Development of an IT tool to support post-market surveillance and expert panels in detecting sentinel signals relevant to serious incidents in high-risk medical devices: pilot on italian data

Abstract Introduction The EU Medical Device Regulation 2017/745, in force since May 26th 2021, defines new rules for medical device (MD) certification and post-market surveillance (PMS). For high-risk implantable MD, the process notified bodies are obliged to consult clinical Expert Panels (EP) that could decide proceed an extensive review of supporting data provided by manufacturer, also view sentinel signals, such as significantly increased rate reported serious incidents a specific MD groups. Purpose 1) To develop ICT tool automatically collect display aggregated way accessible curated regulatory information on alerts recalls capture possible trends be used both scientific analysis source EP. 2) conduct pilot feasibility study Italian data, characterized already same European Nomenclature (EMDN), organized multi-level hierarchical tree code define it will Europe. Methods Web scraping was retrieve 7622 safety notices (SN) from 2009 2021 Ministry Health website. EMDN missing 68% cases: it, best match searched within separate public list about 1.5M market containing code, using Natural Language Processing techniques pairwise entity resolution with Cosine similarity identify similar manufacturers MD. performance this approach tested 2440 SN which available gold standard. A mash up then performed integrate present final user through graphical interface. Results implemented method able correctly assign correct manufacturer each 99% cases. Moreover, at level 1 (22 categories available) assigned 2382 (97.62%), 2 (146 anatomical or functional groups 2366 (96.97%), 3 (multiple types 2304 (94.45%). developed interface (Figure allows querying database manufactures, devices, type (1a), selecting nomenclature fourth (1b). As result, relevant is shown, including over selected period link original website (2). Conclusions proposed cope uncompleteness publicly SN, thus allowing proper matching its very good performance. In way, grouping category/group/type rates Extension aggregate other nations result effective support PMS. Funding Acknowledgement Type funding sources: Public grant(s) – funding. Main source(s): Horizon 2020 - Project CORE-MD