Hernández-Cruz, Lace

Publication Search Results

Now showing 1 - 1 of 1

A predictive model to estimate FDA’S decision time for medical device development of 510(k)’s
(2017) Hernández-Cruz, Lace; Medina-Aviles, Lourdes A.; College of Engineering; Dávila, Saylisse; Resto, Pedro; Department of Industrial Engineering; Rodriguez, Daniel
This research aims to identify the relevant factors affecting FDA’s decision time for medical devices and to develop a predictive model for the decision time of 510(k) submissions. Development companies are concerned for the time it takes the FDA to clear a device. Data shows that, since 2006, 510(k)’s decision time has exceeded the goal of 90 days, averaging 150 in 2015. To understand the factors affecting the decision time, characteristics of cleared devices and the approval process are studied. Data was extracted from FDA’s database including 510(k) submissions from 2000 through 2015 for 8 medical disciplines. Traditional statistical analysis tools and ensemble machine learning methods were implemented as prospective predictors for the FDA’s decision time of 510(k)’s. By testing hypothesis about relevant factors, it is found that applicant experience, submission year, product code, submission subtype and third party review influence the decision time. In terms of predictability, we provide a comparison and discussion on the adequacy of the models fitted, for which random forest ensemble learning methods resulted in the best performer.