Tracking the opioid epidemic with social media: an early warning system

项目摘要 阿片类药物在美国国内的死亡人数继续居高不下，甚至还在加速 在大流行期间。遏制和阻止疫情将需要多管齐下的努力。 沿着许多健康、社会、甚至文化层面。一个重要的目标是提高我们的能力 为了监测疫情-阿片类药物使用热点的检测速度和精确的地理位置- 本地化。社交媒体数据代表着一个耐人寻味的机会，因为它提供直接报告 来自人群的帖子--这可能有助于检测和定位阿片类药物活动。 然而，还没有对以下方面的效用(和缺陷)进行仔细的系统评估 广泛使用社交媒体。在这份提案中，我们概述了一项将Reddit和Twitter结合起来的计划 评估跟踪阿片类药物相关死亡率的能力。我们将专注于提取提及的 芬太尼在社交媒体上的应用，因为它是大多数死亡的罪魁祸首。我们还将重点关注 可能预示致命使用的修饰物，特别是单独或成群使用阿片类药物， 纳洛酮的功能性可利用性，以及合并症药物的使用。我们将探索极限 通过在多个地理空间尺度上聚合社交媒体数据的效用：国家、地区、 州和县(或聚集人口稀少的县的县集合)。我们还将 探索我们可以看到相关信号的时间尺度：年度、半年度、季度 而且是每月一次。因此，我们有两个目标： 1.评估社交媒体共同提供阿片类药物使用信号的能力(通过 提到芬太尼和相关的合成阿片类药物)与死亡率或 在多种地理空间和时间分辨率下的死亡率。 2.为了评估社交媒体在提供关于集体使用与单独使用的信息方面的效用， 纳洛酮和多药联用改变阿片类药物的强度或紧迫度的功能可用性- 使用信号。 我们的工作目标是生成一个全面的初步评估的效用 社交媒体，因为它存在于现实世界和多个规模，作为阿片类药物的一个组成部分 使用监控系统。来自民众的直接报告可能会提供有用的“早期预警” 当与其他来源相结合时，将为政策制定者和公共卫生提供 官员们拥有一套强大的工具来监测这场公共卫生危机。

Project Summary The opioid epidemic has continued to take a high death toll within the US, even accelerating during the pandemic. Stemming and stopping the pandemic will require a multi-prong effort along many health, social, even cultural dimensions. One important goal is to improve our ability to monitor the epidemic–both the speed of detection of opioid-use hotspots and precise geo- localization. Social media data represents an intriguing opportunity as it provides direct-report posts from the population–which may be useful for detecting and localizing opioid activity. However there has not been a careful systematic assessment of the utility (and pitfalls) for broad use of social media. In this proposal, we outline a plan to combine Reddit and Twitter to assess the ability to track opioid-associated mortality. We will focus on extracting mentions of fentanyl in social media, as it is responsible for the majority of deaths. We will also focus on modifiers that may be predictive of deadly use, particularly the use of opioids alone or in groups, the functional availability of naloxone, and the use of comorbid drugs. We will explore the limits of utility of social media data by aggregating it at multiple geo-spatial scales: national, regional, state and county (or county-sets that aggregate sparsely populated counties). We will also explore the time scales at which we can see relevant signals: annual, semi-annual, quarterly and monthly. Thus, we have two aims: 1. To assess the ability of social media together to provide signals of opioid use (through mentions of fentanyl and related synthetic opioids) that are correlated with mortality or proxies of mortality at multiple geo-spatial and temporal resolutions. 2. To assess the utility of social media to provide information about group vs. solo use, functional availability of naloxone and polypharmacy to modify the strength or urgency of opioid- use signals. The goal of our work is to generate a comprehensive preliminary assessment of the utility of social media, as it exists in the real world and at multiple scales, as a component of the opioid use surveillance system. Direct reports from the population may provide a useful “early warning” system that when combined with other sources, will provide policy makers and public health officials with a powerful set of tools for monitoring this public health crisis.