de Swart Wieske K, Loog Marco, Krijthe Jesse H
Institute for Computing and Information Sciences, Radboud University, Nijmegen, Netherlands.
Pattern Recognition Laboratory, Delft University of Technology, Delft, Netherlands.
Front Neurol. 2025 Feb 18;16:1504535. doi: 10.3389/fneur.2025.1504535. eCollection 2025.
Dynamic survival analysis has become an effective approach for predicting time-to-event outcomes based on longitudinal data in neurology, cognitive health, and other health-related domains. With advancements in machine learning, several new methods have been introduced, often using a two-stage approach: first extracting features from longitudinal trajectories and then using these to predict survival probabilities.
This work compares several combinations of longitudinal and survival models, assessing their predictive performance across different training strategies. Using synthetic and real-world cognitive health data from the Alzheimer's Disease Neuroimaging Initiative (ADNI), we explore the strengths and limitations of each model.
Among the considered survival models, the Random Survival Forest consistently delivered strong results across different datasets, longitudinal models, and training strategies. On the ADNI dataset the best performing method was Random Survival Forest with the last visit benchmark and super landmarking with an average tdAUC of 0.96 and brier score of 0.07. Several other methods, including Cox Proportional Hazards and the Recurrent Neural Network, achieve similar scores. While the tested longitudinal models often struggled to outperform simple benchmarks, neural network models showed some improvement in simulated scenarios with sufficiently informative longitudinal trajectories.
Our findings underscore the importance of aligning model selection and training strategies with the specific characteristics of the data and the target application, providing valuable insights that can inform future developments in dynamic survival analysis.
动态生存分析已成为基于神经学、认知健康及其他健康相关领域的纵向数据预测事件发生时间结果的有效方法。随着机器学习的发展，引入了几种新方法，通常采用两阶段方法：首先从纵向轨迹中提取特征，然后利用这些特征预测生存概率。
本研究比较了纵向模型和生存模型的几种组合，评估它们在不同训练策略下的预测性能。利用来自阿尔茨海默病神经影像学倡议（ADNI）的合成和真实世界认知健康数据，我们探究了每个模型的优势和局限性。
在考虑的生存模型中，随机生存森林在不同数据集、纵向模型和训练策略中始终表现出色。在ADNI数据集上，表现最佳的方法是采用末次访视基准的随机生存森林和超级地标法，平均tdAUC为0.96，布里尔评分0.07。其他几种方法，包括Cox比例风险模型和递归神经网络，也取得了类似的分数。虽然经过测试的纵向模型往往难以超越简单基准，但神经网络模型在具有足够信息的纵向轨迹的模拟场景中显示出一些改进。
我们的研究结果强调了使模型选择和训练策略与数据的特定特征及目标应用保持一致的重要性，提供了有价值的见解，可为动态生存分析的未来发展提供参考。
