Suicide of Virginia Woolf could have been predicted using Artificial Intelligence
Researchers say a predictive algorithm would have flagged the potential for self-harm
What if an algorithm could be used to predict a person’s risk of suicide? Imagine being able to apply the process to text messages or social media posts to get a real sense of a person’s mental health to create the potential to intervene before tragedy strikes? Researchers at St. Joseph’s Healthcare Hamilton have developed a machine-learning algorithm that can predict individual suicidal behavior.
A test case involving historical writings of acclaimed author Virginia Woolf has been published this month in the journal “PLOS ONE.” Woolf died by suicide in 1941. Her diaries and letters were carefully analyzed using a Naïve-Bayes machine-learning algorithm to identify written patterns associated with her death.
The research team led by Dr. Flavio Kapczinski of St. Joseph’s compared 46 written entries from the two months before Woolf’s death with 54 pieces of her writing randomly selected from other periods of her life. The text classification algorithm was able to identify with high accuracy the potential of suicide two months prior to her death by suicide.
“Sometimes we need to dig into examples of the past with new technology to find answers to our current clinical dilemmas,” says Kapczinski.
St. Joseph’s Healthcare is among many research institutions looking into Artificial Intelligence as a means to deliver better health care. This innovative work is an example of the potential offered by text analysis. The use of an algorithm to predict a possible suicide would allow clinical experts to respond with appropriate interventions with the intent of trying to prevent untimely deaths.
Read more about the study here.