I have been absent as I have been steadily working on finalizing my paper as well as preparing my defense. The paper has been submitted to my committee and I have an official defense date set.
Some interesting takeaways now that research is complete:
Cardiac medical devices account for nearly 20% of the total dataset. This result supports my original assumption given the prevalent nature of cardiovascular disease worldwide. Cardiovascular disease is the leading cause of death globally, affecting 1 in 4 adults.
Machine Learning allowed for a much more efficient way to analyze a dataset consisting of million of records. However, if the dataset is fraught with errors, we cannot expect true accuracy in the result. MAUDE dataset would need standardization in how data is collected, as well as WHAT is collected. There were many columns that were left empty which proved to be challenging for analysis.
Cardiac medical device migration had the highest impact on patients as far as frequency and risk of injury. More than 70% of those that experienced cardiac device migrations were classified as “injured”, and nearly 3% died. The results were similar in the non-cardiac device group as well. In conclusion, device migrations are harmful to patients, so more research would be interesting to understand location, facilities, physicians, etc. where this is occurring.
In order to derive insight from MAUDE, data analysis was required, which is something that needs improvement. The average American needs to have information about device risks at their fingertips in a user-friendly manner, so they are empowered to make informed decisions about their health.