AI-assisted MS care is moving from a future vision toward practical clinical decision support — and this episode of the MS-Perspektive Podcast, supported by the European Charcot Foundation, explores how the CLAIMS project could help make MS treatment more precise and individualized.
I speak with Prof. Friedemann Paul, neurologist and neuroimmunologist at Charité – Universitätsmedizin Berlin, about how CLAIMS combines MRI, clinical data, OCT, digital health information and patient-reported outcomes to better predict MS progression and support treatment decisions.
The goal is not to replace doctors. The goal is to give clinicians better tools and people with MS more clarity when making important choices about their care.
You are currently viewing a placeholder content from Default. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationTable of Contents
In short
CLAIMS is a European project that aims to support more precise MS care by combining MRI, OCT, blood biomarkers, clinical data and patient-reported outcomes. The goal is not to replace neurologists, but to provide better decision support for individualized treatment choices. In this interview, Prof. Friedemann Paul explains how AI-assisted MS care could help identify progression earlier, distinguish RAW and PIRA, and support shared decision making between doctors and people with MS.
Introduction and personal background
Could you briefly introduce yourself, and what drew you personally to working on the CLAIMS project?
Prof. Friedemann Paul:
I have been involved with neuroinflammatory disorders of the central nervous system, including multiple sclerosis, for more than 20 years. At Charité in Berlin, Germany, we see around 1,500 to 2,000 patients with MS every year.
One of the biggest unmet needs is disease prediction. People want to know what their prognosis will look like, which treatment they should choose, and how MS may affect their private, professional and family life.
That is exactly where the CLAIMS project comes in.
Could you explain in simple terms what the CLAIMS project aims to achieve?
Prof. Friedemann Paul:
CLAIMS is an EU-funded project with funding from several industry partners. It is an AI-assisted MS care project.
The basic idea is to provide a better prognosis for people with multiple sclerosis by integrating many large datasets.
This includes clinical and epidemiological data, imaging data such as MRI scans and OCT, as well as patient-reported outcomes.
MRI, or magnetic resonance imaging, visualizes the brain and spinal cord and is central in MS diagnosis and monitoring.
OCT, or optical coherence tomography, is a non-invasive eye scan that measures retinal structures and can provide information about neurodegeneration, meaning damage or loss of nerve tissue.
CLAIMS aims to create a tool that can support treating physicians with more precise, more individualized information about the disease course and possible treatment decisions.
From your clinical and research perspective, how would you describe the promise of AI-assisted care for MS patients in one sentence?
Prof. Friedemann Paul:
The promise is to make predictions for individual patients more precise and to support better treatment decisions in daily clinical practice.
AI can help us analyze complex data that no physician could fully process manually within a short consultation.
But this is very important: AI should not replace doctors. It should support them.
What does the term precision medicine mean to you in MS care — and why is it such an important goal?
Prof. Friedemann Paul:
Precision medicine means looking at the individual person and their specific situation.
For example, someone is newly diagnosed with MS after optic neuritis and has recovered well. MRI, spinal fluid, blood tests and other assessments point toward MS.
Now the physician and the patient need to decide whether to start immunotherapy and, if so, which treatment is best.
This decision should not be based on one single factor. It should include all relevant aspects: MRI findings, spinal fluid results, clinical symptoms, age, sex, individual risks, comorbidities and personal circumstances.
Precision medicine also means taking the patient’s preferences into account.
For example, one therapy may be more effective but also have more risks. Another may be easier to integrate into everyday life. For one person, family planning may be important. For another, work, travel or lifestyle may play a major role.
So precision medicine means making an individualized treatment decision, together with the patient, based on the best available data.
Unfortunately, some patients are also at risk of converting from relapsing-remitting MS to progressive disease after years of living with MS. We still need better tools to detect that risk early and to find the right point to adapt treatment.
Project vision and implementation
How do you integrate diverse data types — like MRI scans, clinical scores and digital health tools — and how do you ensure that this information is actually usable in a busy clinical routine?
