TwinRehab connects patient and clinician through home exercises, sensors and a monitoring dashboard. Here, in silico, we validate that the platform transmits and displays every value faithfully - tested at scale on synthetic patients, the digital twins of the people it will one day serve.
A synthetic patient is a digital twin: a generated profile that mirrors the clinical reality of someone recovering from - or preparing for - hip or knee surgery, without being any real person. It lets us stress-test TwinRehab thoroughly and safely, long before and far beyond what a small real cohort allows.
Same data shape · same clinical realism · no real identity
Level 1 has one job: prove that data captured at home reaches the clinician exactly as sent, and that anomalies are surfaced - never interpreted.
Synthetic twins validate technical & analytical performance. Real patients validate clinical performance and usability.
At Level 1 TwinRehab transmits and displays - it does not interpret. Every dashboard value equals the value received.
Built to EU MDR 2017/745, MDCG 2020-1, IEC 62304, ISO 14971, IEC 62366 - with HL7 FHIR payloads.
Twins span phases (prehab & post-op), joints (hip & knee), sex, age and severity - plus the edge cases real cohorts rarely show.
Each requirement maps to a test case with an expected output. Pass or fail is explicit, automated, repeatable.
Level 1 (transmit & display) is validated before Level 2 (AI decision support). Twins never replace clinical evidence.
TwinRehab is a single platform with two role-based interfaces. The clinician prescribes a plan from the dashboard; the patient app turns it into guided home sessions and sends the results back. Each link of this loop is validated independently, with explicit pass/fail criteria traced to a software requirement.
Receives the plan, guides the session, collects and sends data - also offline / poor network.
The clinician's CarePlan reaches the right patient, intact and versioned.
Transport, authentication, acceptance of session payloads at cohort scale.
Every payload validated against the FHIR profile: fields, types, units, ranges.
Sent = stored = shown. Field-by-field comparison on every record.
No loss, no truncation; order and timestamps preserved, duplicates handled.
Every field rendered in the right place; flagged values shown as flagged.
Patients see only themselves; clinicians only their patients. Anomalies flagged, never interpreted.
Every validation run produces these artefacts automatically. They feed the technical–analytical performance section of the Clinical Evaluation Report (MDCG 2020-1) and the IEC 62304 software file.
The loop runs both ways. Press “Prescribe plan” to send a clinician's CarePlan down to the patient app - or “Transmit session” to send a home session up to the dashboard. Verify that what arrives equals what was sent, inject an anomaly to see it flagged without interpretation, or run the whole cohort at scale.
New to the terms? Open the glossary ↗
Clinical partner, investor or fellow IRHIS beneficiary? Request a guided walkthrough of the platform and the in silico validation, or ask about collaboration on the musculoskeletal DemoScale Lab.
EUC Inovação Portugal, Unipessoal Lda · info@twinrehab.care
