Session (CTI)

A challenge to generating real-world evidence for medical devices and diagnostics (MDD) is the ability to identify unique MDD products in real-world data. Device specific information, which is not readily available in structured claims data, is typically manually abstracted from unstructured data such as medical charts and physician notes, before being placed into a database for research. This process is lengthy, costly, and can lead to inaccuracies in data capture. Artificial intelligence (AI) methods, such as natural language processing (NLP) and machine learning (ML), have been shown to be a potential solution to this challenge. These methods can lead to fit-for-purpose curated databases that can be used for generating real-world evidence that may meet the needs of regulators, payers, and other healthcare decision makers. AI methods have the potential to identify MDD products within unstructured data and create structured databases that provide information not only on the device of interest, but also on comparators, patient characteristics, and outcomes. The panel will discuss the challenges and opportunities associated with these techniques, including current practice, reliability and validity, and chances of acceptance by regulators and payers. Panel members will discuss use cases from their own experiences. Each speaker will present their perspective for 15 minutes with an additional 15 minutes reserved for audience questions facilitated by the moderator towards the end of the session. Individuals interested in the challenges in extracting unique MDD product and patient data from real world data would benefit from attending.