Introduction:

The beginning of 2020 was accompanied by a pandemic caused by the virus called SARS-CoV-2. With social distancing measures implemented to prevent the spread of this virus, mental health problems arose, such as anxiety, depression, etc., resulting in a need for telemedicine. Given the alarming numbers of suicide incidences in today's society, coupled with these distancing measures, support tools are required to identify individuals at risk of committing suicide.

Objective:

To propose and evaluate a new methodology to calculate suicide risk in Twitter users, based on the analysis of emotions.

Materials and Methods:

Using statistical learning models (supervised and unsupervised), the proposed methodology identifies the level of risk in the analyzed text of 77 tweets from regular users and political figures in Mexico and Latin America.

Results:

It was found that, when comparing the methods used, the percentage of coincidence in classification is close to 96%, being the supervised non-parametric and unsupervised methods those that detected the extreme levels of suicide risk. Conclusions: the proposed methodology is a tool that can be of great support for specialists in the mental health area by helping to identify, in a massive way, the presence of signs of mental illness, for its subsequent diagnosis.

        · abstract in Spanish | Portuguese     · text in Spanish     · Spanish ( pdf )