SciLifeLab NBIS and Data Centre contribute to the development of an AI model improving MS diagnostics
A new AI-based tool developed by researchers at Uppsala University helps clinicians diagnose progressive forms of multiple sclerosis (MS) earlier and with greater accuracy. Published in npj Digital Medicine, the project was supported by the SciLifeLab Bioinformatics platform (NBIS) and uses resources from the SciLifeLab Data Centre.
Identifying when a patient transitions from relapsing-remitting to secondary progressive MS is key to providing the right treatment. Today, this shift is often detected several years too late. The new model can determine disease status with around 90 percent accuracy, offering a chance to adapt treatment plans in time.
SciLifeLab contribution
The research team, led by Kim Kultima (associate professor) at Uppsala University, received bioinformatics support through SciLifeLab’s peer-reviewed WABI track. Payam Emami, expert at the SciLifeLab Bioinformatics platform, contributed to the machine learning methods behind the model. The WABI track is funded by the Knut and Alice Wallenberg Foundation and supports data-driven life science projects across Sweden.
The model is available via SciLifeLab Serve, a platform from the SciLifeLab Data Centre that enables hosting and sharing of scientific tools and services.
“One of the challenges in applying machine learning to healthcare is ensuring that models are not only accurate in a statistical sense but also useful in practice. We provided methodological guidance on how to work with temporal clinical data, how to structure the model to reflect real-world scenarios, how to incorporate explainability, and how to evaluate outcomes in a way that matters to clinicians. The result is a tool that reflects both analytical rigor and clinical insight showing how infrastructure-supported expertise can bridge the gap between data science and patient care,” Payam explains.
Built on Swedish registry data
The AI model was trained using clinical data from over 22,000 patients in the Swedish MS Registry. It draws on routinely collected data such as neurological tests, MRI scans, and treatment history. In nearly 87 percent of cases, the model identified the transition to secondary progressive MS earlier than what was documented in the medical records. The overall accuracy was close to 90 percent.
“For patients, this means that the diagnosis can be made earlier, which can make it possible to adjust the patient’s treatment in time and hopefully slow down the progression of the disease. This also reduces the risk of patients receiving medicines that are no longer effective. In the long term, the model could also be used to identify suitable participants for clinical trials, which could contribute to more effective and individualised treatment strategies,” Kim Kultima says in an interview with Uppsala University.
An open, anonymised version of the model is now available to researchers via the web service msp-tracker hosted on SciLifeLab Serve.
