Evidence

We draw a hard line between three different kinds of claim: published clinical decision tools, our own models (not yet independently validated), and forward-looking design targets. Here's exactly where that line sits.

1. Published clinical decision tools we apply

The symptom-check assessment is instructed to apply these published, widely used clinical decision rules where relevant to a patient's presentation — they are the validated part of this product, not something we invented:

HEART score
Chest pain risk stratification
Backus et al., International Journal of Cardiology, 2013
Wells criteria
DVT / pulmonary embolism risk
Wells et al., The Lancet / NEJM, 1997–2000
Ottawa ankle/knee rules
Ankle and knee injury imaging decisions
Stiell et al., JAMA, 1992–1996
ABCD²
TIA stroke-risk stratification
Johnston et al., The Lancet, 2007
Cincinnati Prehospital Stroke Scale (FAST)
Rapid stroke symptom screening
Kothari et al., Academic Emergency Medicine, 1997
PECARN head injury rules
Pediatric head trauma — CT imaging decisions
Kuppermann et al., The Lancet, 2009

2. Our own models: shadow-mode only, not yet validated

JegsMediScan's own risk-scoring and admission-prediction models have never influenced a real patient's ESI level, routing, or care. They run only in shadow mode — evaluated against already-triaged real patients' outcomes, purely for internal review, with no path to a live recommendation until a clinical team signs off on a go/no-go review. Shadow evaluation can measure agreement with an ED's own actual triage decisions; it cannot, by itself, prove the model would have caught something a clinician missed. That requires a real clinical validation study, which hasn't been run.

3. Design targets are projections, not measured results

Numbers like reduced door-to-triage time or a lower left-without-being-seen rate, shown elsewhere on this site, are explicitly labeled design targets — goals the product is built toward, not outcomes measured at a live site. No hospital has deployed this product's hardware-sensor concept yet, so there is no live-site data to report.

For hospitals →Request a pilot →