Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Learning the irreversible progression trajectory of Alzheimer's disease

Yipei Wang, Bing He, Shannon Risacher, Andrew Saykin, Jingwen Yan, Xiaoqian Wang

TL;DR

The proposed novel regularization approach outperforms existing techniques in capturing the progressiveness of disease risk, and at the same time preserves prediction accuracy, and is evaluated using the longitudinal structural MRI and amyloid-PET imaging data from the ADNI.

Abstract

Alzheimer's disease (AD) is a progressive and irreversible brain disorder that unfolds over the course of 30 years. Therefore, it is critical to capture the disease progression in an early stage such that intervention can be applied before the onset of symptoms. Machine learning (ML) models have been shown effective in predicting the onset of AD. Yet for subjects with follow-up visits, existing techniques for AD classification only aim for accurate group assignment, where the monotonically increasing risk across follow-up visits is usually ignored. Resulted fluctuating risk scores across visits violate the irreversibility of AD, hampering the trustworthiness of models and also providing little value to understanding the disease progression. To address this issue, we propose a novel regularization approach to predict AD longitudinally. Our technique aims to maintain the expected monotonicity of increasing disease risk during progression while preserving expressiveness. Specifically, we introduce a monotonicity constraint that encourages the model to predict disease risk in a consistent and ordered manner across follow-up visits. We evaluate our method using the longitudinal structural MRI and amyloid-PET imaging data from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Our model outperforms existing techniques in capturing the progressiveness of disease risk, and at the same time preserves prediction accuracy.

Learning the irreversible progression trajectory of Alzheimer's disease

TL;DR

The proposed novel regularization approach outperforms existing techniques in capturing the progressiveness of disease risk, and at the same time preserves prediction accuracy, and is evaluated using the longitudinal structural MRI and amyloid-PET imaging data from the ADNI.

Abstract

Alzheimer's disease (AD) is a progressive and irreversible brain disorder that unfolds over the course of 30 years. Therefore, it is critical to capture the disease progression in an early stage such that intervention can be applied before the onset of symptoms. Machine learning (ML) models have been shown effective in predicting the onset of AD. Yet for subjects with follow-up visits, existing techniques for AD classification only aim for accurate group assignment, where the monotonically increasing risk across follow-up visits is usually ignored. Resulted fluctuating risk scores across visits violate the irreversibility of AD, hampering the trustworthiness of models and also providing little value to understanding the disease progression. To address this issue, we propose a novel regularization approach to predict AD longitudinally. Our technique aims to maintain the expected monotonicity of increasing disease risk during progression while preserving expressiveness. Specifically, we introduce a monotonicity constraint that encourages the model to predict disease risk in a consistent and ordered manner across follow-up visits. We evaluate our method using the longitudinal structural MRI and amyloid-PET imaging data from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Our model outperforms existing techniques in capturing the progressiveness of disease risk, and at the same time preserves prediction accuracy.
Paper Structure (11 sections, 4 equations, 4 figures, 1 table)

This paper contains 11 sections, 4 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methods
  3. Datasets
  4. Model Formulation
  5. Irreversible Subject Trajectories
  6. Regularization
  7. Experiments
  8. Expressiveness vs Violation Ratio
  9. Total Normalized Violation Gap
  10. Visualizations of Trajectories
  11. Conclusions

Figures (4)

  • Figure 1: Illustration of fluctuating and expected individual progression trajectories across follow-up visits. Each connected line indicates one subject with multiple visits. Left: fluctuating trajectories commonly seen in raw data and from existing classification models which rely on baseline data only. Right: Expected individual trajectory with predicted risk monotonically increasing across follow-up visits.
  • Figure 2: Illustration of the violation ratio -- accuracy of comparing methods on the Amyloid-PET (left) and the MRI (Right) datasets. The top left corner (higher accuracy, lower violation ratio) indicates better results.
  • Figure 3: Illustration of the relation among neighbor/complete violation ratio/gap. They are represented by $r_{nb,cp},\omega_{nb,cp}$, respectively. Results show that the linear relation between the violation gap and the violation ratio is of great significance ($p=$9.995e-14, 2.609e-15). And the measurements based on the neighbor pairs are linearly consistent with complete pairs.
  • Figure 4: The progression trajectory of individual subjects learned from MLP (left) and RMLP (right) using Amyloid-PET (top) and MRI (bottom) test data. Red connected dots represent subjects with multiple follow-up visits. The black dashed horizontal line represents the decision boundary.