Incremental value of machine learning for risk prediction in tetralogy of Fallot
Authors' conclusion
Robust prediction of 5-year MACE in rTOF was achieved using either ML or a multidisciplinary team of ACHD experts.
Study context
- Population
- Adults with repaired tetralogy of Fallot (rTOF) from an institutional databasepatient reviews by 5 observers
- Participants
- 25 adults
- Comparison
- Machine learning (ML) model vs expert clinical judgement for 5-year MACE prediction
- Outcome
- MACE (composite of mortality, resuscitated sudden death, sustained VT and heart failure (HF)) prediction
- Follow-up
- 5 years
Methodological detailsclinicians blinded to outcome, single institutional database
- Clinicians blinded to outcome
- Single institutional database
Performance
As reported in the abstract
Discrimination for 5-year MACE
| Test or model | AUC |
|---|---|
| ML | 0.85 (95% CI 0.58 to 0.96) |
| Experts | 0.92 (0.72 to 0.98) |
AUC
ML0.85 (95% CI 0.58 to 0.96)
Experts0.92 (0.72 to 0.98)
- ML vs Experts: P = 0.315 — similar
Further findings
| Outcome | Values | P value |
|---|---|---|
| Discrimination by expert experience (≥20 vs <20 years) | ≥20 years AUC 0.98 (95% CI 0.86 to 0.99) · <20 years AUC 0.80 (0.56 to 0.93) | P = 0.027 |
| Combined clinical+ML in clinicians with <20 years of experience<20 years clinical+ML vs ≥20 years · approached ≥20 years | <20 years clinical+ML AUC 0.85 (95% CI 0.61 to 0.95) | P = 0.055 |
Original publication
Incremental value of machine learning for risk prediction in tetralogy of Fallot. Heart. Published 1 December 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →