Why clinical AI needs data from the time between visits

01The episodic-data problem

A clinical model is only as good as the signal it is given, and healthcare's signal is fundamentally episodic. A blood pressure at an annual physical. An ECG during a symptomatic episode that may never recur on cue. A sleep study for a single night in an unfamiliar bed. Between those points, the record is empty.

But the events that matter rarely schedule themselves for the appointment. The deterioration, the arrhythmia, the fall, the fertile window, the recovery plateau — they happen in the weeks of ordinary life that no instrument is watching. AI trained on episodes inherits the blind spots of episodes.

The patient between visits is where the signal lives — and where the instruments aren't.

02Continuous signal, from apparel

Textile sensing narrows that gap by making the garment the instrument. Identified SKIIN™ implementations have been developed to capture combinations of ECG, heart rate and variability, respiration, temperature, and movement from what a patient already wears; the exact signals, intended use, evidence, and regulatory status depend on the product and configuration. And in tested constructions the signal can hold up — textile ECG comparable to gel electrodes in one study (r² = 0.93 after 30 washes), pediatric heart rate at 3.6–3.8% NRMSE in twenty children, with a body of ambulatory-ECG literature behind the approach.

That approach is being tested clinically rather than only in the lab. An ongoing registered trial (ClinicalTrials.gov NCT05983484), run with Southlake Regional Health Centre, is comparing continuous textile monitoring against conventional Holter recording — the premise being that a multi-day continuous record can surface arrhythmic events that a 24-to-48-hour Holter window can miss. That is a hypothesis under evaluation, not a settled result.

The result the approach aims at is a stream rather than a snapshot — continuous physiology captured across the ordinary weeks episodic care never sees. The objective is to reduce the friction associated with episodic or separately attached measurement systems while preserving the signal quality required for the intended use.

03From monitoring to intervention

Across separate textile-sensing and textile-actuation programs, MyantX has explored both measurement and response. In one study, closed-loop neuromuscular stimulation from a garment regulated grasp force in a participant with quadriplegia to under 15% steady-state error. Functional electrical stimulation garments for stroke and spinal-cord-injury recovery were developed with rehabilitation clinicians, with a qualitative study of nineteen patients and clinicians shaping their design. In fertility care, continuous temperature and heart rate from textile sensors predicted ovulation in a cohort undergoing IUI — a non-invasive alternative to repeated blood draws.

In that direction, monitoring and intervention could become two sides of one loop rather than separate products — each combination requiring its own architecture, validation, and regulatory pathway.

04The decisions continuous data could support

The value of a continuous stream is not more charts — it is better decisions. Four in particular become answerable with data from between visits, provided the signal is validated for the purpose.

Detection — does a meaningful change appear between scheduled encounters, rather than only at the next appointment?

Baseline — how does an individual differ from their own normal pattern, rather than from a population average?

Escalation — when is a change significant enough that a clinician or service should intervene?

Response — can a connected product adapt safely within a defined, validated protocol?

None of these is automatic. Each depends on signal quality, validation for the intended use, and governance. But continuous physiology is the input they require, and episodic care cannot supply it.

05The direction

The forward path is a governed foundation where clinicians and developers could reach the body safely and by permission — aimed at continuous, non-invasive care rather than a single device. Because the interface is fabric, it could meet patients in their own lives without asking them to accept anything more than the textiles they already wear. How far that reaches — and at what scale — is what the work is still to prove.