Descriptive and predictive statistical analysis of suicide attempts in Colombia and the identification of suicidal ideation in social networks using deep learning models.

Abstract

Objective: descriptive and predictive statistical analysis of data on suicide attempts in Colombia between 2016 and 2023. Additionally, develop the study of unstructured data from a social network, using a deep learning model. Methodology: the CRISP-DM process (Cross Industry Standard Process for Data Mining) is used as a proven method to guide data mining projects, providing a structured and efficient approach. Results: data on suicide attempts show an upward trend, with a cumulative total of 237,924 reported cases. In 2023, the monthly average was 3,404 attempts, which equates to approximately 111 cases per day. In addition, sentiment analysis conducted on social network comments revealed that 87.35% of these present a high probability of being associated with suicidal ideation or negative intent. Discussions: in relation to discussions about suicidal ideation on social networks, it is observed that most of the conversations contain expressions of this type. It has been pointed out that more and more people are using social networks to express suicidal feelings. Conclusions: the findings reinforce the need to implement more dynamic public health policies, focused on prevention and monitoring of vulnerable groups, both in the physical and digital environment.