Ethnic disparity in diagnosing asymptomatic bacterial vaginosis using machine learning.

While machine learning (ML) has shown great promise in medical diagnostics, a major challenge is that ML models do not always perform equally well among ethnic groups. This is alarming for women's health, as there are already existing health disparities that vary by ethnicity. Bacterial Vaginosis (BV) is a common vaginal syndrome among women of reproductive age and has clear diagnostic differences among ethnic groups. Here, we investigate the ability of four ML algorithms to diagnose BV. We determine the fairness in the prediction of asymptomatic BV using 16S rRNA sequencing data from Asian, Black, Hispanic, and white women. General purpose ML model performances vary based on ethnicity. When evaluating the metric of false positive or false negative rate, we find that models perform least effectively for Hispanic and Asian women. Models generally have the highest performance for white women and the lowest for Asian women. These findings demonstrate a need for improved methodologies to increase model fairness for predicting BV.

A multiplex assay for detection of SHIV plasma and mucosal IgG and IgA.

Evaluating antibody maturation provides valuable data to characterize immune responses to HIV infection and can provide insight into biomedical intervention efficacy. It is important to develop assays that evaluate antibody maturation in both plasma and mucosal compartments. The nonhuman primate model provides a controlled system to collect temporal data that are integral to assessing intervention strategies. We report the development of a novel multiplex assay, based on the Bio-Plex platform, to evaluate plasma and mucosal IgG and IgA avidity and maturation against simian/human immunodeficiency virus (SHIV) in this controlled system. Vaginal mucosa and plasma samples were collected from a prior study evaluating the efficacy of a tenofovir disoproxil fumarate (TDF) intravaginal ring (IVR) against SHIVSF162P3 challenge in female pigtailed macaques. For validation of the multiplex assay, specimens from six SHIV-infected placebo animals and one TDF breakthrough animal were evaluated. For SHIV and HIV envelope analytes, antibody levels and avidity in both compartments continued to mature post-infection. Maturation of IgG and IgA levels was similar in each compartment, however, mucosal antibody levels tended to be more variable. This SHIV assay elucidates IgG/IgA antibody kinetics in the plasma and vaginal mucosa and will be a valuable tool in vaccine and other biomedical intervention studies in the nonhuman primate model.

Low-Cost Method to Monitor Patient Adherence to HIV Antiretroviral Therapy Using Multiplex Cathepsin Zymography.

Monitoring patient adherence to HIV antiretroviral therapy (ART) by patient survey is inherently error prone, justifying a need for objective, biological measures affordable in low-resource settings where HIV/AIDS epidemic is highest. In preliminary studies conducted in Ethiopia and South Africa, we observed loss of cysteine cathepsin activity in peripheral blood mononuclear cells of HIV-positive patients on ART. We optimized a rapid protocol for multiplex cathepsin zymography to quantify cysteine cathepsins, and prospectively enrolled 350 HIV-positive, ART-naïve adults attending the Themba Lethu Clinic, Johannesburg, South Africa, to test if suppressed cathepsin activity could be a biomarker of ART adherence (103 patients were included in final analysis). Poor adherence was defined as detectable viral load (>400 copies/ml) or simplified medication adherence questionnaire, 4-6 months after ART initiation. 86 % of patients with undetectable viral loads after 6 months were cathepsin negative, and cathepsin-positive patients were twice as likely to have detectable viral loads (RR 2.32 95 % CI 1.26-4.29). Together, this demonstrates proof of concept that multiplex cathepsin zymography may be an inexpensive, objective method to monitor patient adherence to ART. Low cost of this electrophoresis-based assay makes it a prime candidate for implementation in resource-limited settings.