Life Sciences Blog

Webinar Recording: Predicting Medical Device Recalls

by | Jan 17, 2021

Is predicting medical device recalls possible?

We have some compelling research that indicates-yes!

Did you know? We have been designing and curating an expansive database of medical device companies and products since 2016. Over time, we have been developing further analysis of our proprietary database and have formed a research hypothesis, methodology and have found some startling results revealing leading indicators for the potential of a recall.

www.reedtech.com/navigator

In our initial webinar on the subject in early 2020, we provided a summary of research into the topic of predicting medical device recalls and the agenda included:

  • An overview of a proprietary database of medical device companies and products
  • Predictive analytics methodology
  • Leading indicators of a medical device recall
  • Which devices are at the highest risk for recall in the coming year across all industries
  • How advance notice of a potential recall could help your business
  • Polling on current awareness (view results in recording)

Analyzing Medical Device Recalls–Update

The project has continued and with the available history of adverse event reports, product design and performance defects captured and, in many cases, reported patient problems, we posed a hypothetical question: Can these data points be modeled to help understand if recall events are predictable in nature?

We have taken a deep-dive into the trends leveraging our proprietary database to create a prediction model. The main source for this database is the FDA. The FDA provides a large variety of medical device data; however, the data is siloed and kept in different locations. Using natural language processing and the LexisNexis® company relationship database, we can match the FDA data together and create a single relational database. A machine-learning algorithm uses this database to train a classifier model for determining if a medical device is likely to have a recall in the next 365 days. Our latest testing results have shown that the model can correctly identify 13.18% of the medical devices that account for 75.6% of all medical device recalls.

Download the white paper  ‘Predicting Recalls Using Publicly Available Data’ to see how Reed Tech is using the expansive data properties and analytics within Navigator to reliably test and predict medical device recall probabilities.

Questions? We are here to help. Visit our Knowledge Center or Contact Us.