Randomised trial of AI wearable monitoring after heart-failure discharge finds no fall in rehospitalisation (LINK-HF2)
In a randomised trial of 171 veterans discharged after a heart-failure admission at five VA medical centres, alerts from a personalised AI model reading a wearable patch, routed to clinicians with a response algorithm, were feasible but did not reduce heart-failure rehospitalisation (16.3% against 18.7%, hazard ratio 0.90, 95% CI 0.43 to 1.84) or a composite of rehospitalisation, emergency visits and death.
- Model
- Personalised similarity-based anomaly model on wearable multisensor data, not named in the abstract
- Field
- Medicine
- Date
- 2026-10-09
- Human collaborators
- Konstantinos Sideris, Josef Stehlik
What was found
LINK-HF2 was a prospective, randomised, open-label, blinded-endpoint study. All participants wore a continuous multisensor patch for up to 90 days; in the intervention arm a similarity-based model learned each patient's baseline cardiorespiratory and activity signals and notified clinicians of deviations, and clinicians responded within a median of one day. Median adherence to the patch was 72.4% of follow-up days. The intervention arm had fewer AI-detected episodes of physiological instability (hazard ratio 0.76, P = 0.025), but neither heart-failure rehospitalisation nor the composite of all-cause unplanned rehospitalisation, emergency visits and death moved (hazard ratio 1.03). Interviews named workflow integration and uncertainty about how to respond as the main barriers. It follows the 2020 LINK-HF study, in which the same approach detected precursors of heart-failure rehospitalisation with 76% to 88% sensitivity.
Novelty check
Europe PMC, LINK-HF2 and the trial title, to 2026-10-20: this publication and its single-arm predecessor, LINK-HF (Circulation: Heart Failure, 2020, NCT03037710). ClinicalTrials.gov lists NCT04502563 as completed with 176 enrolled.
Caveats and known objections
A negative result on the outcome that matters; the one positive finding is a surrogate the AI itself defines, fewer AI-detected instability episodes. The publication frames the trial as a feasibility study with heart-failure rehospitalisation as the main exploratory endpoint, but the ClinicalTrials.gov record gives the 90-day heart-failure hospitalisation rate between arms as the primary outcome, so the framing changed between registration and report. With 171 patients the trial was not powered for clinical events, and the confidence interval runs from roughly halving to nearly doubling the risk. Five VA medical centres, a veteran population, open-label allocation, sponsored by the VA Office of Research and Development per its registration. The author list includes Stephan Wegerich, listed in the 2020 LINK-HF paper at PhysIQ, an analytics company; affiliations for this paper were not available to check, so any developer role in it is not established here. Autonomy is ai-assisted: the model flagged deviations and clinicians decided what to do.
Trial registration
NCT04502563
That a study was registered, with the outcomes it declared and the date it declared them, so a published endpoint can be checked against the one planned. It does not certify the results, their analysis, or any clinical benefit; registry records are submitted by the sponsor and are not peer reviewed.
Nobody outside the lab has checked this yet.
Reading the primary source closely enough to say whether it supports the claim counts as a check, and you are credited on the entry.
Or on GitHub: submit a check challenge the grade send a correction or send a pull request
Flag this for triage
Signals order the review queue and nothing else. They are never published, and they never move a grade: that takes a citation.
Entry history (1 event)
- AddedEntered the registry graded Peer reviewed and AI-assisted.
Entries are never deleted. A grade that does not hold up is downgraded on the record, with the reason beside it.
Graded peer reviewed for verification and ai-assisted for autonomy. What these mean.
Cite this entry
whataifound.org. (2026). Randomised trial of AI wearable monitoring after heart-failure discharge finds no fall in rehospitalisation (LINK-HF2). whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2026-10-09-link-hf2-wearable-monitoring-rctBibTeX
@misc{whataifound-usdepartmentofveteransaffairs-2026-rct,
title = {Randomised trial of AI wearable monitoring after heart-failure discharge finds no fall in rehospitalisation (LINK-HF2)},
author = {{whataifound.org}},
year = {2026},
howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
note = {Result by US Department of Veterans Affairs / University of Utah. Verification: Peer reviewed. Autonomy: AI-assisted.},
url = {https://whataifound.org/finding/2026-10-09-link-hf2-wearable-monitoring-rct}
}