Strategies for integrating artificial intelligence and cognitive assessment to predict disability progression in relapsing-remitting multiple sclerosis: A model development study
18 affiliations
- 1Department of Neuroscience, School of Translational Medicine, Monash University, Melbourne, Victoria, Australia
- 2Department of Electrical and Computer Systems Engineering, Monash University, Melbourne, Victoria, Australia
- 3AIM for Health Lab, Faculty of IT, Monash University, Melbourne, Victoria, Australia
- 4Discipline of Pediatric and Child Health, School of Clinical Medicine, University of New South Wales, Sydney, New South Wales, Australia
- 5Department of Neurology, Alfred Hospital, Melbourne, Victoria, Australia
- 6Menzies Institute for Medical Research, University of Tasmania, Hobart, Tasmania, Australia
- 7Department of Neurology, Neuroimmunology Centre, Royal Melbourne Hospital, Melbourne, Victoria, Australia
- 8CORe, Department of Medicine, University of Melbourne, Melbourne, Victoria, Australia
- 9Hunter New England Health, New Lambton, New South Wales, Australia
- 10University of Newcastle, Newcastle, New South Wales, Australia
- 11Eastern Health Clinical School, Monash University, Box Hill, Victoria, Australia
- 12Florey Department of Neuroscience and Mental Health, University of Melbourne, Melbourne, Victoria, Australia
- 13Brain and Mind Centre, University of Sydney, Sydney, New South Wales, Australia
- 14Sydney Neuroimaging Analysis Centre, Camperdown, New South Wales, Australia
- 15School of Public Health and Preventive Medicine, Monash University, Melbourne, Victoria, Australia
- 16Department of Data Science and AI, Faculty of IT, Monash University, Melbourne, Victoria, Australia
- 17School of Computing Technologies, RMIT University, Victoria, Australia
- 18Faculty of IT, Monash-Airdoc Research Center, Monash University, Melbourne, Victoria, Australia
Abstract
Serial cognitive assessment in multiple sclerosis can help identify patients at higher risk of disability progression, but high dimensionality of reaction-time data makes analysis challenging. We developed a method to convert longitudinal reaction-time data into images and applied deep survival modelling to predict disability progression in relapsing-remitting MS (RRMS). In this cohort study, clinical data were obtained from the MSBase registry and cognitive data from the MSReactor computerised cognitive battery between February 2016 and September 2022, with a median follow-up of 3.2 years. The serial reaction-time data from three tasks (psychomotor function [R], attention [G], and working memory [B]) were converted into multicolour images using RGB encoding. We extracted the key features from the images using a convolutional neural network and combined them with clinical variables in a Transformer-based survival model (MS-TranSurv) to predict time to confirmed disability progression. Discrimination, calibration, and accuracy were assessed using the C-index, integrated Brier score (iBS), and time-dependent area under the receiver operating characteristic (tAUROC). Performance was compared with other machine learning models, including Dynamic DeepHit, Recurrent Deep Survival Machines, a Cox proportional hazards model in traditional statistics using clinical variables only, and a version of MS-TransSurv using mean test values. A total of 746 RRMS patients were included. MS-TranSurv showed slightly higher discrimination compared with benchmark models, with a C-index of 0.61 (95% CI 0.54–0.68) and tAUROC of 0.74 (95% CI 0.63–0.85), as well as comparable calibration, with an iBS of 0.24 (95% CI 0.16–0.32). Using individual test-level data provided slightly better performance than using summary measures. Longitudinal cognitive reaction-time data can be used for survival-based prediction of disability progression in RRMS. This framework supports remote cognitive monitoring and may improve risk stratification and clinical trial enrichment, with potential applicability to other high-dimensional digital biomarkers. Author summary: People with multiple sclerosis often experience changes in thinking speed, attention, and memory, sometimes years before clear physical disability appears. Digital cognitive tests make it possible to measure these changes repeatedly, including at home, but the large amount of data they produce is difficult to analyse and is often simplified in ways that may lose important information. In this study, we developed a new approach to make better use of these data. We transformed repeated reaction-time measurements from simple cognitive tasks into colour-based images that capture patterns of performance over time. We then used a machine learning method designed to analyse changes over time to estimate the risk of future disability progression. By keeping more detailed information from each test, our approach performed better than methods that rely on summary measures. Our findings show that repeated, low-burden cognitive testing can be used to help identify people with multiple sclerosis who are at higher risk of worsening disability. This approach could support earlier clinical review, improve the selection of participants for clinical trials, and may be adapted for use with other types of digital health data in the future.
