Prevention & lipidsDiagnostic or prognostic study
Comparison of Artificial Intelligence-Driven Echocardiographic Assessment of Diastolic Function With the ASE Guidelines
- Comparison of the prognostic and diagnostic performance of a previously validated deep learning (DL) model with the 2016 ASE/EACVI and 2025 ASE guidelines
- exploratory
- 1 prognostic cohort (ARIC) and 2 hemodynamic validation cohorts
Authors' conclusion
In settings where not all guideline-recommended parameters are available, the DL model may improve prognostic and diagnostic performance compared with both guideline versions. Nevertheless, these exploratory findings require further validation to confirm that the observed advantages persist when all guideline-recommended parameters are systematically acquired.
Study context
- Population
- Participants of the ARIC (Atherosclerosis Risk In Communities) cohort, for prognosis, and patients in 2 hemodynamic validation cohorts from the United States and Japan, for diagnosing elevated left ventricular filling pressure; in all 3 cohorts, ≥1 guideline-recommended parameters were unavailable in a substantial proportion of patientsper cohort · ARIC cohort (prognostication) 5,450 · United States hemodynamic validation cohort 83 · Japan hemodynamic validation cohort 130
- Comparison
- DL model vs the 2016 ASE/EACVI and 2025 ASE guidelines for prognostication and for diagnosing elevated left ventricular filling pressure
Performance
As reported in the abstract
Prognostic performance (C-index), ARIC cohort
DL model
0.676
C-index
2016 ASE/EACVI guidelines
0.638
C-index
2025 ASE guidelines
0.602
C-index
DL model vs 2016 ASE/EACVI guidelines: P < 0.001 · DL model vs 2025 ASE guidelines: P < 0.001
Further findings
AUC
| Outcome | DL model | 2016 ASE/EACVI guidelines | 2025 ASE guidelines | P value |
|---|---|---|---|---|
| Diagnosis of elevated left ventricular filling pressure (AUC), United States cohort — DL model vs 2025 ASE guidelines · higher with the DL model | 0.879 | 0.812 | 0.822 | P = 0.041 |
| Diagnosis of elevated left ventricular filling pressure (AUC), United States cohort — DL model vs 2016 ASE/EACVI guidelines · similar AUC | P = 0.138 | |||
| Diagnosis of elevated left ventricular filling pressure (AUC), Japanese cohort — DL model vs 2016 ASE/EACVI guidelines · higher with the DL model; both P < 0.05 | 0.816 | 0.694 | 0.634 | P < 0.05 |
| Diagnosis of elevated left ventricular filling pressure (AUC), Japanese cohort — DL model vs 2025 ASE guidelines · higher with the DL model; both P < 0.05 | P < 0.05 |
Original publication
Comparison of Artificial Intelligence-Driven Echocardiographic Assessment of Diastolic Function With the ASE Guidelines. JACC Adv. doi:10.1016/j.jacadv.2026.103115
Summarised from the published abstract; every number is checked against it