Natural Language Processing to Adjudicate Heart Failure Hospitalizations in Global Clinical Trials.

Autor: Marti-Castellote PM; Division of Cardiovascular Medicine, Brigham and Women's Hospital, Boston, MA; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA., Reeder C; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA; Data Sciences Platform, Broad Institute of Harvard and the Massachusetts Institute of Technology, Cambridge, MA., Claggett BL; Division of Cardiovascular Medicine, Brigham and Women's Hospital, Boston, MA., Singh P; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA; Data Sciences Platform, Broad Institute of Harvard and the Massachusetts Institute of Technology, Cambridge, MA., Lau ES; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA; Division of Cardiology, Massachusetts General Hospital, Boston, MA., Khurshid S; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA; Demoulas Center for Cardiac Arrhythmias, Massachusetts General Hospital, Boston, MA., Batra P; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA; Data Sciences Platform, Broad Institute of Harvard and the Massachusetts Institute of Technology, Cambridge, MA., Lubitz SA; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA; Demoulas Center for Cardiac Arrhythmias, Massachusetts General Hospital, Boston, MA., Maddah M; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA; Data Sciences Platform, Broad Institute of Harvard and the Massachusetts Institute of Technology, Cambridge, MA., Vardeny O; Minneapolis VA Hospital, University of Minnesota, Minneapolis, MN., Lewis EF; Stanford University School of Medicine, Palo Alto, CA., Pfeffer MA; Division of Cardiovascular Medicine, Brigham and Women's Hospital, Boston, MA., Jhund PS; British Heart Foundation Cardiovascular Research Centre, University of Glasgow, United Kingdom., Desai AS; Division of Cardiovascular Medicine, Brigham and Women's Hospital, Boston, MA., McMurray JJV; British Heart Foundation Cardiovascular Research Centre, University of Glasgow, United Kingdom., Ellinor PT; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA; Demoulas Center for Cardiac Arrhythmias, Massachusetts General Hospital, Boston, MA., Ho JE; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA; CardioVascular Institute and Division of Cardiology, Department of Medicine, Beth Israel Deaconess Medical Center, Boston, MA., Solomon SD; Division of Cardiovascular Medicine, Brigham and Women's Hospital, Boston, MA., Cunningham JW; Division of Cardiovascular Medicine, Brigham and Women's Hospital, Boston, MA; Cardiovascular Disease Initiative, Broad Institute of Harvard University and the Massachusetts Institute of Technology, Cambridge, MA.
Jazyk: angličtina
Zdroj: Circulation. Heart failure [Circ Heart Fail] 2024 Nov 16. Date of Electronic Publication: 2024 Nov 16.
DOI: 10.1161/CIRCHEARTFAILURE.124.012514
Abstrakt: Background: Medical record review by a physician clinical events committee is the gold standard for identifying cardiovascular outcomes in clinical trials, but is labor-intensive and poorly reproducible. Automated outcome adjudication by artificial intelligence (AI) could enable larger and less expensive clinical trials, but has not been validated in global studies. Methods: We developed a novel model for automated AI-based heart failure adjudication ("HF-NLP") using hospitalizations from three international clinical outcomes trials. This model was tested on potential heart failure hospitalizations from the DELIVER trial, a cardiovascular outcomes trial comparing dapagliflozin with placebo in 6063 patients with heart failure with mildly reduced or preserved ejection fraction. AI-based adjudications were compared with adjudications from a clinical events committee that followed FDA-based criteria. Results: AI-based adjudication agreed with the clinical events committee in 83% of events. A strategy of human review for events that the AI model deemed uncertain (16%) would have achieved 91% agreement with the clinical events committee while reducing adjudication workload by 84%. The estimated effect of dapagliflozin on heart failure hospitalization was nearly identical with AI-based adjudication (hazard ratio 0.76 [95% CI 0.66-0.88]) compared to clinical events committee adjudication (hazard ratio 0.77 [95% CI 0.67-0.89]). The AI model extracted symptoms, signs, and treatments of heart failure from each medical record in tabular format and quoted sentences documenting them. Conclusions: AI-based adjudication of clinical outcomes has the potential to improve the efficiency of global clinical trials while preserving accuracy and interpretability.
Databáze: MEDLINE