Opportunistic Screening for Chagas Disease Using an Artificial Intelligence–Enabled ECG: Prospective Evaluation of Feasibility and Diagnostic Accuracy
Authors' conclusion
In this community-based study, an AI-ECG-EPI model integrated into a public telecardiology network enabled feasible opportunistic screening for ChD in primary care. It showed acceptable diagnostic accuracy, with higher discrimination than epidemiological questions and AI ECG-only models, and identified many previously undiagnosed patients in endemic regions, supporting its potential role in scalable screening strategies.
Study context
- Population
- Adults undergoing routine tele-ECGs in hyperendemic and endemic regions of Brazil, October 2023 to September 2024completed serology testing, 75,779 eligible ECGs, AI-ECG-EPI model available for 59,154 individuals, 7804 invited for serology · AI-ECG-EPI positive, completed testing 2,517 · AI-ECG-EPI negative, completed testing 992
- Participants
- 3,509 adults
- Exposure
- AI-ECG-EPI model against serology (the reference standard); compared with both an epidemiological questions model and an artificial intelligence (AI) ECG-only model
Methodological detailsembedded in the Telehealth Network of Minas Gerais, Brazil, 107 municipalities in hyperendemic and endemic regions, index test-dependent sampling, Weighted analyses (inverse probability)
- Embedded in the Telehealth Network of Minas Gerais, Brazil
- 107 municipalities in hyperendemic and endemic regions
- Index test-dependent sampling
- Weighted analyses (inverse probability)
Main finding
As reported in the abstract
Screened positive by AI-ECG-EPI model
Screened positive
6812
11.5%
Observational data: associations, not effects.
Further findings
AUROC
| Outcome | AI-ECG-EPI model | Epidemiological questions model | AI ECG-only model |
|---|---|---|---|
| Discrimination (area under the receiver operating characteristics curve) | 77.6% | 70.0% | 64.5% |
- Chagas disease (ChD) confirmed by serology — Weighted prevalence of ChD
- 12.2%
- Diagnostic accuracy of the AI-ECG-EPI model
- 34.3% · Sensitivity · Specificity 91.6% · Harmonic mean of precision and recall score 0.52 · Diagnostic odds ratio 5.69
- Discrimination (area under the receiver operating characteristics curve) — AI-ECG-EPI model showed higher discrimination than both comparators
- P <0.001
- ChD confirmed by serology: AI-ECG-EPI positive (tested) 36.8%
- ChD confirmed by serology: AI-ECG-EPI negative (tested) 9.3%
Precision recall curves
- Higher precision across most recall levels
Risk reclassification (exploratory)
- Net reclassification improvement
- 0.503
- Exploratory analysis
Performance by subgroup
- Performance was higher in women and in individuals with Chagas cardiomyopathy
Original publication
Opportunistic Screening for Chagas Disease Using an Artificial Intelligence–Enabled ECG: Prospective Evaluation of Feasibility and Diagnostic Accuracy. Circulation. Published 8 September 2026. doi:10.
Summarised from the published abstract; every number is checked against it