Epidemiological vigilance of transmissible disease in Colombia through generative models based on Artificial Intelligence to follow donors in Colombia

The case of HTLV-1 and 2

Abstract

Introduction: human T-cell lymphotropic virus (HTLV) types 1 and 2 belong to a group of globally distributed oncogenic retroviruses. This viral infection is associated with adult T-cell lymphoma. This virus is endemic in Colombia; however, data regarding its incidence and prevalence is scarce.

Method: this study was carried out in two phases. The first consisted of a descriptive observational study of a retrospective cohort of blood bank donors at Fundación Hematológica de Colombia from 2012 to 2022, to determine the seroprevalence of HTLV-1 and 2. In the second phase, the data of women with the potential for vertical transmission of the virus was transformed into an “InovaD Epidemiology” platform based on generative artificial intelligence (AI) 3.5 models and natural language, which permits epidemiological surveillance in this country.

Results: during the study, 1,553,478 patients were recruited, 3,484 of whom were seropositive, for a global prevalence of 0.22% over 10 years in the donor population. The periods with the highest incidence were 2014-2015 (0.26%) and 2018-2019 (0.25%). The cities with the highest prevalence were Bogotá with 48% and Ibagué with 13%. Fifty-four percent of the seropositive women were of reproductive age. A generative AI model was constructed that allows epidemiological surveillance of this virus in our country for the first time, with 100% effectiveness in its pilot test.

Conclusions: our cohort has a higher seroprevalence compared to North America. A total of 1,100 HTLV-1 and 2 seropositive women could transmit the virus (potentially also through breastfeeding). Our digital platform could optimize reporting of positive tests to public health agencies, to implement pertinent measures and conduct epidemiological surveillance in our country