Abstract
Malaria remains among the world's deadliest diseases, and control efforts depend critically on the availability of effective diagnostic tools, particularly for the identification of asymptomatic infections, which play a key role in disease persistence and may account for most instances of transmission but often evade detection by current screening methods. Research on humans and in animal models has shown that infection by malaria parasites elicits changes in host odors that influence vector attraction, suggesting that such changes might yield robust biomarkers of infection status. Here we present findings based on extensive collections of skin volatiles from human populations with high rates of malaria infection in Kenya. We report broad and consistent effects of malaria infection on human volatile profiles, as well as significant divergence in the effects of symptomatic and asymptomatic infections. Furthermore, predictive models based on machine learning algorithms reliably determined infection status based on volatile biomarkers. Critically, our models identified asymptomatic infections with 100% sensitivity, even in the case of low-level infections not detectable by microscopy, far exceeding the performance of currently available rapid diagnostic tests in this regard. We also identified a set of individual compounds that emerged as consistently important predictors of infection status. These findings suggest that volatile biomarkers may have significant potential for the development of a robust, noninvasive screening method for detecting malaria infections under field conditions.
Highlights
Malaria remains among the world’s deadliest diseases, and control efforts depend critically on the availability of effective diagnostic tools, for the identification of asymptomatic infections, which play a key role in disease persistence and may account for most instances of transmission but often evade detection by current screening methods
We found that predictive models based on machine learning algorithms reliably determined infection status based on volatile biomarkers and, critically, identified asymptomatic infections with 100% sensitivity, even in the case of low-level infections not detectable by microscopy
These findings suggest that volatile biomarkers have significant potential for the development of robust, noninvasive screening methods for detecting symptomatic and asymptomatic malaria infections under field conditions
Summary
Sample Collection and Determination of Malaria Infection Status. Between 2013 and 2016, we collected samples of skin volatiles from more than 400 primary-school children (aged ≤12 y) at 41 schools across 21 localities within the Mbita area of western Kenya (Fig. 1). To characterize the volatile signatures associated with each category of infection status, we used machine learning algorithms that develop tree-based ensemble classification models, which aim to identify a minimal set of compounds that correctly classify individuals These algorithms were used to “train” models on 70% of samples from K2, iteratively eliminating the least important compounds (i.e., those making the smallest contribution to accuracy) to obtain a subset resulting in the best. We observed differences in the effects of symptomatic and asymptomatic infections on volatile profiles, we tested the ability of our algorithms to detect all malaria cases without regard to symptom status and including submicroscopic infections. Compound 14 (ethylcyclohexane) appeared as a predictor in some models and, along with compound 38 (propylcyclohexane), exhibited an interesting pattern in which its emission exhibited relatively strong up-regulation in AS individuals and relatively strong suppression in S individuals (Fig. 4)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
