ABSTRACT: The opioid epidemic continues to plague the United States. Unreliable and slow data systems pose a continuing challenge to the public health response. There is a paucity of real-time data that can be used to detect or forecast increases in opioid overdoses and coordinate timely community resource mobilization efforts. Household surveys underestimate use and use disorder rates, mortality data has a time lag that makes evaluating policy impact difficult, and new illicit drugs are identified too slowly. Creating valid, sensitive, real-time data systems is thus a critical priority. Kentucky has suffered greatly during the opioid epidemic. In response, the Rapid Actionable Data for Opioid Response in Kentucky (RADOR-KY; 1-R01 DA057605-01) system enhances surveillance capacity with machine learning models designed to forecast future trends. The system currently uses data from 11 sources and integrates them for forecasting county-level risk of opioid overdose. The proposed administrative supplement of RADOR-KY will evaluate the added value of real-time social media data to improve opioid overdose surveillance and forecasting. Recent studies have shown that social media can signal opioid trends. Mentions of opioids on Reddit and language markers of distress on Twitter/X correlate with regional opioid-related overdose deaths. It is unclear if these signals are captured by current RADOR-KY data streams or if they provide independent information that could be useful for improving performance of the system. Evaluating whether social media data can improve the RADOR-KY prediction models is thus a major public health priority for Kentucky, and to the extent it can be shown effective, could be applied nationwide. In this supplement, we propose to work with Stanford University to (1) analyze the utility of counting direct social media mentions of opioids for forecasting overdoses across Reddit and Twitter/X, (2) analyze the utility of indirect population sentiment signals for forecasting overdoses, and (3) analyze the value of considering additional social media context around opioid use (e.g., addiction vs. recovery). This work will measure the potential value of social media data for opioid surveillance and estimate its incremental value to RADOR-KY.

Project Narrative RADOR-KY (Rapid Actionable Data for Opioid Response in Kentucky) is a statewide population-based surveillance system with a key feature of forecasting opioid-related overdose mortality and morbidity, so that public health resources can be deployed timely to the areas and populations that need them most to save lives. The system currently integrates 11 sources of data, but these do not include social media. This supplement proposes to investigate the value of two sources of (in principle, real-time) social media data (Reddit, Twitter/X) to improve the performance of forecasting algorithms, potentially providing a model for real-time opioid overdose epidemic surveillance.