JMIR Publications

ABSTRACT

Background:

Due to concerns about privacy breaches and fear of social stigma, some individuals engaging in high-risk behaviors—particularly men who have sex with men (MSM)—may avoid in-person HIV testing services and instead choose to purchase HIV self-test kits online for self-testing. Unlike traditional e-commerce platforms that directly provide self-test kits without conducting risk assessments, the “Easy Test Know” platform integrates a structured behavioral risk assessment prior to HIVST kit purchase, allowing users to be categorized into different HIV risk groups.

Objective:

This study aimed to evaluate the performance of a digital HIV behavioral risk stratification tool integrated with online self-testing among MSM, and to determine whether incorporating multi-dimensional behavioral indicators—particularly high-risk venue diversity score—could improve the prediction of HIV positivity compared with the platform’s existing rule-based risk scoring algorithm.

Methods:

This retrospective observational study utilized data from the “Easy Test Know” platform between November 2023 and February 2025. The study included MSM classified as medium or high risk by the platform algorithm. Firth logistic regression models were developed to predict HIV positivity using behavioral variables, including high-risk venue diversity score. Model performance was evaluated using area under the receiver operating characteristic curve (AUC) and calibration analyses.

Results:

A total of 9,961 participants were included, of whom 90.8% were classified as high risk by the platform. The overall HIV positivity rate was 0.59%, with a higher rate in the high-risk group. The platform’s baseline risk score demonstrated moderate discriminatory power for predicting HIV positivity (AUC = 0.646). High-risk venue diversity score showed a dose–response association with HIV positivity. The data-driven model incorporating the platform’s existing variables together with the newly derived high-risk venue diversity score improved discrimination compared with the rule-based platform algorithm (ΔAUC = +0.082). The fully adjusted model achieved the highest predictive performance (AUC = 0.752).

Conclusions:

Digital behavioral risk stratification demonstrated moderate ability to differentiate HIV risk among MSM using online HIV self-testing services. Compared with the platform’s original rule-based scoring system, data-driven behavioral models incorporating high-risk venue diversity score showed improved predictive performance. These findings support the potential value of data-driven approaches for optimizing digital HIV risk assessment tools.