Use of the energy waveform electrocardiogram to detect subclinical left ventricular dysfunction in patients with type 2 diabetes mellitus
Authors' conclusion
Machine learning based modelling using additional ewECG extracted features are superior to NT-proBNP and ARIC HF in SBHF screening among patients with T2DM, providing an alternative HF screening strategy for asymptomatic patients and potentially act as a guidance tool to determine those who required echocardiogram to confirm diagnosis.
Study context
- Population
- Patients with T2D in two clinical trials of SBHF (DD, reduced GLS [≤18%] or LVH), with 12-lead electrocardiogram (ECG) and echocardiography178 training, 97 validation
- Comparison
- ewECG machine-learning model vs NT-proBNP and Atherosclerosis Risk in Communities (ARIC) HF risk score for SBHF screening
- Outcome
- Accuracy (AUC) in identifying SBHF
Methodological detailssupervised machine learning, training and external validation datasets from two clinical trials
- Supervised machine learning
- Training and external validation datasets from two clinical trials
Performance
As reported in the abstract
SBHF identification
| Test or model | AUC |
|---|---|
| ewECG | 0.81 (95% CI 0.787-0.794) |
| NT-proBNP | 0.56 (95% CI 0.44-0.68) |
| ARIC HF | 0.67 (95%CI 0.56-0.79) |
AUC
ewECG0.81 (95% CI 0.787-0.794)
NT-proBNP0.56 (95% CI 0.44-0.68)
ARIC HF0.67 (95%CI 0.56-0.79)
- ewECG vs NT-proBNP: P < 0.001 — significantly better with ewECG
- ewECG vs ARIC HF: P = 0.002 — significantly better with ewECG
Further findings
- ewECG screening for SBHF components
- 0.74 (95% CI 0.73-0.74) · AUC for DD · AUC for reduced GLS 0.76 (95% CI 0.73-0.74) · AUC for LVH 0.90 (95% CI 0.88-0.89)
SBHF prevalence
- SBHF in training dataset (median 72 years, 41% female)
- 128 (72%)
- SBHF in validation dataset (median 70 years, 43% female)
- 64 (66%)
Original publication
Use of the energy waveform electrocardiogram to detect subclinical left ventricular dysfunction in patients with type 2 diabetes mellitus. Cardiovasc Diabetol. Published 6 March 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →