経時的測定データの個人内変動パターンに基づくイベント発生のリスク評価

概要

論文の詳細を見る
The effects of risk factors on the development of disease are usual1y assessed based on the baseline value of each risk factor in cohort studies. However, such analysis has the potential limitation that the baseline values may not sufficiently explain the development of a disease when the value of the risk factor changes markedly according to time. The Cox proportional hazards model with time-dependent covariates is sometimes applied to analyze the data of cohort studies where the risk factors are measured repeatedly. The repeated measurements are included as time-dependent covariates to consider the changes in risk factors over time. However, the model assesses the risk of a disease according to the value of a risk factor at each time point, but not the pattern of change. Here, we propose a model to assess the effects of risk factors on the development of disease in terms of the baseline value and the rate of change(i.e., intercept and slope)assuming that the value of a risk factor changes linearly according to time. The performance of our proposed model was compared with models that include only the baseline value and time-dependent covariates by Monte-Carlo simulation.
国立保健医療科学院の論文