American Federation for Medical Research

Back to 2019 Abstracts


Reduction of Heart Failure Admissions by 40% Using KenCor Artificial Intelligence Software: A Pilot Study of 30 Patients
Maya J. Dhond1, Aedan Enriquez3, Yardana Gill1, Milind Dhond1, 2, Cyrus Mancherje1, Hossein Dehghani1, Terra Hadsall1, Jeff Breneisen1. 1Cardiology, Northbay Healthcare, Fairfield, California, United States, 2UC Davis, Davis, California, United States, 3UC Berkeley, Berkeley, California, United States

Purpose of Study Congestive heart failure (CHF) causes significant morbidity and mortality with costs exceeding $30 billion annually in the United States. We used KenCor Artificial Intelligence Software (KAIS) on outpatients with CHF to evaluate reductions in hospital admission rates.
Methods Used We enrolled 30 patients with known CHF (systolic and diastolic) from our CHF clinic in KAIS. KAIS software was downloaded onto patients' smartphones with on-site training in its use. Patients were provided with a Wifi-enabled blood pressure (BP) cuff, weigh scale, and pulse oximeter. Patients completed the daily 5 minute program with transmission of data to the CHF nurse. KAIS stratified the patients into low (green), medium (yellow), or high (red) risk for admission. Patients with red alerts were contacted by the CHF nurse. Patients at high risk for more than 1 day were seen in the clinic for medication adjustment and intravenous diuresis. We collected data up to 6 months pre and post KAIS enrollment.
Summary of Results 29/30 (97%) were compliant with KAIS. The patients (16 women) were aged 44-88 years (average: 63.8 years). Patients had systolic CHF (72.4%) and diastolic CHF (37.6%). The average ejection fraction (EF) was 41.5%. The average pre and post enrollment systolic/diastolic BPs were 106.9/63.6 and 113.4/66.5 (p=NS). The average serum creatinine pre and post KAIS enrollment were 1.43 and 1.28 (p=NS). The average clinical brain natriuretic peptides (BNP) pre and post enrollment were 4399 and 2696 (p=NS). The pre enrollment CHF admission rate was 0.108/month (203 months total). The post enrollment CHF admission rate was 0.077/month (155 months total). Paired t-test compared pre and post enrollment CHF admission rates, demonstrating a 40% reduction in the admission rate per month per patient (p=0.03).
Conclusions In a small pilot study of outpatients with CHF, enrollment in KAIS reduced CHF hospital admissions by 40%. This technology may result in significant reduction in CHF-associated costs to the patient, hospital, and insurance companies. It may also allow more efficacious use of CHF nurses.


Back to 2019 Abstracts