A Blueprint for Continuous MS Care — and Why a Blueprint Is Not Yet Evidence

Imagine a triage rule with three steps. The patient self-monitors at home through validated apps and a wearable. An automated layer watches the incoming data and flags anything that looks clinically relevant. And only then, on a validated alert, does a neurologist's time get spent. That tiered logic is the load-bearing idea in a recent perspective on multiple sclerosis care in npj Digital Medicine, and it is a good one — because continuous data is an asset only if most of it can be safely ignored, and the authors are honest that someone, or something, has to do the ignoring.

Before going further, it is worth being precise about what this paper is. MS360°, as the framework is called, is a perspective — a conceptual model authored by neurologists at the university hospitals of Dresden and Düsseldorf, set out as how hybrid MS care ought to be organised. PubMed indexes it as a review. It enrols no patients, reports no prospective outcomes, and describes no implementation. To read it as evidence would be a category error. To read it as a carefully argued blueprint is fair, and repays the time.

The signal, not the calendar

The structural problem MS360° names is real and familiar. Multiple sclerosis does not progress on a quarterly schedule, but its follow-up usually does: a visit every three or four months, a neurological examination, an EDSS score, and the tacit assumption that whatever changed in between either did not matter or will surface on its own. A patient whose gait has slowed or whose cognition has slipped in week six waits for the calendar to catch up.

The framework's answer is to let a measured signal, rather than the diary, decide when the patient is seen. It assembles components that already exist — on-site examination, imaging and laboratory work; remote monitoring through wearables and validated smartphone apps; telemedicine; patient-reported outcomes — and connects them so that the next on-site visit becomes a response to something measured. The triggers are specific: a clinically meaningful change in gait speed, defined as more than twenty percent deviation from the patient's own baseline; a drop in cognitive performance beyond the minimal clinically important difference for the test used; a patient-reported score for fatigue, depression or functional decline crossing a validated cut-off. Data moves in both directions, and the appointment stops being a fixture.

Where a threshold is not yet a trigger

Those thresholds are the place to slow down. A gait-speed change past twenty percent, or a cognitive drop past a test's minimal clinically important difference, sounds operational. But the paper does not establish what fraction of such alerts would turn out to be genuine progression rather than a bad night's sleep, an unrelated infection, or sensor noise. That fraction is the entire difference between a threshold and a validated trigger. Set the cut-offs too tight and tier two buries tier three under false alarms — the alert fatigue that has defeated continuous monitoring in other specialties. Set them too loose and the system misses the early progression it was built to catch. Nothing in the paper locates these particular cut-offs on that trade-off, because no prospective data was collected to do so.

The barriers the authors list are, if anything, more telling than the thresholds — and the candour with which they list them is to their credit. Reimbursement still rewards the on-site visit and pays for remote monitoring only patchily. Evidence, they note especially from the United States, suggests telemonitoring can raise clinician workload through extra data review and administrative load rather than lower it. Patient participation depends on access to devices, broadband and comfort with digital tools across ages and incomes. And proprietary platforms do not interoperate, producing data silos and duplicate documentation that erode the very efficiency the model promises. None of these is a footnote. Each is a precondition. A signal-driven pathway that doubles a neurologist's screen time, or that only the digitally fluent can use, is not the pathway on the page.

On artificial intelligence and digital twins, raised as future layers, the paper shows the restraint one wants to see. The authors state plainly that current AI systems cannot reliably produce accurate, guideline-based answers to specific neurological questions and may return erroneous or outdated recommendations, and that any such layer would need robust validation against clinical standards, alongside compliance with the EU AI Act and the Medical Device Regulation, before it is added. That sentence carries more weight than the architecture diagram.

What to do with it

For European neurology services the diagnosis is correct: episodic, calendar-bound follow-up captures a chronic, fluctuating disease only in snapshots, and a model that reserves scarce specialist attention for validated alerts is worth designing and worth testing. This paper has done the first and not the second. Its real value is as a research agenda with concrete, falsifiable thresholds that someone can now go and measure — in a prospective cohort, against progression outcomes that reach the patient, with the alert burden and the equity gap counted rather than assumed. Until that work is done, MS360° is a good map of a road no one has yet driven.

Source: Voigt I, Masanneck L, Pawlitzki M, Inojosa H, Meuth SG, Ziemssen T. MS360°: a conceptual digital-first, data-driven hybrid care framework for personalised multiple sclerosis management. npj Digital Medicine 2026;9:229. A perspective paper: it proposes a framework and carries no prospective outcome data, so its thresholds and workflow remain hypotheses to be validated rather than findings.