Related
- Converging Paths: Artificial Intelligence for Integrating Biological and Digital Biomarkers in Personalized Longevity Medicine
- Artificial Intelligence for Predictive Microbiome Engineering: Integrating Community Design and Environmental Validation
- Adaptive Precision Nutrition: Integrating Dynamic Phenotyping, Multi-Omics, Artificial Intelligence, and the Allostatic Viability Principle
- Artificial intelligence-based decision support for axial spondyloarthritis integrating inflammatory and structural MRI information
- Contribution of artificial intelligence research to the development of inclusive applications for people with disabilities: A scoping review (2015–2025)
BibTeX
@article{zhu2026strategies,
title = {{Strategies for integrating artificial intelligence and cognitive assessment to predict disability progression in relapsing-remitting multiple sclerosis: A model development study}},
author = {Chao Zhu and Xin Zhang and Anneke van der Walt and Bruce Taylor and Mastura Monif and Tomas Kalincik and Jeanette Lechner-Scott and Katherine Buzzard and Trevor Kilpatrick and Michael Barnett and Zhen Zhou and Vilija Jokubaitis and Melissa Gresle and David Darby and Deval Mehta and Zongyuan Ge and Helmut Butzkueven and Daniel Merlo},
journal = {PLOS Digital Health},
year = {2026},
month = sep,
doi = {10.1371/journal.pdig.0001722},
url = {https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0001722},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Strategies for integrating artificial intelligence and cognitive assessment to predict disability progression in relapsing-remitting multiple sclerosis: A model development study
AU - Chao Zhu
AU - Xin Zhang
AU - Anneke van der Walt
AU - Bruce Taylor
AU - Mastura Monif
AU - Tomas Kalincik
AU - Jeanette Lechner-Scott
AU - Katherine Buzzard
AU - Trevor Kilpatrick
AU - Michael Barnett
AU - Zhen Zhou
AU - Vilija Jokubaitis
AU - Melissa Gresle
AU - David Darby
AU - Deval Mehta
AU - Zongyuan Ge
AU - Helmut Butzkueven
AU - Daniel Merlo
PY - 2026
DA - 2026/09/25
JO - PLOS Digital Health
DO - 10.1371/journal.pdig.0001722
UR - https://journals.plos.org/digitalhealth/article?id=10.1371/journal.pdig.0001722
AB - Serial cognitive assessment in multiple sclerosis can help identify patients at higher risk of disability progression, but high dimensionality of reaction-time data makes analysis challenging. We developed a method to convert longitudinal reaction-time data into images and applied deep survival modelling to predict disability progression in relapsing-remitting MS (RRMS). In this cohort study, clinical data were obtained from the MSBase registry and cognitive data from the MSReactor computerised cognitive battery between February 2016 and September 2022, with a median follow-up of 3.2 years. The serial reaction-time data from three tasks (psychomotor function [R], attention [G], and working memory [B]) were converted into multicolour images using RGB encoding. We extracted the key features from the images using a convolutional neural network and combined them with clinical variables in a Transformer-based survival model (MS-TranSurv) to predict time to confirmed disability progression. Discrimination, calibration, and accuracy were assessed using the C-index, integrated Brier score (iBS), and time-dependent area under the receiver operating characteristic (tAUROC). Performance was compared with other machine learning models, including Dynamic DeepHit, Recurrent Deep Survival Machines, a Cox proportional hazards model in traditional statistics using clinical variables only, and a version of MS-TransSurv using mean test values. A total of 746 RRMS patients were included. MS-TranSurv showed slightly higher discrimination compared with benchmark models, with a C-index of 0.61 (95% CI 0.54–0.68) and tAUROC of 0.74 (95% CI 0.63–0.85), as well as comparable calibration, with an iBS of 0.24 (95% CI 0.16–0.32). Using individual test-level data provided slightly better performance than using summary measures. Longitudinal cognitive reaction-time data can be used for survival-based prediction of disability progression in RRMS. This framework supports remote cognitive monitoring and may improve risk stratification and clinical trial enrichment, with potential applicability to other high-dimensional digital biomarkers. Author summary: People with multiple sclerosis often experience changes in thinking speed, attention, and memory, sometimes years before clear physical disability appears. Digital cognitive tests make it possible to measure these changes repeatedly, including at home, but the large amount of data they produce is difficult to analyse and is often simplified in ways that may lose important information. In this study, we developed a new approach to make better use of these data. We transformed repeated reaction-time measurements from simple cognitive tasks into colour-based images that capture patterns of performance over time. We then used a machine learning method designed to analyse changes over time to estimate the risk of future disability progression. By keeping more detailed information from each test, our approach performed better than methods that rely on summary measures. Our findings show that repeated, low-burden cognitive testing can be used to help identify people with multiple sclerosis who are at higher risk of worsening disability. This approach could support earlier clinical review, improve the selection of participants for clinical trials, and may be adapted for use with other types of digital health data in the future.
SN - 2767-3170
ER -