Quality
Image adequacy, reviewer agreement and the need for repeat acquisition help assess whether trained operators can produce interpretable studies.
H-TRIAGE™ and INSIGHT
Test the model. Understand its limits.
Internal validation and workforce experience inform the next questions. Independent and prospective evaluation remain distinct steps.
Internal validation
Testing whether H-TRIAGE™ identifies cardiovascular risk appropriately.
0.946AUC
94.6%Sensitivity
93.1%Specificity
95.3%Negative predictive value
View the study population and methods
| Study context | Reported scope |
|---|---|
| Sample | 3,799 patients across six facilities |
| Held-out site | Dodowa / Shai Osudoku |
| Held-out sample | 677 patients |
| Reference | Clinician adjudication; request the study methods for endpoint and adjudication definitions. |
| Period | The public study summary does not specify enrollment dates. |
| Status | Internal validation; independent external evaluation remains necessary. |
Source: Internal validation results reported by HeartLINK+, summarized in September 2026. Held-out-site evaluation within this study is not independent external validation.
The public summary does not identify whether these performance values refer to the full study population or the 677-patient held-out site. Request the analysis details before applying the figures to a particular cohort. These findings are study-specific. Performance may vary with prevalence, population and clinical setting. AUC measures discrimination; it is not a percentage of correct diagnoses.
Workforce and diagnostic quality
Building local capacity for cardiac imaging.
INSIGHT examines competency-based training, AI-assisted acquisition, specialist interpretation and operator learning.
Explore quality, efficiency and operator learning
Efficiency
Acquisition time is considered alongside scan type, operator experience and image quality, so speed is interpreted in the context of a usable study.
Learning
Changes in competency and study quality over successive scans help describe learning. Training experience alone does not establish improved patient outcomes.
Request the INSIGHT methods to examine training, quality scoring, acquisition time and operator learning in context.
AI supports the pathway; clinicians remain responsible. Clinical responsibility
Connect the next step
Look closely at the methods.
Discuss study design, analysis populations and the next independent evaluation.