Description

A web-based application to help researchers explore participants' online information seeking behavior by observing their online search history.

Background

Each person has their own digital footprint like their search history on Google or their viewing preferences on TikTok. While individual searches, like “how to wash your hands” may reveal very little about you, over time these searches build up and with enough data, they can be examined to reveal patterns. 

One particular pattern that researchers have been interested in is suicidal ideation, and for us in particular, whether a person’s Google search history can reveal patterns about their suicidal ideation in the past. This has inspired us to start the POISE project, short for Personalized Online Information Seeking Explorer. 

Purpose

Individualized search data could reveal the key topics that are being searched on a day-to-day basis along with the time of day they are using the most. This data can then be observed parallelly with the timeline of their past behavior anomalies. The goal is to understand changes in an individual’s behavior using a continual stream of data and use that insight to improve their mental healthcare.

History

The first use case of POISE emerged from data collected as part of a recent study. Our team used Google search histories to observe patterns related to suicidal ideation. The participants consented to providing their online search data around the time of their suicidal attempts. The analysis revealed distinctive online information seeking behavior across individuals.

Benefits

POISE will be available to the study team and partners at the University of Washington and the Centre for Addiction and Mental Health (CAMH). However, our long term goal is to enable broader access to including individuals to use POISE to enable individuals to examine their own search data and to understand their own patterns. We also take data privacy and security seriously and will not require the users to upload their data to our servers in order to use POISE.

How

By using the search data from patients (that have consented), we’ll be developing ways to reveal patterns based on their search data, like showing the day of the week that they tend to Google the most or highlighting the top 25 words used in their day-to-day Googling.

Further reading

https://www.jmir.org/2021/5/e27918

Why:

• We seek to understand information regarding how Google searches relate to one's suicidal ideation in the past

What:

• Create a Shiny dashboard for a particular individual/cohort to show and understand the patterns
  
  • Should help clinicians explore the clinical value of these datasets

...

Resource Stack

...