Evidence built from the recovery journey itself.
Crely's evidence base is generated from longitudinal physiological data collected across the postoperative course — spanning retrospective clinical proof-of-concept, technology validation, translational work, and a prospective validation pathway.
Four layers, from data to prospective study.
The evidence is organised into four layers — each described below with exactly the detail it supports.
Clinical proof-of-concept
Model performance developed on retrospective, continuously-monitored post-CABG cohorts.
Technology validation
Sensing benchmarked against cleared reference devices and assessed for signal quality.
Translational
Physiological mechanism connecting preclinical work to the human-cohort models.
Prospective pathway
A validation programme with pre-specified endpoints, now under way.
Built on longitudinal data from ~450 patients across cardiac, abdominal, and organ-transplant surgery. All performance shown below is development-stage and retrospective — always paired with its cohort and design.
Performance, inside its study context.
Each application is shown with its cohort, the number of outcome events, warning-time ahead of the clinical reference, coverage of those events, and sensitivity / specificity — all on post-cardiac-surgery (CABG) development cohorts.
IRIS
Infection Risk Intelligence System160 post-CABG patients · 16 infection events · retrospective development
CLARIS
Critical Care Alert & Risk Intelligence System45 post-CABG patients · 41 deterioration events · 100% event coverage · retrospective development
MERIS
Mortality Event Risk Intelligence System364 post-CABG patients · 11 deaths · retrospective development
Does the device measure what it should?
Beyond the clinical models, the underlying sensing is assessed against established references and for signal quality.
Device benchmarking
Tissue-oximetry measurements assessed for substantial equivalence against FDA-cleared tissue oximeters (Medtronic INVOS 7100, Masimo O3).
Signal quality
Continuous multimodal signals are evaluated for quality and stability, so downstream models act on reliable inputs rather than artefact.
From mechanism to human study.
Crely's models rest on physiological mechanism, not correlation alone. Each signal is included because there is a reason a complication would change it — perfusion, inflammation, autonomic response — so a flagged change is explainable rather than a black box.
Preclinical work supports that a measurable physiological change can appear before a complication becomes clinically overt — the same mechanism the human-cohort models rely on. This translational thread links laboratory findings to the retrospective clinical results, and grounds the prospective studies now under way.
From retrospective signal to prospective proof.
Each capability is labelled by exactly where it stands today — retrospective development complete, and prospective validation now in progress.
Retrospective development
CompleteSignals established on continuously-monitored cohorts.
Prospective validation
In progressHospital studies with pre-specified endpoints, testing the signals prospectively.
The record, as it grows.
Manuscripts describing the development cohorts and model performance are in preparation. Each reference is labelled by stage — published, accepted, submitted, or in preparation — and full methods and results are shared under an evidence briefing.
Want the full methods and results?
We share study design, cohort detail, and complete performance — with all context intact — under an evidence briefing for clinical and partner conversations.
Request an evidence briefing