Dental University, Bunkyo-ku, Tokyo, Japan.
Innovation Center for Translational Research, National Center for Geriatrics and Gerontology, Obu, Aichi, Japan.
Stat Med. 2022 Sep 20;41(21):4200-4214. doi: 10.1002/sim.9504. Epub 2022 Jun 24.
Alzheimer's disease (AD) is a chronic neurodegenerative disease characterized by a gradual decline in cognitive function over a few decades. The Mini-Mental State Examination (MMSE) is a widely used measure for evaluating global cognitive functioning. Characterizing the longitudinal trajectory of the MMSE in the population of interest is important to detect AD onset for preventive intervention. In this study, we formulate a new class of longitudinal trajectory modeling for MMSE from short-term individual data based on an ordinary differential equation. The proposed method models the relationship between individual decline speed of MMSE and the average MMSE using the fractional polynomial function model and subsequently estimates the longitudinal trajectory of MMSE by solving the ordinary differential equation for the estimated model. The appropriate model for trajectory estimation is selected based on the proposed criterion for quantifying the goodness of trajectory fit. The accuracy of the trajectory estimation of the proposed method was demonstrated via simulation studies. The proposed method was successfully applied to MMSE data from the Japanese Alzheimer's Disease Neuroimaging Initiative study.
阿尔茨海默病（AD）是一种慢性神经退行性疾病，其认知功能在几十年内逐渐下降。简易精神状态检查（MMSE）是评估整体认知功能的常用方法。描述感兴趣人群中 MMSE 的纵向轨迹对于检测 AD 发病并进行预防性干预很重要。在这项研究中，我们基于常微分方程，为基于短期个体数据的 MMSE 提出了一类新的纵向轨迹建模方法。该方法使用分数多项式函数模型来模拟 MMSE 个体下降速度与平均 MMSE 之间的关系，然后通过求解所估计模型的常微分方程来估计 MMSE 的纵向轨迹。基于用于量化轨迹拟合优劣的标准，选择适当的轨迹估计模型。通过模拟研究验证了该方法的轨迹估计准确性。该方法成功应用于日本阿尔茨海默病神经影像学倡议研究中的 MMSE 数据。
