Mexican students develop bot to detect the risk of life threatening

Mexico .- Graduates of the University of Guadalajara (UdeG) are working on the development of a bot that, through a series of algorithms, manages to detect risk factors and suicidal behaviors in young people, reported in A statement from the National Council of Science and Technology (CONACYT).

The goal, said students Manuel Santana Castolo and Patricia Brand, project manager, is that the bot can interact with users through the messaging service offered by Facebook and that when there is risk factors for suicide, the person can be referred to the necessary care.

"The idea is that we can detect whether there is depression or risk of suicide, it is not a diagnosis, only a detection and if possible a pipeline with health professionals appropriate, "says Brand, a graduate in psychology and a graduate of biomedical engineering.

Brand was responsible for developing the methodology for the detection of suicide risk factors.

Santana, a biomedical engineer, said they're looking to implement artificial intelligence algorithms that help the bot deal with users' issues and, over time, better understand the answers.

According to the most recent data from the World Health Organization, Health (WHO) in 2015, the average suicide rate in Latin America was 9.8 per 100,000 population.

This means about 65,000 deaths a year.

In Mexico, according to the National Institute of Statistics and Geography (Inegi), 6,285 suicides were reported in 2015 across the country, which corresponds to a rate of 5.2 per 100,000 population.

Santana, Ph.D. student in electronics and computer science from the University Center of Exact Sciences and Engineering of UdeG explained that, as a first step, the chatbot will receive the basic user information to develop a first profile about the person interacting.

Thereafter, work will be done with questions based on medical manuals to map out a route that helps detect risk factors.

Once artificial intelligence detects risk factors, it will contact previously established users. as a reference, or with institutions that provide mental health treatment and counseling.

The two students hope that in mid-October a first phase of this bot will be published in control patients with antecedents.

Brand said that for the chatbot has a database to react to users, he is working on the development of a series of information that strengthens artificial intelligence, based on available manuals and tests that are the cornerstone of the design. Ways to build the bot perspective.

"We designed it to be able to detect and evaluate responses, we assign a score to the answers users give and we also want to incorporate an algorithm that starts to detect language indicators, which are the types of words used by users, people who have been detected with depression, for example, "explained the student.

Both developers said that at the In the future, it is thought to integrate the voice analysis function of users to detect, by factors such as tone of voice, the possible risks of depression to be treated preventively.