Decision support environment for medical product safety surveillance

Abstract

We have developed a Decision Support Environment (DSE) for medical experts at the US Food and Drug Administration (FDA). The DSE contains two integrated systems: The Event-based Text-mining of Health Electronic Records (ETHER) and the Pattern-based and Advanced Network Analyzer for Clinical Evaluation and Assessment (PANACEA). These systems assist medical experts in reviewing reports submitted to the Vaccine Adverse Event Reporting System (VAERS) and the FDA Adverse Event Reporting System (FAERS). In this manuscript, we describe the DSE architecture and key functionalities, and examine its potential contributions to the signal management process by focusing on four use cases: the identification of missing cases from a case series, the identification of duplicate case reports, retrieving cases for a case series analysis, and community detection for signal identification and characterization.

Evaluating automated approaches to anaphylaxis case classification using unstructured data from the FDA Sentinel System

“…This original project was a collaboration among the FDA, the Mini‐Sentinel Operations Center, and selected Mini‐Sentinel Academic and Data Partners. Five Mini‐Sentinel Data Partners participated in this project: (1) HealthCore, Inc.; (2) Humana; (3) three member health plans within the Kaiser Permanente Center for Effectiveness and Safety Research; (4) two member health plans within the Health Care Systems Research Network; and (5) Vanderbilt University (Tennessee Medicaid data).As illustrated in Figure , the text files were merged and then processed using two platforms: (1) the FDA's Event‐based Text‐mining of Health Electronic Records (ETHER) Natural Language Processing (NLP) system that extracts key information from clinical texts; and (2) the National Library of Medicine (NLM) MetaMap tool that converts medical terms to the codes of selected terminologies, such as the Medical Dictionary for Regulatory Activities (MedDRA) . ETHER accomplishes feature extraction based on a set of rules.…”

“…When a clinical term cannot be associated with any trigger word, it will be extracted as a symptom. Similarly, ETHER will retrieve clinical terms negated with the appropriate keywords, such as “without” and “denied.” The full information for the ETHER algorithm can be found in our previously published work …”

“…As MetaMap returns terms from any level of the MedDRA hierarchy, we had to further map them to the MedDRA Preferred Term (PT) Level. Each PT was linked to an ETHER diagnostic feature (primary diagnosis [PDx], secondary diagnosis [SDx], and symptoms [SYM]) . This supported the generation of a list of PTs (with their feature type) for each section per chart as well as a list of PTs (with their feature type) for each chart.…”

“…In addition to ETHER and MetaMap, the similarity‐based approach was supported by the FDA's Pattern‐based and Advanced Network Analyzer for Clinical Evaluation and Assessment (PANACEA) that allows the identification of cases associated with selected conditions by performing the corresponding similarity calculations …”

“…• Improvements in performance of the algorithms will depend on incorporating additional information that human experts use to distinguish anaphylaxis from other conditions. Figure 1, the text files were merged and then processed using two platforms: (1) the FDA's Event-based Text-mining of Health Electronic Records (ETHER) Natural Language Processing (NLP) system that extracts key information from clinical texts 12 ; and (2) the National Library of Medicine (NLM) MetaMap tool that converts medical terms to the codes of selected terminologies, such as the Medical Dictionary for Regulatory Activities (MedDRA). 13 ETHER accomplishes feature extraction based on a set of rules.…”