Use of Machine Learning Techniques to Identify HIV Predictors for Screening

Abstract

Aim: HIV prevention measures at sub-Saharan Africa are still short of attaining the UNAIDS 90-90-90 fast track targets set in 2014. Identifying predictors for HIV status may facilitate targeted screening interventions that improve health care. We aimed at identifying HIV predictors as well as predicting persons at high risk of the infection.Method: We applied six machine learning approaches for building models using population-based HIV Impact Assessment (PHIA) data for 41,939 male and 45,105 female respondents with 24 and 29 variables respectively from four countries in sub-Saharan countries. We trained and validated the six algorithms on 80% of data and tested on the remaining 20% where we rotated around the left-out country. An algorithm with the best mean f1 score was retained and trained on the most predictive variables. We used the model to identify people living with HIV and individuals with a higher likelihood of contracting the disease.Results: Application of XGBoost algorithm appeared to significantly improve identification of HIV positivity over the other six algorithms by f1 scoring mean of 78.9% and 92.8% for males and females respectively. Amongst the eight most predictor features in both sexes were: age, relationship with family head, the highest level of education, highest grade at that school level, work for payment, avoiding pregnancy, age at the first experience of sex, and wealth quintile. Model performance using these variables increased significantly compared to having all the variables included. We identified five males and seven females individuals that would require testing to find one HIV positive individual. We also predicted that 4·14% of males and 10.81% of females are at high risk of the infection.Conclusion: Our findings provide a potential use of XGBoost algorithm with socio-behavioural-driven data at substantially identifying HIV predictors and predicting individuals at high risk of infection for targeted screening.

A machine learning model to explore individual risk factors for tuberculosis treatment non-adherence in Mukono district

“…These results indicate that classification machine learning models can be used to map the different predictors to their respective classes. Likewise, these capabilities have been reported in similar rather challenging scenarios like predicting viral failure (53), tweets classification for disease surveillance (59), identification of HIV predictors for screening (60), and dermatology conditions (67). Therefore, this demonstrates the potential and applicability of machine learning algorithms to provide insights in scenarios where human decision-making would be limited.…”

“…Lastly, this study notes the increasing use of ML for specific customer loan default predictions in the banking sector (58), multilingual tweets classification for disease surveillance (59), and predicting an individual HIV/AIDs patient likely not to adhere to treatment (60). Similarly, this could be replicated in TB management and research.…”

A machine learning approach to explore individual risk factors for tuberculosis treatment non-adherence in Mukono district

“…These results indicate that classification machine learning models can be used to map the different predictors to their respective classes. Likewise, these capabilities have been reported in similar rather challenging scenarios like predicting viral failure [51], tweets classification for disease surveillance [43], identification of HIV predictors for screening [52], and cancer [53]. Therefore, this demonstrates the potential and applicability of machine learning algorithms to provide insights in scenarios where human decision-making would be limited.…”