Inexpensive, non-invasive biomarkers predict Alzheimer transition using machine learning analysis of the Alzheimer’s Disease Neuroimaging (ADNI) database

Abstract

The Alzheimer's Disease Neuroimaging (ADNI) database is an expansive undertaking by government, academia, and industry to pool resources and data on subjects at various stage of symptomatic severity due to Alzheimer's disease. As expected, magnetic resonance imaging is a major component of the project. Full brain images are obtained at every 6-month visit. A range of cognitive tests studying executive function and memory are employed less frequently. Two blood draws (baseline, 6 months) provide samples to measure concentrations of approximately 145 plasma biomarkers. In addition, other diagnostic measurements are performed including PET imaging, cerebral spinal fluid measurements of amyloid-beta and tau peptides, as well as genetic tests, demographics, and vital signs. ADNI data is available upon review of an application. There have been numerous reports of how various processes evolve during AD progression, including alterations in metabolic and neuroendocrine activity, cell survival, and cognitive behavior. Lacking an analytic model at the onset, we leveraged recent advances in machine learning, which allow us to deal with large, non-linear systems with many variables. Of particular note was examining how well binary predictions of future disease states could be learned from simple, non-invasive measurements like those dependent on blood samples. Such measurements make relatively little demands on the time and effort of medical staff or patient. We report findings with recall/ precision/area under the receiver operator curve after application of CART, Random Forest, Gradient Boosting, and Support Vector Machines, Our results show (i) Random Forests and Gradient Boosting work very well with such data, (ii) Prediction quality when applied to relatively easily obtained measurements (Cognitive scores, Genetic Risk and plasma biomarkers) achieve results that are competitive with magnetic resonance techniques. This is by no

Predicting progression and cognitive decline in amyloid-positive patients with Alzheimer’s disease

“…We found that the predictability of changes in cognitive test scores was low, leaving a large portion of variance unexplained. Our results complement previous works which show good discrimination of progressing and non-progressing subjects [ 16 , 21 ] in cohorts comprising both A β -positive and A β -negative subjects. In particular, we show that discriminating between subjects who are potential candidates for drugs designed to reverse or slow down A β plaque formation presents a harder prediction task.…”

“…Most predictive models of neurodegenerative diseases are based on recent advances in machine learning (ML) models by obtaining data sets with measurements of cognition and neuropathology from large cohorts [ 16 , 20 – 22 ]. In this context, classification methods such as random forest [ 13 , 21 , 23 , 24 ] and logistic regression (LR) [ 21 , 25 – 27 ] have been used to predict whether individuals will decline or remain stable in their diagnosis.…”

Association of peripheral blood DNA methylation level with Alzheimer’s disease progression

“…A neural network model based on data from 1737 participants in the ADNI database was effective in predicting AD progression in a separate set of 110 participants who were CN or had MCI at baseline [ 14 ]. Other machine learning-based models have employed neuroimaging and noninvasive methods using blood biomarkers to predict AD progression [ 15 , 16 ].…”

Predicting Progression & Cognitive Decline in Amyloid-Positive Patients with Alzheimer's Disease

“…The predictive models of neurodegenerative diseases are based on recent advances in machine learning (ML) models by obtaining data sets with measurements of cognition and neuropathology from large cohorts [24,25,20,26]. In this context, most ML models focused on binary classification methods predicting with high accuracy whether individuals diagnosed with MCI will decline or remain stable in their diagnosis [27].…”

“…In this context, most ML models focused on binary classification methods predicting with high accuracy whether individuals diagnosed with MCI will decline or remain stable in their diagnosis [27]. The classification techniques such as random forest [15,28,29,25], logistic regression (LR) [30,31,32,25] and support vector machines (SVM) [33,29] have been used. Most previous AD studies based on ML focus on predicting conversion between the CN, MCI and AD diagnosis [20,34] based on diagnosis classification using various neuroimaging tech-niques such as magnetic resonance imaging (MRI).…”