Zhang Chen, An Lijun, Wulan Naren, Nguyen Kim-Ngan, Orban Csaba, Chen Pansheng, Chen Christopher, Zhou Juan Helen, Liu Keli, Yeo B T Thomas
Centre for Sleep and Cognition (CSC) & Centre for Translational Magnetic Resonance Research (TMR), Yong Loo Lin School of Medicine, National University of Singapore, Singapore.
Department of Electrical and Computer Engineering, National University of Singapore, Singapore.
medRxiv. 2025 Jun 11:2024.11.18.24317513. doi: 10.1101/2024.11.18.24317513.
Accurately predicting Alzheimer's Disease (AD) progression is useful for clinical care. The 2019 TADPOLE (The Alzheimer's Disease Prediction Of Longitudinal Evolution) challenge evaluated 92 algorithms from 33 teams worldwide. Unlike typical clinical prediction studies, TADPOLE accommodates (1) variable number of observed timepoints across patients, (2) missing data across modalities and visits, and (3) prediction over an open-ended time horizon, which better reflects real-world data. However, TADPOLE only used the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset, so how well top algorithms generalize to other cohorts remains unclear.
We tested five algorithms in three external datasets covering 2,312 participants and 13,200 timepoints. The algorithms included FROG, the overall TADPOLE winner, which utilized a unique Longitudinal-to-Cross-sectional (L2C) transformation to convert variable-length longitudinal histories into feature vectors of the same length across participants (i.e., same-length feature vectors). We also considered two FROG variants. One variant unified all XGBoost models from the original FROG with a single feedforward neural network (FNN), which we referred to as L2C-FNN. We also included minimal recurrent neural networks (MinimalRNN), which was ranked second at publication time, as well as AD Course Map (AD-Map), which outperformed MinimalRNN at publication time. All five models - three FROG variants, MinimalRNN and AD-Map - were trained on ADNI and tested on the external datasets.
L2C-FNN performed the best overall. In the case of predicting cognition and ventricle volume, L2C-FNN and AD-Map were the best. For clinical diagnosis prediction, L2C-FNN was the best, while AD-Map was the worst. L2C-FNN also maintained its edge over other models, regardless of the number of observed timepoints, and regardless of the prediction horizon from 0 to 6 years into the future.
L2C-FNN shows strong potential for both short-term and long-term dementia progression prediction. Pretrained ADNI models are available: https://github.com/ThomasYeoLab/CBIG/tree/master/stable_projects/predict_phenotypes/Zhang2025_L2CFNN.
准确预测阿尔茨海默病（AD）的进展对临床护理很有帮助。2019年的TADPOLE（阿尔茨海默病纵向演变预测）挑战赛评估了来自全球33个团队的92种算法。与典型的临床预测研究不同，TADPOLE考虑了以下几点：（1）患者观察到的时间点数量可变；（2）不同模态和访视中的数据缺失；（3）在无固定期限的时间范围内进行预测，这能更好地反映现实世界的数据。然而，TADPOLE仅使用了阿尔茨海默病神经影像学倡议（ADNI）数据集，因此顶级算法在其他队列中的泛化能力如何仍不清楚。
我们在三个外部数据集中测试了五种算法，这些数据集涵盖2312名参与者和13200个时间点。这些算法包括FROG（TADPOLE挑战赛的总冠军），它利用独特的纵向到横断面（L2C）转换，将可变长度的纵向病史转换为所有参与者长度相同的特征向量（即等长特征向量）。我们还考虑了两种FROG变体。一种变体用单个前馈神经网络（FNN）统一了原始FROG中的所有XGBoost模型，我们将其称为L2C-FNN。我们还纳入了当时排名第二的最小递归神经网络（MinimalRNN），以及在当时表现优于MinimalRNN的AD病程图（AD-Map）。所有五个模型——三种FROG变体、MinimalRNN和AD-Map——均在ADNI上进行训练，并在外部数据集上进行测试。
总体而言，L2C-FNN表现最佳。在预测认知和脑室体积方面，L2C-FNN和AD-Map表现最佳。对于临床诊断预测，L2C-FNN最佳，而AD-Map最差。无论观察到的时间点数量如何，也无论预测期是从0到未来6年，L2C-FNN都保持着相对于其他模型的优势。
L2C-FNN在短期和长期痴呆进展预测方面都显示出强大的潜力。预训练的ADNI模型可在以下网址获取：https://github.com/ThomasYeoLab/CBIG/tree/master/stable_projects/predict_phenotypes/Zhang2025_L2CFNN 。
