EAGER: Collaborative Research: CRUFS: A Unified Framework for Social Media Analysis of Adverse Drug Events

EAGER：协作研究：CRUFS：药物不良事件社交媒体分析的统一框架

• 批准号: 1553109
• 负责人: Ahmed Abbasi
• 金额: $ 11万
• 依托单位: University of Virginia Main Campus
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-08-15 至 2018-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1553109&HistoricalAwards=false
• 关键词: EAGER; Collaborative; Research; CRUFS; Unified

An adverse drug reaction (ADR) is any undesired response to a medication. ADRs have been linked with significant morbidity and mortality, accounting for as much as 5% of hospital admissions. The problem stems from the fact that the ADR profile of a given drug is rarely complete at the time of official approval. The typically limited pre-approval evaluation often results in the possibility that when the drug is finally approved for use in the general population (with significant diversity in race, gender, age, lifestyle), some previously unidentified ADRs are often observed. For psychotropic medications, the problem becomes compounded by the fact that most people with psychiatric diseases tend to have other health issues, with the individual taking multiple medications (both psychotropic and non-psychotropic) at the same time, with often unknown interactions between them. Given the huge quantities of data on drugs, drug interactions, and diseases, and the possibility offered by social media sources in obtaining more information about particular drugs and their side effects, the problem of post-marketing drug surveillance could be turned into a computational problem. This work will have relevance to government agencies charged with drug approval and disease monitoring (e.g., the Food and Drug Administration (FDA), Centers for Disease Control (CDC), public health agencies), pharmaceutical companies, and the general public. The proposed work will have impact beyond drug surveillance as the methods can be applied to other scenarios such as financial markets, national security, or other healthcare problems. Graduate and undergraduates students will be involved in the project, thereby gaining experience in doing research. Journal papers and conference presentations will be used to disseminate research results. The project takes a new approach to the problem of adverse drug event surveillance by relying heavily on the collective intelligence of the web community, with significant emphasis on social media and online sources. This calls for more serious attention to the ubiquity, veracity and diversity of data from these sources. Thus the general goal is to develop the CRUFS (credibility, recency, uniqueness, frequency and salience) framework as a uniform and innovative foundation for assessing different data channels in social media analysis of adverse drug events. The project will study methods to extract reliable signals from unreliable, noisy, redundant, and potentially deceptive online data, a core challenge in social media analytics. The project also proposes novel methods for ADR signal detection and signal fusion based on causality networks. The results will change the current passive surveillance that relies on voluntary reports, by making the public an integral part of a proactive drug surveillance system.

药物不良反应（ADR）是对药物的任何不良反应。ADR与显著的发病率和死亡率相关，占住院人数的5%。这个问题源于这样一个事实，即给定药物的ADR概况在正式批准时很少完整。通常有限的批准前评价通常导致当药物最终批准用于一般人群（种族、性别、年龄、生活方式具有显著差异）时，经常观察到一些先前未识别的ADR。对于精神药物来说，由于大多数患有精神疾病的人往往会出现其他健康问题，而且个人同时服用多种药物（精神药物和非精神药物），而且它们之间往往存在未知的相互作用，因此问题变得更加复杂。考虑到关于药物、药物相互作用和疾病的大量数据，以及社交媒体来源提供的获取有关特定药物及其副作用的更多信息的可能性，上市后药物监测的问题可能会变成一个计算问题。这项工作将与负责药物批准和疾病监测的政府机构（例如，食品和药物管理局（FDA）、疾病控制中心（CDC）、公共卫生机构）、制药公司和公众。拟议的工作将产生超越药物监测的影响，因为这些方法可以应用于其他场景，如金融市场，国家安全或其他医疗保健问题。研究生和本科生将参与该项目，从而获得做研究的经验。期刊论文和会议报告将用于传播研究成果。该项目对药品不良事件监测问题采取了一种新的方法，主要依靠网络社区的集体智慧，并特别强调社交媒体和在线来源。这就要求更加认真地关注这些来源的数据的普遍性、准确性和多样性。因此，总体目标是开发CRUFS（可信度、新近度、独特性、频率和显著性）框架，作为评估药物不良事件社交媒体分析中不同数据渠道的统一和创新基础。该项目将研究从不可靠、嘈杂、冗余和潜在欺骗性的在线数据中提取可靠信号的方法，这是社交媒体分析的核心挑战。该项目还提出了基于因果网络的ADR信号检测和信号融合的新方法。研究结果将改变目前依赖自愿报告的被动监测，使公众成为主动药物监测系统的一个组成部分。

项目成果

A Deep Learning Architecture for Psychometric Natural Language Processing

• DOI: 10.1145/3365211
• 发表时间: 2020-02
• 期刊: ACM Transactions on Information Systems (TOIS)
• 作者: Faizan Ahmad;A. Abbasi;Jingjing Li;David G. Dobolyi;Richard G. Netemeyer;G. Clifford;Hsinchun Chen