Prof. Friedemann Paul:
This is one of the key challenges.
In clinical routine, physicians often have very little time. You cannot expect an experienced neurologist, even an MS specialist, to integrate all complex data manually in five, ten or fifteen minutes.
A decision support tool could bring these different data together in the background. It could then provide clear, usable information to the treating physician.
For example, the platform might show that the OCT is stable, the MRI looks stable, the EDSS has not changed, and the overall risk is low.
EDSS stands for Expanded Disability Status Scale. It is a commonly used scale to describe disability in MS.
In another case, the tool might show new lesions, increasing brain volume loss or worsening patient-reported outcomes. Then the physician could discuss whether treatment should be changed.
The aim is to give the physician a concrete, accessible risk profile that can support the conversation with the patient.
The tool needs to work in the background. It must integrate different data sources and then deliver information in a format that is easy to use.
For example, MRI findings should not just be described in long, narrative reports. The tool should be able to quantify lesions, compare scans over time and detect relevant changes.
Instead of only saying that there are lesions in certain brain regions, it could show whether there are new T2 lesions, whether lesions are located in relevant areas, and whether brain atrophy has progressed.
T2 lesions are areas visible on specific MRI sequences that indicate tissue changes related to MS inflammation or damage.
The same applies to OCT, clinical scores and patient-reported outcomes. All these data should be combined into an understandable clinical picture.
This should not make the consultation more complicated. It should make it more focused.
The platform distinguishes between RAW and PIRA types of MS progression. Could you briefly define these terms and explain why this distinction is clinically important?
Prof. Friedemann Paul:
RAW means relapse-associated worsening. It describes disability worsening that happens in connection with a relapse.
A relapse is a new or clearly worsening neurological symptom caused by inflammatory activity in MS, usually lasting at least 24 hours and not explained by infection or fever.
PIRA means progression independent of relapse activity. It describes worsening that happens without a clear relapse.
This distinction is important because we now know that MS progression is not only driven by relapses.
Some people become worse even though they do not have obvious attacks. They may notice worsening walking ability, more fatigue, cognitive problems or other changes.
Cognition refers to mental functions such as memory, attention, concentration and processing speed.
PIRA may also be reflected in biomarkers, such as neurofilament light chain, often abbreviated as NfL. Neurofilament light chain is a marker of nerve cell damage that can be measured in blood or cerebrospinal fluid.
The challenge is that PIRA can be difficult to recognize in everyday clinical care. But it matters, because patients with early PIRA may need a more active treatment approach.
Research suggests that a substantial proportion of MS patients show PIRA. That is why it is important to identify these patients earlier and more reliably.
From a patient’s perspective, how would interacting with this platform change their care experience?
Prof. Friedemann Paul:
If CLAIMS is successful, care could become more precise and more transparent.
At the moment, MRI reports are often descriptive. They may say that there are multiple lesions in the brain, some in the frontal region, some in the parietal region, some in the cerebellum, and so on.
A structured tool could provide more specific information. It could state how many lesions are present, whether new lesions appeared since the last MRI, and whether brain volume loss is higher than expected.
This would give the neurologist more concrete information to discuss with the patient.
The platform could also include what patients report themselves. For example, a patient may say that fatigue has increased, sleep quality is poor, or cognitive problems at work are becoming more noticeable.
If these patient-reported outcomes are combined with MRI, OCT and clinical data, the neurologist may have stronger reasons to discuss a change of treatment.
For patients, this could mean feeling more seen and better understood — not only through scans, but also through their lived experience.
What makes CLAIMS unique and practically useful?
Many technologies in MS research sound promising — what makes CLAIMS unique and practically useful?
Prof. Friedemann Paul:
CLAIMS is unique because it tries to combine several relevant data sources into one practical decision support system.
It is not only about MRI. It is not only about one biomarker. It is not only about a questionnaire.
It brings together imaging, clinical information, OCT and patient-reported outcomes.
The project also aims to be useful in real clinical routines. That is essential. A tool may be scientifically impressive, but if it cannot be used in a busy outpatient clinic, it will not improve patient care.
