A Power Tracking Algorithm for Early Detection of Centrifugal Flow Pump Thrombosis

Area of Science:

• Cardiovascular Engineering
• Biomedical Data Science
• Medical Device Monitoring

Background:

• Logfiles from the HeartWare HVAD System offer crucial operational pump trend data for patient management.
• Pump thrombosis (PT) is a serious complication often indicated by increased logfile power preceding clinical symptoms.

Purpose of the Study:

• To develop and validate a Power Tracking algorithm for early detection of pump thrombosis (PT) using HVAD system logfiles.
• To assess the algorithm's sensitivity, timing of detection, and false alarm rate in clinical trial data.

Main Methods:

• A Power Tracking algorithm was developed to identify significant deviations in pump power indicative of PT.
• The algorithm was retrospectively applied to logfiles from the ENDURANCE, ENDURANCE Supplemental, and LATERAL clinical trials.
• Two cohorts were analyzed: suspected PT events (n=70) and control patients without adverse events (n=106).

Main Results:

• The Power Tracking algorithm detected PT cases on or before the recorded adverse event date with 85.7% sensitivity.
• Detection occurred an average of 3.9 days prior to clinical presentation.
• The algorithm generated approximately one false alarm per 6.85 patient-years of monitoring in non-event logfiles.

Conclusions:

• The Power Tracking algorithm demonstrates favorable performance for early PT detection in HVAD patients.
• Earlier detection facilitates timely medical management, potentially avoiding surgical intervention.
• This algorithm can enhance patient management by providing advance warning of potential pump thrombosis.