Machine-learning-based prediction of disability progression in multiple sclerosis: An observational, international, multi-center study
Status PubMed-not-MEDLINE Jazyk angličtina Země Spojené státy americké Médium electronic-ecollection
Typ dokumentu časopisecké články
PubMed
39052668
PubMed Central
PMC11271865
DOI
10.1371/journal.pdig.0000533
PII: PDIG-D-23-00247
Knihovny.cz E-zdroje
- Publikační typ
- časopisecké články MeSH
BACKGROUND: Disability progression is a key milestone in the disease evolution of people with multiple sclerosis (PwMS). Prediction models of the probability of disability progression have not yet reached the level of trust needed to be adopted in the clinic. A common benchmark to assess model development in multiple sclerosis is also currently lacking. METHODS: Data of adult PwMS with a follow-up of at least three years from 146 MS centers, spread over 40 countries and collected by the MSBase consortium was used. With basic inclusion criteria for quality requirements, it represents a total of 15, 240 PwMS. External validation was performed and repeated five times to assess the significance of the results. Transparent Reporting for Individual Prognosis Or Diagnosis (TRIPOD) guidelines were followed. Confirmed disability progression after two years was predicted, with a confirmation window of six months. Only routinely collected variables were used such as the expanded disability status scale, treatment, relapse information, and MS course. To learn the probability of disability progression, state-of-the-art machine learning models were investigated. The discrimination performance of the models is evaluated with the area under the receiver operator curve (ROC-AUC) and under the precision recall curve (AUC-PR), and their calibration via the Brier score and the expected calibration error. All our preprocessing and model code are available at https://gitlab.com/edebrouwer/ms_benchmark, making this task an ideal benchmark for predicting disability progression in MS. FINDINGS: Machine learning models achieved a ROC-AUC of 0⋅71 ± 0⋅01, an AUC-PR of 0⋅26 ± 0⋅02, a Brier score of 0⋅1 ± 0⋅01 and an expected calibration error of 0⋅07 ± 0⋅04. The history of disability progression was identified as being more predictive for future disability progression than the treatment or relapses history. CONCLUSIONS: Good discrimination and calibration performance on an external validation set is achieved, using only routinely collected variables. This suggests machine-learning models can reliably inform clinicians about the future occurrence of progression and are mature for a clinical impact study.
1 Biostat Hasselt University Belgium
AHEPA University Hospital Thessaloniki Greece
American University of Beirut Medical Center Beirut Lebanon
AZ Alma Ziekenhuis Sijsele Damme Belgium
Azienda Ospedaliera di Rilievo Nazionale San Giuseppe Moscati Avellino Avellino Italy
Azienda Sanitaria Unica Regionale Marche AV3 Macerata Italy
Bakirkoy Education and Research Hospital for Psychiatric and Neurological Diseases Istanbul Turkey
BAZ County Hospital Miskolc Hungary
Biobix Department of Data Analysis and Mathematical Modelling Ghent University Belgium
Biomedical Research Institute Hasselt University Belgium
Box Hill Hospital Melbourne Australia
Brain Ghent University Belgium
Centro Hospitalar Universitario de Sao Joao Porto Portugal
Charles University Prague and General University Hospital Prague Czech Republic
CHUM and Université de Montreal Montreal Canada
CISSS Chaudière Appalache Levis Canada
Cliniques Universitaires Saint Luc Brussels Belgium
College of Medicine and Health Sciences and Sultan Qaboos University Hospital SQU Oman
Concord Repatriation General Hospital Sydney Australia
CORe Department of Medicine University of Melbourne Melbourne Australia
Data Science Institute Hasselt University Belgium
Department of Medical and Surgical Sciences and Advanced Technologies GF Ingrassia Catania Italy
Department of Neurology Buffalo General Medical Center Buffalo United States of America
Department of Neurology Ghent University Belgium
Dept of Rehabilitation CRFF Mons Luigi Novarese Moncrivello Italy
ESAT STADIUS KU Leuven Belgium
Geneva University Hospital Geneva Switzerland
Groene Hart Ziekenhuis Gouda Netherlands
Hospital Clinic de Barcelona Barcelona Spain
Hospital de Galdakao Usansolo Galdakao Spain
Hospital Fernandez Capital Federal Argentina
Hospital General Universitario de Alicante Alicante Spain
Hospital Universitario Donostia San Sebastián Spain
Hospital Universitario Virgen Macarena Sevilla Spain
Jahn Ferenc Teaching Hospital Budapest Hungary
KERMIT Department of Data Analysis and Mathematical Modelling Ghent University Belgium
King Fahad Specialist Hospital Dammam Khobar Saudi Arabia
Koc University School of Medicine Istanbul Turkey
Liverpool Hospital Sydney Australia
Mater Dei Hospital Msida Malta
Mayis University Samsun Turkey
Melbourne MS Centre Department of Neurology Royal Melbourne Hospital Melbourne Australia
MS center UOC Neurologia ARNAS Garibaldi Catania Italy
Nemocnice Jihlava Jihlava Czech Republic
Noorderhart ziekenhuizen Pelt Belgium
Ospedali Riuniti di Salerno Salerno Italy
Royal Hobart Hospital Hobart Australia
Royal Victoria Hospital Belfast United Kingdom
School for Mental Health and Neuroscience Maastricht University Maastricht The Netherlands
Semmelweis University Budapest Budapest Hungary
South Eastern HSC Trust Belfast United Kingdom
St Michael's Hospital Toronto Canada
St Vincent's University Hospital Dublin Ireland
SUMO IDLAB Ghent University imec Belgium
The Alfred Hospital Melbourne Australia
Universidade Metropolitana de Santos Santos Brazil
Universitair MS Centrum Hasselt Pelt Belgium
Universitary Hospital Ghent Ghent Belgium
University Hospital Reina Sofia Cordoba Spain
University Newcastle Newcastle Australia
Mateen BA, Liley J, Denniston AK, Holmes CC, Vollmer SJ. Improving the quality of machine learning in health applications and clinical research. Nature Machine Intelligence. 2020;2(10):554–556. doi: 10.1038/s42256-020-00239-1 DOI
De Brouwer E, Gonzalez J, Hyland S. Predicting the impact of treatments over time with uncertainty aware neural differential equations. In: International Conference on Artificial Intelligence and Statistics. PMLR; 2022. p. 4705–4722.
De Brouwer E, Krishnan RG. Anamnesic Neural Differential Equations with Orthogonal Polynomial Projections. In: The Eleventh International Conference on Learning Representations; 2022.
Vaswani A, Shazeer N, Parmar N, Uszkoreit J, Jones L, Gomez AN, et al. Attention is all you need. In: Advances in neural information processing systems; 2017. p. 5998–6008.
Gal Y, Ghahramani Z. Dropout as a bayesian approximation: Representing model uncertainty in deep learning. In: international conference on machine learning. PMLR; 2016. p. 1050–1059.
Platt JC. Probabilistic outputs for support vector machines and comparisons to regularized likelihood methods. Advances in large margin classifiers. 1999;10(3):61–74.
Zadrozny B, Elkan C. Transforming classifier scores into accurate multiclass probability estimates. In: Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining; 2002. p. 694–699.