Analysis of social determinants of health based on patterns identified with Machine Learning techniques in the municipality of Tumaco, Nariño

Abstract

San Andrés de Tumaco, located in the Southern Pacific region, is the coastal municipality of Nariño most affected by violence and its repercussions on public health. With a population of 217,079 inhabitants and a slight male predominance of 50.1%, 57% of its inhabitants reside in the municipal seat. The region faces a high incidence of diseases such as snakebite accidents, dengue, and tuberculosis, in addition to elevated levels of residual chlorine in the water, making it unsuitable for human consumption. Likewise, the municipality has a rate of Unsatisfied Basic Needs of 48.74% and only 11.6% of its population has formal employment. Between 2013 and 2018, Tumaco was the scene of the highest number of violent acts in Nariño, which has severely impacted the quality of life of its inhabitants.

The health of women in Tumaco has been particularly affected by the armed conflict and humanitarian crises stemming from the structural violence that persists in the territory. With the aim of mitigating these impacts, various organizations and territorial entities have implemented strategies to strengthen the protection and well-being of the female population, focusing on training in violence care, recognition of rights, awareness of physical and mental integrity, and education to promote financial independence. However, these efforts still require reinforcement in various strategic areas to ensure sustainable and effective results in reducing gender-based violence and other social issues.

This article analyzes the social determinants of health in San Andrés de Tumaco through a data model based on unsupervised learning, specifically the clustering technique. The implementation of this model allowed for the identification of the effects of violence on health from the perspective of social determinants in a territory historically impacted by armed conflict, drug trafficking, and violence perpetrated by illegal armed groups. These conditions have deteriorated collective well-being and the health of the population, increasing the challenges for the institutions responsible for social protection. The study addresses violence against women and other forms of violence from a comprehensive perspective that considers social, familial, economic, cultural, physical, and psychological factors, highlighting the importance of establishing specialized and comprehensive care pathways for victims.

The present study is based on the findings of the project "Strengthening the implementation of strategies for violence control through data science models that allow the study of social determinants of health." Its purpose was to develop a tool that would allow the identification and analysis of the social determinants of health and, based on that, generate effective strategies to mitigate violence against women in Tumaco using artificial intelligence. For its development, three main phases were established: the first consisted of describing the events recorded in the SIVIGILA 356, 365, and 875 surveillance systems; the second phase included the analysis of the data obtained through the implemented data science model, and finally, the third stage focused on designing strategies for the territorial entity based on the results of the analysis conducted.

As a result, a hermeneutic-phenomenological analysis was developed that allowed for the interpretation and understanding of social phenomena in health, providing a solid foundation for designing strategies aimed at mitigating violence and improving the living conditions of the population. This approach allowed for the identification and strengthening of actions that directly impact the social determinants of health in the district of Tumaco, providing tools for decision-making in public policies and the design of more effective interventions in the fight against gender-based violence and other associated issues.