How can we identify people who are at high risk for infectious diseases? This lesson and emphasis on predictive measures has not been specific to COVID-19, although it has certainly created a setting for paying more attention to how we can intervene in transmission. infectious diseases.
Whether it’s a respiratory disease like COVID-19, a sexually transmitted disease like the Human Immunodeficiency Virus (HIV), or a zoonotic pathogen like the Ebola virus infection, there are factors that increase danger for its spread. Despite working to combat HIV / AIDS since it was first identified decades ago, there are still major hurdles that we need to overcome.
A research team recently demonstrated their efforts to integrate machine learning to get people at risk of HIV in rural Kenya and Uganda, and they appear in a journal of the recent publication, Clinical infectious diseases. UNAIDS reports that, in 2019, in Kenya there were approximately 1.5 million people living with HIV and an incidence rate in adults of 4.5. In the same year, UNAIDS reported in Uganda, approximately 1.5 million people live with HIV and an adult incidence rate of 5.8.
The authors of the study noted that, of these 16 communities within the SEARCH study, more than 75% were diagnosed each year between 2013-2017. The team used three strategies to incorporate demographic factors into a prediction model for HIV seroconversion over one year at a time.
They noted, “in this population, we evaluated 3 strategies for using demographic factors to predict 1-year risk of HIV seroconversion: membership in ≥1 a known ‘risk group’ (e.g., a spouse living with HIV), a model-based risk score constructed with logistic regression, and a ‘machine learning’ risk score constructed with the Super Learner algorithm. We thought that machine learning would identify people at high risk in a more effective way (fewer people targeted for established sensitivity) and with higher sensitivity (for a targeted number) than the other approach. “
During this study period, more than 75000 people were assessed, with more than 166000 years later, and a total of 519 seroconversion events. Through this evaluation, the authors noted that machine learning improved efficiency. A 50% established sensitivity was achieved through the risk group strategy targeting 42% of the population, a model-based strategy targeting 27%, and the machine learning strategy was targeted. aimed at 18% of the population.
Simply put, machine learning, which is the next stage of automation that allows machines to make more decisions, allowed more sensitivity and sorting of those at higher risk of getting HIV. This strategy was more effective than traditional methods such as modeling or trust in identified risk groups. Here is another useful example of how artificial intelligence can play a role in the treatment of infectious diseases and biotreats.
It will be crucial, however, to consider the use of bioethics and the fundamental limitations of these technologies. Efforts like this should be evaluated not only for their effectiveness, but also their potential for harm, meaning that these should include wholistic evaluation groups. There is great potential for machine learning in responding to infectious diseases, and now more than ever, we need to carefully consider and evaluate such technologies.