Artificial Intelligence (AI) pilot project for surveillance of suicide-related behaviours using social media.

1. Requirements

The following are suggested work phases and time frames, to serve as a general guideline. This project is anticipated to begin in January 2018 and be completed on or before June 30, 2018. The following are suggested work phases, time frames and deliverables:

Phase 1. Project Initiation

a) The contractor will work with the client to define ‘suicide-related behavior’ terms broadly under following terms: ideation (i.e., thoughts), behaviors (i.e., suicide attempts, self-harm, suicide) and communications (i.e., suicidal threats, plans).

b) The contractor will build a classifier in order to implement these definitions of ‘suicide-related behaviour’ and use that classifier to conduct market research on the general population of Canada in order to identify patterns associated with users who discuss suicide-related behavior, based on online data.

Phase 2. Baseline Report & Living Survey

Baseline Report: This will be an initial report from the contractor, 5 months after receiving required inputs from the Client, and the Client will have the opportunity to probe the findings and produce two (2) more iterative reports over the course of the following sixty (60) days, at no extra cost. An iterative report is a “drill down” into specific areas of the report that the client may find interesting. The contractor understands that the project might reveal some unexpected findings or that the original question posed was vague and requires more “drill down” in order to glean more useful information.

Living Survey: The contractor will use initial findings developed in task (a) to establish a ‘living survey’ pilot for a period of three (3) months in order to establish its value in obtaining a deeper understanding of associated risk and protective factors.

Phase 3. Final Report

The contractor will prepare a final report summarizing findings to date. Based on efforts in Phases 1 and 2, the contractor would include a mock-up of a potential monthly reporting mechanism, with the goal of summarizing the magnitude of discussion around suicide by age group and gender (e.g. 14-18-year-old males), changes in patterns and available risk and protective factors. This report will be used to assess what, if any, potential utility there would be for ongoing surveillance at a national level. Based on this assessment, PHAC will determine if future work would be useful for ongoing suicide surveillance.