Heart failure & cardiomyopathies
Natural Language Processing to Adjudicate Heart Failure Hospitalizations in Global Clinical Trials
Authors' conclusion
AI-based adjudication of clinical outcomes has the potential to improve the efficiency of global clinical trials while preserving accuracy and interpretability.
Study context
- Population
- Potential HF hospitalizations in the DELIVER trial (dapagliflozin vs placebo in HF with mildly reduced or preserved EF)
- Participants
- 6,063 DELIVER patients
- Comparison
- Artificial intelligence (AI)-based HF adjudication vs clinical events committee adjudication
- Outcome
- Agreement of AI-based adjudication with clinical events committee
Methodological detailsmodel development in 3 international outcomes trials, tested in DELIVER trial
- Model development in 3 international outcomes trials
- Tested in DELIVER trial
Performance
As reported in the abstract
Agreement with clinical events committee
AI-based adjudication
83%
Further findings
| Outcome | HR (95% CI) | |
|---|---|---|
| Effect of dapagliflozin on HF hospitalization, by adjudication methodDapagliflozinvs PlaceboAI-based adjudication · nearly identical | 0.76 (0.66–0.88) | |
| Effect of dapagliflozin on HF hospitalization, by adjudication methodDapagliflozinvs PlaceboClinical events committee adjudication · nearly identical | 0.77 (0.67–0.89) | |
| 0.512Hazard ratio (log scale) |
Hybrid strategy: human review of events AI deemed uncertain
- Events deemed uncertain
- 16%
- Agreement with clinical events committee
- 91%
- Reduction in adjudication workload
- 84%
Interpretability
- AI model extracted HF symptoms, signs and treatments in tabular format and quoted documenting sentences
Original publication
Natural Language Processing to Adjudicate Heart Failure Hospitalizations in Global Clinical Trials. Circ Heart Fail. Published 16 November 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →