DigiTriage

Accurate CD4 testing is essential for identifying Advanced HIV Disease (AHD) and preventing life-threatening opportunistic infections such as tuberculosis (TB). The VISITECT® CD4 Advanced Disease assay (VISITECT-CD4-AD), developed by AccuBio, is currently the only World Health Organization (WHO)-prequalified point-of-care (POC) CD4 test available that works without electricity or advanced equipment, making it especially valuable in remote and low-resource settings.

The VISITECT-CD4-AD varies from other POC tests (e.g. malaria, COVID) in that the result requires comparison of intensity of two test lines, rather than simple presence of a line. The visual interpretation of VISITECT-CD4-AD test results can vary by user and experience level, which may lead to misclassification and missed opportunities for timely care. To address this, the Digital Health Solutions and Diagnostics teams at KNCV TB Plus have come up with a potential solution: the DigiTriage tool.

The DigiTriage tool is being piloted in Addis Ababa, Ethiopia, with the support of a local partner from the KNCV Network: KNCV Ethiopia. During this phase, the project will include selected health facilities already using VISITECT-CD4-AD for routine care, and will also conduct limited CD4 count testing via FACSPresto for comparison. This project will validate:

1. Diagnostic accuracy of the DigiTriage app compared to laboratory-based CD4 results; and
2. Agreement between human-read VISITECT results, AI-interpreted VISITECT test results, and FACSPresto CD4 count outcomes.

Expected Outcomes

• A working prototype of the DigiTriage tool to improve timely and accurate detection of AHD.
• Insights into the usability and feasibility of AI-assisted testing from health worker feedback.
• An optimized DigiTriage tool for possible national rollout in routine care settings.

Why Telegram?

In Ethiopia, Telegram is one of the most widely used free messaging platforms, making it an ideal choice for frontline health workers. The DigiTriage tool integrates directly within this familiar app for ease of use and accessibility. Lab technicians conduct the VISITECT test as normal, interpret the result, and then open a dedicated Telegram chat to send a photo of the VISITECT test cartridge. This is automatically processed in the back end by Audere’s specialised AI computer vision model. For the purposes of this research, both the human readers and the AI model are blinded to each other’s test interpretations. 

Looking Ahead

The Ethiopian Ministry of Health has already endorsed the VISITECT-CD4-AD test as a standard CD4 testing method. DigiTriage aims to complement this rollout by improving test interpretation reliability, bringing digital innovation to the front lines of HIV care.

KNCV and partners also recognize the opportunities around refining the computer vision models to reliably support in test interpretation of TB-LAM, which is a key diagnostic in Ethiopia and other settings to facilitate comprehensive TB care. Combined with the VISITECT support, improved TB-LAM interpretation could provide a powerful multi-disease tool for many health systems.