Two graduate students at the University of Guadalajara are developing a bot that will be capable of detecting potential suicide risks among users of Facebook’s popular Messenger application.
Manuel Santana Castolo and Patricia Brand said in a statement issued by the National Council for Science and Technology (Conacyt) that the aim of their automated program is to interact with users on the social media site and refer them to medical professionals if suicide warning signs are detected.
Brand, who has a degree in psychology and is now studying biomedical engineering, is responsible for developing the methodology the bot will use to detect suicide risks.
Santana, a biomedical engineering graduate and PhD student in computer science, said the bot will function by asking users questions that are based on medical knowledge and practice.
Artificial intelligence algorithms he is developing will help the bot to interpret the responses it receives and over time it will be able to better understand them.
The bot also has a database that it can access to respond to the social media user in a relevant way.
Certain words that have been identified among those commonly used by people with depression could act as warning signs that alert the bot to the possibility that the user is suffering from that condition.
If artificial intelligence built into the so-called chatbot detects suicide risk factors, users will be provided with the contact details of medical facilities where they can be assessed and/or treated for mental illness.
The students said they hope to start testing a pilot version of the bot in the middle of October. People with and without a history of mental illness are slated to participate.
Santana and Brand said a future function might allow it to analyze speech and pick up on additional potential warning signs of mental illness such as tone of voice.
According to the National Institute of Statistics and Geography (Inegi), there were 6,285 suicides in Mexico in 2015, a rate of 5.2 per 100,000 inhabitants.
Source: EFE (sp)