Machine Learning and Artificial Intelligence
Theo Pezel, MD
Head of the Cardiovascular Imaging department
Lariboisiere University Hospital, APHP, Paris, France
Paris, Ile-de-France, France
Theo Pezel, MD
Head of the Cardiovascular Imaging department
Lariboisiere University Hospital, APHP, Paris, France
Paris, Ile-de-France, France
Philippe Garot, MD
Cardiologist
Institut Cardiovasculaire Paris Sud (ICPS), Cardiovascular Magnetic Resonance Laboratory, Hôpital Privé Jacques CARTIER, Massy, France
Thomas Hovasse, MD
Cardiologist
Institut Cardiovasculaire Paris Sud (ICPS), Cardiovascular Magnetic Resonance Laboratory, Hôpital Privé Jacques CARTIER, Massy, France
Solenn Toupin, PhD
Clinical scientist
Siemens Healthcare France, Scientific partnerships, Saint-Denis
Bordeaux, Aquitaine, France
Kenza Hamzi
Research Scientist
Université de Paris, Service de Cardiologie, Hôpital Lariboisière – APHP, Inserm UMRS 942, 75010, Paris, France., France
Thierry Lefevre, MD, PhD
Cardiologist
Hôpital Privé Jacques Cartier, Institut Cardiovasculaire Paris Sud, Cardiovascular Magnetic Resonance Laboratory, Ramsay Santé, 91300, Massy, France, France
Bernard Chevalier, MD, PhD
Cardiologist
Hôpital Privé Jacques Cartier, Institut Cardiovasculaire Paris Sud, Cardiovascular Magnetic Resonance Laboratory, Ramsay Santé, 91300, Massy, France, France
Thierry Unterseeh, MD
Cardiologist
Institut Cardiovasculaire Paris Sud (ICPS), Cardiovascular Magnetic Resonance Laboratory, Hôpital Privé Jacques CARTIER, Massy, France
Francesca Sanguineti, MD
Cardiologist
Institut Cardiovasculaire Paris Sud (ICPS), Cardiovascular Magnetic Resonance Laboratory, Hôpital Privé Jacques CARTIER, Massy, France
Stéphane Champagne, MD
Cardiologist
Institut Cardiovasculaire Paris Sud (ICPS), Cardiovascular Magnetic Resonance Laboratory, Hôpital Privé Jacques CARTIER, Massy, France
Hakim Benamer, MD
Cardiologist
Hôpital Privé Jacques Cartier, Institut Cardiovasculaire Paris Sud, Cardiovascular Magnetic Resonance Laboratory, Ramsay Santé, 91300, Massy, France, France
Antoinette Neylon, MD
Cardiologist
Hôpital Privé Jacques Cartier, Institut Cardiovasculaire Paris Sud, Cardiovascular Magnetic Resonance Laboratory, Ramsay Santé, 91300, Massy, France, France
Tania Ah-Sing, MD
Radiologist
Service de Radiologie, Hôpital Lariboisière – APHP, Paris, France., France
Lounis Hamzi, MD
Radiologist
Service de Radiologie, Hôpital Lariboisière – APHP, Paris, France
Trecy Goncalvez, MD
Cardiologist
Université de Paris, Service de Cardiologie, Hôpital Lariboisière – APHP, Inserm UMRS 942, 75010, Paris, France., France
Jean Guillaume Dillinger, MD, PhD
Cardiologist
Université de Paris Cité, Service de Cardiologie, Hôpital Lariboisière – APHP, France
Valérie Bousson, MD, PhD
Head of the Department
Service de Radiologie, Hôpital Lariboisière – APHP, Paris, France
Jerome Garot, PhD
Head
ICPS - Massy
Massy, Ile-de-France, France
In patients with suspected or known coronary artery disease (CAD), traditional prognostic risk assessment is based upon a limited selection of clinical and imaging findings. Machine learning (ML) methods can take into account a greater number and complexity of variables. We aimed to investigate the feasibility and accuracy of ML-score using simultaneously stress CMR, coronary computed tomography angiography (CCTA). and clinical data to predict the occurrence of cardiovascular events in patients with suspected or known CAD, and compared its performance to existing scores.
Methods:
Between 2008 and 2020, consecutive symptomatic patients without known CAD referred for CCTA were screened in Institut Cardiovasculaire Paris Sud (Massy, France). Patients with obstructive CAD (at least one ≥50% stenosis on coronary CTA) were further referred for stress CMR and followed for the occurrence of major adverse cardiovascular events (MACE), defined as cardiovascular death or nonfatal myocardial infarction. Twenty-three clinical, 11 stress CMR and 11 CCTA parameters were evaluated. Machine learning involved automated feature selection and model building by random survival forest. The external validation cohort of the ML score was performed in another center (N= 274 patients, University Lariboisiere Hospital, APHP, Paris).
Results:
Of 2,038 consecutive patients (46.5% men; mean age 69.8 ± 12.2 years), 281 (13.8%) patients experienced a MACE after a median follow-up of 6.7 years (interquartile range: 5.9-9.1). Our ML score exhibited a higher area-under-the-curve compared with stress CMR data alone, CCTA data alone, and traditional Cox model for prediction of 10-year MACE(ML: 0.88 vs. CMR data alone: 0.79, CCTA data alone: 0.72; traditional Cox model: 0.81, all p< 0.001). The ML score assessed in the derivation cohort (AUC: 0.88, F1-score 0.80) exhibited also a good area-under-the-curve in the external cohort for prediction of 10-year MACE (AUC: 0.86, F1-score 0.80).
Conclusion:
The ML score including clinical, stress CMR and CCTA data exhibited a higher prognostic value to predict 10-year MACE compared with all traditional clinical data, CMR data or CCTA data alone. These findings reinforce the importance of multimodality cardiovascular imaging integrating both CMR and CCTA data to improve our cardiovascular risk stratification tools.