CLAIMS tries to provide information that is clinically meaningful, understandable and actionable.
Innovation, challenges and patient impact
What have been the biggest challenges so far — scientifically, logistically or in translating research into clinical tools?
Prof. Friedemann Paul:
One of the biggest challenges in European projects is often the administrative and contractual setup.
There are many partners. There are pharmaceutical partners, academic partners and technical partners. Data protection and legal agreements must be handled carefully.
This took time.
Another challenge is that datasets are heterogeneous. That means they are not all collected in the same way. In retrospective datasets, some information may be missing because the data were collected in the past for clinical reasons, not specifically for this project.
The RECLAIM part of the project was retrospective. The PROCLAIM study is prospective and randomized. This means patients are enrolled and followed according to a defined study plan.
The goal is to see whether the CLAIMS tool has an effect on treatment decisions made by physicians, and ultimately on disease outcomes after several years.
So far, the project has already gained first insights into risk profiles and how certain immunotherapies may influence these risks.
AI raises questions of trust. How do you ensure transparency and confidence among clinicians and patients? How do you communicate AI-based insights to patients in a way they can understand and trust?
Prof. Friedemann Paul:
Several aspects are important.
First, it must be transparent how the algorithms work, which data the algorithms were trained with, and where these data came from.
Second, the source data must be good. If the source data are poor, the outcome will also be poor.
Third, the neurologist needs to understand how the algorithm operates. The physician does not need to be able to code the algorithm, but should understand which kinds of data contributed to the recommendation.
Imagine the algorithm recommends switching to a more potent MS therapy.
The physician should not simply say: “The algorithm says so, so we do it.”
Instead, the physician needs to ask whether the recommendation makes sense. Was the MS diagnosis correct? Were other diseases excluded? Were comorbidities considered? Is the patient at higher risk because of age, vascular risk factors or other health issues?
Comorbidities are other medical conditions that exist alongside MS, such as high blood pressure or diabetes.
AI should support the doctor’s thinking, not replace it.
The best approach is that the doctor interprets the AI-supported insight and then discusses it clearly with the patient.
This is also why AI will not replace neurologists. It may help save time and improve the precision of care. It may also help identify patients who need specialist attention earlier.
For example, in rural areas, where there may be a shortage of neurologists, a screening tool could help show which patients need urgent specialist care.
But the final decision still belongs to the experienced physician and the patient.
AI is only as good as the data it receives. If the source data are poor, the output will also be poor.
So AI should not replace doctors. Hopefully, it will become a useful tool that supports better decision making.
Timeline, accessibility and participation
What is the anticipated timeline for key milestones — like validation, approval and rollout?
Prof. Friedemann Paul:
At the time of the interview, enrollment into the PROCLAIM study was ongoing.
The hope was to complete enrollment by the end of the year and then analyze the data during the following year.
The broader aim mentioned in the conversation was to have the tool ready toward the end of the next year and to move toward regulatory approval and reimbursement by around 2028.
But of course, this depends on data analysis, validation, regulatory processes and many factors that are not fully under the project team’s control.
Where will CLAIMS become available first — and what barriers do you still see for implementation?
Prof. Friedemann Paul:
First, the project has to demonstrate that it makes a real difference for medical decisions in MS.
If that works, the next step is regulatory approval.
Some of the icometrix MRI tools have already been through regulatory clearance, but this specific decision support tool would need additional approval.
Another key issue is reimbursement. Even if a tool is approved, it also needs to be paid for by healthcare systems.
The aim should be that patients can benefit from the tool without additional personal costs.
Whether this becomes possible across Europe and beyond will depend on healthcare systems, approval pathways and reimbursement decisions.
Can people with MS or clinicians still participate in CLAIMS-related studies?
Prof. Friedemann Paul:
The PROCLAIM study is open in Berlin, Dresden and other centers (as of June 2026).
Interested people with MS can check the CLAIMS website, contact participating centers or reach out to the study teams to see whether participation is possible.
