Using Search Engine Data for Detection and Early Intervention in Suicide Prevention

摘要。这是一份使用搜索引擎数据补充R 01 MH 123484补助金的请求， 针对NOT-OD-22-026的自杀预防检测和早期干预 与生物伦理问题有关的研究和能力建设工作补编。这一补充建议 专注于生物伦理学研究。父母奖将决定是否互联网搜索历史和 搜索行为的谷歌搜索引擎捐赠和前瞻性收集的人与不同的 自杀风险的程度将成功地确定自杀的近端风险。在之前的一项研究中， 最近的自杀企图捐赠了从谷歌外卖工具（GTO）下载的回顾性数据。 我们能够识别30-60天前预测自杀企图的行为和语言模式 事件发生了。目前资助的研究将要求1,000名有不同自杀风险的人捐款 回顾性数据，并继续捐赠这些数据1年。参与者完成回顾 每两周进行一次访谈和前瞻性调查，了解自杀行为和企图的发生情况。 我们是否能够证明我们在以前的试点项目中发现的相同结果， 可能会改变近端自杀风险检测的游戏规则。鉴于77%的 的美国人口1几乎完全使用谷歌搜索在线查找信息， 算法和随后的干预应该能够接触到处于危险之中的美国人，以防止这种严重的公众 健康结果。然而，如果我们成功了，有一些道德的，法律的，和社会的 仍然需要解决的影响。为了理解这些含义，我们将对50名 一系列焦点小组的研究参与者（25名以前没有自杀治疗经验，25名 有这样的经验）和20名干预者（临床医生和社区工作者）关于道德和公平的问题 将这种算法应用于预防自杀的干预措施。我们包括以下观点： 干预者在这项研究中，以确定消费者和干预者同意的道德，法律的， 社会影响以及可能存在意见分歧的地方。与伦理学家的磋商将指导 提出问题和解释结果。在数字健康框架的指导下，我们将 向参与者提供有关隐私问题的不同场景（选择共享，共享哪些数据）， 风险/利益问题（哪一代理人应有权获得司法协助并负责根据司法协助行事 建议），可访问性和可用性问题（多样性代表性和访问; 干预措施是可以接受的，但要特别注意道德和公平的资源分配）， 管理问题（数据应存储在何处以及如何存储）。与会者还将被要求考虑 应采用哪些潜在解决方案来解决这些问题。

ABSTRACT. This is a request to supplement grant award R01MH123484 Using Search Engine Data for Detection and Early Intervention in Suicide Prevention in response to NOT-OD-22-026 Administrative Supplement for Research and Capacity Building Efforts Related to Bioethical Issues. This supplement proposal focuses on Bioethics Research. The parent award will determine whether internet search histories and on search behavior on the Google Search Engine donated by and prospectively collected by people with varying degrees of suicide risk will be successful in determining proximal risk of suicide. In a previous study, people with a recent suicide attempt donated retrospective data downloaded from the Google Take Out tool (GTO). We were able to identify behavioral and linguistic patterns that predicted suicide attempts 30-60 days before the event occurred. The currently funded study will ask 1,000 people with varying risk for suicide to donate retrospective data and to continue to donate these data for 1 year. Participants complete a retrospective interview and prospective surveys every two weeks about the occurrence of suicidal behavior and attempts. Should we be able to demonstrate to scale the same results we found in the previous pilot project, the data from this current study could be game changing in the detection of proximal suicide risk. Given that 77 percent of the US population1 seek information online almost entirely using Google Search, any risk prediction algorithm and subsequent intervention should be able to reach at-risk Americans to prevent this serious public health outcome. However, should we be successful, there are a number of ethical, legal, and societal implications that still need to be addressed. To understand these implications, we will qualitatively interview 50 study participants in a series of focus groups (25 with no previous experience with treatment for suicide and 25 with that experience) and 20 interventionists (clinicians and community workers) about ethical and equitable application of such an algorithm to interventions to prevent suicide. We include the perspectives of interventionists in this study to identify where consumers and interventionists agree on ethical, legal, and societal implications and where there maybe divergence of opinion. Consultation with ethicists will guide the development of the questions and interpretation of results. Guided by the Digital Health Framework, we will present participants with different scenarios about privacy concerns (choice to share, what data to share), risk/benefit concerns (which agent should have access to the MLA and be responsible for acting on a MLA recommendation), accessibility and usability concerns (diversity representation and access; which interventions are acceptable with a specific eye toward moral and equitable resource allocation), and data management concerns (where and how the data should be stored). Participants will also be asked to consider what potential solutions should be used to address these concerns.