Medical Devices: Evidence and Research (Aug 2017)

Can machine learning complement traditional medical device surveillance? A case-study of dual-chamber implantable cardioverter–defibrillators

  • Ross JS,
  • Bates J,
  • Parzynski CS,
  • Akar JG,
  • Curtis JP,
  • Desai NR,
  • Freeman JV,
  • Gamble GM,
  • Kuntz R,
  • Li SX,
  • Marinac-Dabic D,
  • Masoudi FA,
  • Normand SLT,
  • Ranasinghe I,
  • Shaw RE,
  • Krumholz HM

Journal volume & issue
Vol. Volume 10
pp. 165 – 188

Abstract

Read online

Joseph S Ross,1–4 Jonathan Bates,4 Craig S Parzynski,4 Joseph G Akar,4,5 Jeptha P Curtis,4,5 Nihar R Desai,4,5 James V Freeman,4,5 Ginger M Gamble,4 Richard Kuntz,6 Shu-Xia Li,4 Danica Marinac-Dabic,7 Frederick A Masoudi,8 Sharon-Lise T Normand,9,10 Isuru Ranasinghe,11 Richard E Shaw,12 Harlan M Krumholz2–5 1Section of General Medicine, Department of Medicine, 2Robert Wood Johnson Foundation Clinical Scholars Program, Yale School of Medicine, 3Department of Health Policy and Management, Yale School of Public Health, 4Center for Outcomes Research and Evaluation, Yale–New Haven Hospital, 5Section of Cardiovascular Medicine, Department of Medicine, Yale School of Medicine, New Haven, CT, 6Medtronic Inc, Minneapolis, MN, 7Division of Epidemiology, Center for Devices and Radiological Health, US Food and Drug Administration, Silver Spring, MD, 8Division of Cardiology, Department of Medicine, University of Colorado, Aurora, CO, 9Department of Health Care Policy, Harvard Medical School, 10Department of Biostatistics, Harvard TH Chan School of Public Health, Boston, MA, USA; 11Discipline of Medicine, University of Adelaide, Adelaide, SA, Australia; 12Department of Clinical Informatics, California Pacific Medical Center, San Francisco, CA, USA Background: Machine learning methods may complement traditional analytic methods for medical device surveillance.Methods and results: Using data from the National Cardiovascular Data Registry for implantable cardioverter–defibrillators (ICDs) linked to Medicare administrative claims for longitudinal follow-up, we applied three statistical approaches to safety-signal detection for commonly used dual-chamber ICDs that used two propensity score (PS) models: one specified by subject-matter experts (PS-SME), and the other one by machine learning-based selection (PS-ML). The first approach used PS-SME and cumulative incidence (time-to-event), the second approach used PS-SME and cumulative risk (Data Extraction and Longitudinal Trend Analysis [DELTA]), and the third approach used PS-ML and cumulative risk (embedded feature selection). Safety-signal surveillance was conducted for eleven dual-chamber ICD models implanted at least 2,000 times over 3 years. Between 2006 and 2010, there were 71,948 Medicare fee-for-service beneficiaries who received dual-chamber ICDs. Cumulative device-specific unadjusted 3-year event rates varied for three surveyed safety signals: death from any cause, 12.8%–20.9%; nonfatal ICD-related adverse events, 19.3%–26.3%; and death from any cause or nonfatal ICD-related adverse event, 27.1%–37.6%. Agreement among safety signals detected/not detected between the time-to-event and DELTA approaches was 90.9% (360 of 396, k=0.068), between the time-to-event and embedded feature-selection approaches was 91.7% (363 of 396, k=–0.028), and between the DELTA and embedded feature selection approaches was 88.1% (349 of 396, k=–0.042).Conclusion: Three statistical approaches, including one machine learning method, identified important safety signals, but without exact agreement. Ensemble methods may be needed to detect all safety signals for further evaluation during medical device surveillance. Keywords: implanted cardioverter–defibrillator, methodology, surveillance

Keywords