Machine learning for pattern detection in cochlear implant FDA adverse event reports

Abstract

<b>Importance:</b> Medical device performance and safety databases can be analyzed for patterns and novel opportunities for improving patient safety and/or device design. <b>Objective:</b> The objective of this analysis was to use supervised machine learning to explore patterns in reported adverse events involving cochlear implants. <b>Design:</b> Adverse event reports for the top three CI manufacturers were acquired for the analysis. Four supervised machine learning algorithms were used to predict which adverse event description pattern corresponded with a specific cochlear implant manufacturer and adverse event type. <b>Setting:</b> U.S. government public database. <b>Participants:</b> Adult and pediatric cochlear patients. <b>Exposure:</b> Surgical placement of a cochlear implant. <b>Main Outcome Measure:</b> Classification prediction accuracy (% correct predictions). <b>Results:</b> Most adverse events involved patient injury (<i>n</i> = 16,736), followed by device malfunction (<i>n</i> = 10,760), and death (<i>n</i> = 16). The random forest, linear SVC, naïve Bayes and logistic algorithms were able to predict the specific CI manufacturer based on the adverse event narrative with an average accuracy of 74.8%, 86.0%, 88.5% and 88.6%, respectively. <b>Conclusions & relevance:</b> Using supervised machine learning algorithms, our classification models were able to predict the CI manufacturer and event type with high accuracy based on patterns in adverse event text descriptions.

Ähnliche Arbeiten