Clinicians can also stay involved through conferences, scientific networks, publications and collaboration with participating centers.
What will it take for healthcare providers to adopt this in daily care — training, infrastructure, reimbursement?
Prof. Friedemann Paul:
Infrastructure may be less of a barrier than expected because hospitals and practices are becoming more digital.
The tool should be technically implementable in many settings.
But reimbursement will be an important issue. Training will also be necessary. Physicians need to understand what the tool can do, what it cannot do, and how they should interpret its recommendations.
There may also be differences between generations. Younger physicians who grew up with digital tools may adopt AI-supported systems more quickly.
But acceptance will depend above all on whether the tool is useful, reliable and easy to integrate into clinical practice.
How can listeners stay updated and get involved as the project moves forward?
Prof. Friedemann Paul:
The project publishes newsletters and information through its website.
People can also follow conference presentations, scientific publications and updates from participating institutions.
The European Charcot Foundation supports dissemination through its networks and publications as well.
Anyone interested can sign up for newsletters, follow project updates and look for opportunities to participate in studies if they are available locally.
Reflection and closing
Looking back so far, what is one moment in the project that felt especially meaningful or surprising to you?
Prof. Friedemann Paul:
One surprising aspect was how long the administrative setup took.
This is not specific to CLAIMS. It happens in many European projects. When many countries, institutions and partners are involved, contracts and data protection agreements take time.
But scientifically, it has been meaningful to analyze the RECLAIM data.
The team was able to identify low, intermediate and high-risk categories. These risk groups may be more accurate and more precise than conventional MRI reports alone.
That is promising because it suggests that integrated data can give us more useful information than isolated data points.
What is one thing you wish every MS patient or clinician understood about AI-assisted precision medicine?
Prof. Friedemann Paul:
We cannot make reliable predictions from one single dataset alone.
We need large datasets from clinical trials, real-world cohorts and different sources.
Only then can we make stronger inferences about prognosis, response to immunotherapy and possible disease course.
The more high-quality data we can integrate, the more precise the prediction may become.
That is the core idea of AI-assisted precision medicine.
In the next five years, where do you hope the CLAIMS project — and the broader MS community — will be?
Prof. Friedemann Paul:
I hope we will have a better understanding of disease courses.
I hope we will be more proactive, more precise and more individualized with our treatments.
Ideally, this will mean less disability, a better prognosis for many patients and less burden on healthcare systems.
If you could change one thing right now about the MS care paradigm, what would it be and why?
Prof. Friedemann Paul:
I would like to see more integrative care.
This means using the best possible immunotherapy for a given person in a given situation, but not stopping there.
We also need pregnancy counseling, lifestyle medicine, physical activity, rehabilitation and attention to other factors that influence MS prognosis.
The neurologist should not only give a drug and leave the patient alone. Patients can also actively influence their disease course through lifestyle, activity, rehabilitation and shared care decisions.
This more holistic approach could have a major impact on the prognosis of multiple sclerosis.
What message would you like to leave with our listeners today?
Prof. Friedemann Paul:
There is hope for better management of the disease.
We desperately need this because more and more people are being diagnosed with multiple sclerosis.
And we need more and better data.
If you have time and resources, please consider contributing to research. This could be an observational study, a therapeutic trial, a study on lifestyle or another local research project.
It is really important.
We can only improve outcomes in multiple sclerosis if we do more research.
Farewell
Nele von Horsten:
Thank you, Friedemann. Lovely words at the end. And thank you very much for being my guest and talking about the CLAIMS project. Good luck with everything that is still on the timeline. It would be lovely to soon have such a cool project implemented into real life.
Prof. Friedemann Paul:
Thank you very much. It was good to talk.
See you soon and try to make the best out of your life,
Nele
For more information and positive thoughts, subscribe to my newsletter for free.
Click here for an overview of all podcast episodes published so far.
And at many more places.
* This text contains affiliate links. This means that I get a small compensation if you buy the product recommended by me through the link. For you nothing changes in the price of the product. And it helps me to pay for the blog and to write new posts.
