Fig 1.
Histogram showing the distribution of the number of predicate “parent” devices from 2000 to 2020 for 462 medical devices cleared through the 510(k) regulatory pathway in 2020.
The y axis is a log scale. Note that an outlier of 62 predicates (histogram count of 1) in 2015 has been not indicated, in order to enhance readability of the histograms shown.
Fig 2.
Box and whisker plots for the average equivalence to predicate device across all approved devices each year, from 2003 onwards.
The degree of similarity index (y-axis) indicates the level of equivalence between a device and its predicate (1: Identical, 0: Totally dissimilar).
Fig 3.
Scatter plot of device complaints (since 2006, as the TPLC database retains data for only 15 years) versus recalls for the product codes associated with the 2,721 devices from 2020.
Recalls were split into Class I (most severe), II, and III (least severe), and complaints were divided between device issues and consumer complaints. Complaints were projected onto a log scale because a few product codes had very large numbers of complaints. The size of each point corresponds to the number of different devices in that product code.
Fig 4.
Bar plot comparing seven Class 1 recalls to 87 predicate devices, all under the LZG ProCode.
The results are based on text similarity computed by the cosine similarity algorithm.
Fig 5.
Sankey plot showing the genealogy of medical devices cleared through the 510(k) approach for a De Novo product DEN000001 (which was initially cleared as K974562 and reclassified to a De Novo product with special controls).
The thickness of the connecting edge corresponds to the similarity match between a predicate and a successive device (the exceptions being the incoming/outgoing connections for the product K033779, for which summary statements are not available).
Table 1.
Similarity score for devices descendant from the predicate ancestral tree for the ProCode: MRN.