Device Libraries

The Center is building medical device information libraries, which include information like catalogue numbers and manufacturer names. Device libraries can be used in long-term device surveillance within CRNs, with the flexibility to focus on individual device, device categories, or device characteristics. Device libraries can also enhance the capacity of the FDA’s GUDID for research and surveillance.

• Featured device library:

  • International Consortium of Orthopedic Registries (ICOR) device library 

The ICOR device library is a global, standardized classification system of hip and knee implantable devices, and includes all their clinical attributes and characteristics to advance the implementation of unique device identifiers and FDA post-market surveillance.

 

Data Linkages

One of the main MDEpiNet methodological advancements has been to conduct linkages between registry data and routinely available data sources (e.g. claims and administrative data). The Center has been successfully developing and refining anonymous linkage algorithms to augmenting research capacities of CRNs by bringing together registries, claims data, and electronic health records. Data linkage with indirect identifiers is reliable with high sensitivity and accuracy. It is the most cost-effective way to obtain long-term outcomes and has positive implications for long-term device surveillance.

• Featured projects:

  • Indirect linkage validation showed high sensitivity and accuracy

  • VISION project: Linkage between registries and Medicare claims data enhances long-term surveillance of vascular device

  • Linkage between registry and statewide discharge data enables research among younger patient population undergoing aneurysm repair

  • Three-way linkage between registry, Medicare, and clinical data

  • Linkage between registry and New York City Clinical Research Network data

 

Natural Language Processing

The Center is incorporating natural language processing (NLP) into the building blocks of CRN. NLP is a valuable method that can be used to parse unstructured text data and extract information from unstructured text data, including medical notes, radiology reports, and device adverse event reports. For processing large amount of text data, NLP is efficient and labor saving. The Center is currently also advancing NLP methods that will help with data collections to enhance sustainability of CRNs.

• Featured projects:

  • Assessing adverse events related to hysteroscopic sterilization device removal using NLP

  • Extracting information from prostate cancer biopsy reports, magnetic resonance imaging and partial gland ablation operative reports

 

Active Surveillance

The Center is advancing active surveillance methodologies to support long-term device surveillance based on Coordinated Registry Network. The Center is developing a flexible tool to provide users with timely and comprehensive evaluations of medical device safety signals.

• Featured projects:

  • Outlier methodologies for CRNs or registries

  • In progress: An integrated tool to implement DELTA and outlier detection in CRNs

​