New AI (Artificial Intelligence) model predicts heart failure hospitalizations

  • December 15, 2021

The Selene HF trial included more than 900 patients with a median age of 69 years old and a median ejection fraction of 30%.  Eighty-one percent (81%) of participants were men.  All patients had either an implantable cardioverter-defibrillator (ICD) or cardiac resynchronization therapy defibrillator (CRT-D) manufactured by Bioronik.  Each participant was enrolled from one of 34 sites in Italy and Spain.

The research team’s algorithm focused on seven (7) key factors:

  1. diurnal (day) heart rates;
  2. nocturnal (nighttime) heart rates;
  3. ventricular extrasystoles;
  4. atrial tachyarrhythmia burden;
  5. heart rate variability;
  6. physical activity; and
  7. thoracic impedance.

It was designed to alert clinicians when they needed to step in and provide care, potentially avoiding any further issues.  “A HF (heart failure ) alert would benefit both patients and clinicians,” Lead Author Antonio D’Onofrio, MD, a Cardiologist at Monaldi Hospital in Italy, said in a statement.  “Detecting worsening HF early and proactively stratifying patients at risk may help improve quality of care and avoid re-hospitalizations.  This also may alleviate overloaded clinics and help efficiently allocate resources.”

Overall, the AI model anticipated two-thirds of all first post-implant HF hospitalizations.  The median prediction time was 42 days.  Also, the AI model achieved a specificity that ranged from 76% to 87% and a sensitivity of 65.5%.  Its false alert rate (0.69 per patient-year) and unexplained alert rate (0.63 per patient-year), meanwhile, were “remarkably lower than in other published algorithms.”

“The predicting algorithm showed promising sensitivity and a remarkably low false alert rate,” D’Onofrio et al. wrote.  “First post-implant HF hospitalizations, intravenous interventions, subsequent and terminal HF hospitalizations could be predicted with similar accuracy.  Randomized trials are needed to assess whether the application of the algorithm may be associated with improved outcomes.”

Read the full study in EP Eurospace, a journal from the European Society of Cardiology.

Editor’s note:  Biotronik did fund this analysis.  Some authors reported a prior relationship with Biotronik, but D’Onofrio did not.

REFERENCE:  Innovate Healthcare/Cardiovascular Business (Strategies in economics, practice & technology); 23 AUG 2021; Michael Walter


Leave a Comment