Prediction of high and low disease activity in early MS patients using multiple kernel learning identifies importance of lateral ventricle intensity

Abstract

Background Lack of easy-to-interpret disease activity prediction methods in early MS can lead to worse patient prognosis. Objectives Using machine learning (multiple kernel learning – MKL) models, we assessed the prognostic value of various clinical and MRI measures for disease activity. Methods Early MS patients ( n = 148) with at least two associated clinical and MRI visits were investigated. T2-weighted MRIs were cropped to contain mainly the lateral ventricles (LV). High disease activity was defined as surpassing NEDA-3 Criteria more than once per year. Clinical demographic, MRI-extracted image-derived phenotypes (IDP), and MRI data were used as inputs for separate kernels to predict future disease activity with MKL. Model performance was compared using bootstrapped effect size analysis of mean differences. Results A total of 681 visits were included, where 81 (55%) patients had high disease activity in a combined end point measure using all follow-up visits. MKL model discrimination performance was moderate (AUC ≥ 0.62); however, modelling with combined clinical and cropped LV kernels gave the highest prediction performance (AUC = 0.70). Conclusions MRIs contain valuable information on future disease activity, especially in and around the LV. MKL techniques for combining different data types can be used for the prediction of disease activity in a relatively small MS cohort.

A review on multiple sclerosis prognostic findings from imaging, inflammation, and mental health studies

“…There have been several studies in the last few years that have given rise to interesting information related to brain health and MS. Recently, it has been found that ML can be used to predict MS future disease activity in patients using unprocessed MRIs, where within the periventricular region and CP there seems to be higher T2-weighted intensities in people with higher future disease activity (Chien et al, 2022). Along the same line, unsupervised ML in combination with advanced statistical methods have been used to identify different MS-related MRI-extracted subtypes that lead to different confirmed disability progression and relapse rates (Eshaghi et al, 2021).…”

“…Recently, the choroid plexus (CP) within the lateral ventricles showed enlarged volumes in MS patients (Gauthier, 2023) and were found to occur in conjunction with chronic lesions and brain atrophy (Klistorner et al, 2022). Enlargement of the CP have also been indicated as markers of inflammatory/acute disease activity (Fleischer et al, 2021;Ricigliano et al, 2021;Margoni et al, 2023); where increased gadolinium enhancement in T1-weighted MRIs (Kim et al, 2020) and T2-weighted intensity have predicted higher disease activity in MS patients (Chien et al, 2022).…”

A Review of Pattern Recognition and Machine Learning

“…In their study, Chien, Seiler, Eitel, Schmitz-Hübsch, Paul, and Ritter [10] have suggested MKL (Multiple Kernel Learning) framework as a diagnostic tool for Alzheimer's disease (AD), which involves the integration of diverse data sources. The authors propose a methodology for generating multimodal indicators of Alzheimer's disease (AD) by utilizing diffusion tensor imaging (DTI) to extract biomarkers from adjacent images.…”

Multi-Angle Fast Neural Tangent Kernel Classifier

“…Anirban et al [21] proposed a multiple kernel learning embedded multi-objective swarm intelligence technique to identify the candidate biomarker genes from the transcriptomic profile of arsenicosis samples. Using multiple kernel learning MKL models, Chien et al [22] assessed the predictive value of various clinical and MRI measures for disease activity. Jiang et al [23] proposed a high-order norm-product regularized multiple kernel learning framework to optimize the discrimination performance.…”