Social media and search engine companies, such as Google, use the wealth of data at their disposal to create algorithms to predict suicide risk, and provide interventions based on company policies and outputs from these algorithms. Neither the details of these algorithms nor their accuracy in risk prediction are shared with the public, thus research is needed to verify whether algorithms applied to search engine data can accurately predict suicide risk. To address this gap, the Searchlight Study recruited 485 adults with lived experience of suicidal thoughts and behavior to participate in a gold standard interview of their past year suicidal thoughts and behavior, complete prospective surveys of suicidal thoughts and behavior every two weeks for one year and donate their Google Search and YouTube data for this entire period.
In this presentation, members of the Searchlight Study team will share important lessons learned from conducting research entirely online, recruiting participants, and acquiring Google Search data from said participants. They will present the main findings of the study, including development of ‘features’ (i.e., categorization of searches) from participants’ Search data and detection of heightened suicide risk utilizing the developed features. In addition, we will present the associations between warning signs for suicide attempts and short-term changes in suicidal ideation from our prospective surveys.