Siri and Alexa are part of our daily lives but now Irish scientists are using Artificial Intelligence to help predict national suicide rates in new study.
Irish scientists have discovered how trends for Google search terms can help to predict national suicide rates by using Artificial Intelligence.
In the first Irish study of its type, researchers from NUI Galway examined how certain internet search terms related to monthly suicide deaths reported by the Central Statistics office from the 11-year period from 2004 to 2015.
As the period covered the economic crash, the study also gauged how monthly records relating to unemployment over the time frame related to suicide death rates.
Artificial Intelligence-powered technologies like Siri and Alexa have become household names but now Irish researchers have used a specific AI system to help to crack the complex relationship between internet searches, national suicide deaths and unemployment rates.
Lead author, Joana Barros, PHD student at the Insight Centre for Data Analytics at NUI Galway, said they used an Artificial Intelligence computing system called neural networks which is able to learn in similar way to the human brain.
The scientists first trained the sophisticated system by feeding it Google search terms, past suicide rates and unemployment statistics from the 11 year time frame from 2004 to 2015 so it could learn the relationship between the data - before feeding it more unseen data to get the results.
Over 30 search terms were examined including colloquial terms like ‘top yourself” relating to suicide and ‘feeling down’, ‘got the blues’, and ‘baby blues’ relating to depression along with Irish translations.
It found that direct relations between a spike or drop in Google search trends and a similar rise or fall in suicide deaths “are not frequent”.
But the researchers found the combination of the search term “feeling down” and unemployment statistics was the most effective when it came to the prediction of national suicide occurrences.
“Although the volume of the search terms by themselves might not lead to a conclusive prediction of the suicide deaths, this AI model has been able to learn from past data and to draw conclusions a human might not be able to spot”, said Joana Barros.
She said the model could be an added benefit for public health officials as they can anticipate changes in the number of monthly suicide occurrences, indicating when more attention or caution should be applied.
“It has created a novel tool for improving current health policies in Ireland.
“The search terms that proved more successful in our evaluation were ‘depression’ and ‘feeling down’”, said Joana Barros.
The study, which has just been published in the International Journal of Environmental Research and Public Health, revealed that the total number of reported suicide cases over the 11-year period was 5938.
The CSO figures showed the highest number of deaths occurred in the years 2011, 2006, and 2012, with 554, 552, and 541 cases recorded, respectively.
Overall, May was the month with the higher number of reported cases within this period while the highest number of suicides reported between 2004 and 2015 occurred in January 2011.
The computer scientist noted that suicide is influenced by a variety of factors which makes prediction a highly complex task but their research is an alternative source in the field of suicide prevention.
“Through the identification of search terms related to suicide we can contribute to awareness and prevention campaigns targeting the population at-risk.
When search terms such as ‘depression’ and ‘feeling down’ are used, HSE-approved support websites and helplines can be prioritised in the search result shown to the potentially at-risk user.
“Other research has also shown that search terms can reveal timing and targets for prevention campaigns.
"Publicised suicides - such as with celebrities and in TV shows - can lead to an increased search for suicide related terms.
“The same strategy where support information is prioritised can be applied here as well.”