A comparison of machine learning methods for survival analysis of high-dimensional clinical data for dementia prediction

Abstract

Data collected from clinical trials and cohort studies, such as dementia studies, are often high-dimensional, censored, heterogeneous and contain missing information, presenting challenges to traditional statistical analysis. There is an urgent need for methods that can overcome these challenges to model this complex data. At present there is no cure for dementia and no treatment that can successfully change the course of the disease. Machine learning models that can predict the time until a patient develops dementia are important tools in helping understand dementia risks and can give more accurate results than traditional statistical methods when modelling high-dimensional, heterogeneous, clinical data. This work compares the performance and stability of ten machine learning algorithms, combined with eight feature selection methods, capable of performing survival analysis of high-dimensional, heterogeneous, clinical data. We developed models that predict survival to dementia using baseline data from two different studies. The Sydney Memory and Ageing Study (MAS) is a longitudinal cohort study of 1037 participants, aged 70–90 years, that aims to determine the effects of ageing on cognition. The Alzheimer's Disease Neuroimaging Initiative (ADNI) is a longitudinal study aimed at identifying biomarkers for the early detection and tracking of Alzheimer's disease. Using the concordance index as a measure of performance, our models achieve maximum performance values of 0.82 for MAS and 0.93 For ADNI.

Cure models to estimate time until hospitalization due to COVID-19

“…This leads to conclusions that can be completed with the results of this work, since this will give information of when that hospital admission will take place. However, machine learning-based approaches are not the best option when working with censored data, although they are specially helpful when handling high-dimensional clinical data [ 64 ]. Note that, in a context with censored data, it is not possible to apply directly machine learning classical models since they do not account for censored observations.…”

FRL: An Integrative Feature Selection Algorithm Based on the Fisher Score, Recursive Feature Elimination, and Logistic Regression to Identify Potential Genomic Biomarkers

“…Because of the curse of dimensionality, the high-dimensional data from gene expression profiles present challenges for the use of traditional feature selection methods, including overfitting, weak generalization ability, and high variance [ 36 ]. The relationship between the samples and features of the cancer datasets is formulated by the following matrix: where x m is defined as the m link of the characteristic vector, and y m describes the column vector representing the sample categories.…”

“…Integrative Feature Selection Scheme (FRL) for Identifying Multiple Genomic Biomarkers. Because of the curse of dimensionality, the high-dimensional data from gene expression profiles present challenges for the use of traditional feature selection methods, including overfitting, weak generalization ability, and high variance [36]. The relationship between the samples and features of the cancer datasets is formulated by the following matrix:…”

Current Trends in Readmission Prediction: An Overview of Approaches

“…Without the presence of censoring, standard LR could be used. Traditionally, the Cox proportional hazard (CPH) model has been the most widely used model to analyse censored data, but the CPH model often works for small datasets and does not scale well to high dimensions and large volumes of clinical data [ 26 ].…”