Nanotrap-AI Integration Enables Ultra-Sensitive Point-of-Care HIV Testing

Area of Science:

• Biomedical Engineering
• Infectious Disease Diagnostics
• Artificial Intelligence in Healthcare

Background:

• Early HIV detection is crucial for effective treatment and prevention.
• Current noninvasive methods using oral fluid face challenges due to low antibody concentrations and inhibitors.
• Lateral-flow assays (LFAs) lack sensitivity for early HIV detection in oral samples.

Purpose of the Study:

• To develop a novel diagnostic platform, BE-SMART-HIV, for sensitive and early HIV detection using noninvasive oral fluid samples.
• To overcome the limitations of low antibody levels and inhibitors in oral fluid for HIV diagnostics.
• To improve the accuracy and speed of HIV screening in resource-limited settings.

Main Methods:

• Integration of a BEETLES^2-inspired bioengineered enrichment (BE) nanotrap for antibody concentration and inhibitor removal.
• Utilization of a smartphone-based deep-learning reader (SMART) for automated test line interpretation.
• Validation using a longitudinal seroconversion panel and sample dilutions up to 1,000-fold.

Main Results:

• The BE nanotrap achieved approximately 20-fold antibody enrichment and effective inhibitor removal.
• The AI model demonstrated 98.6% accuracy in interpreting weak test lines, surpassing human performance.
• BE-SMART-HIV detected antibody emergence up to 8 days earlier than conventional LFAs, capturing IgM and IgG kinetics.

Conclusions:

• BE-SMART-HIV offers a practical and noninvasive approach for high-fidelity HIV screening.
• The platform enables early detection by capturing systemic antibody kinetics from enriched oral fluid specimens.
• This technology holds significant potential for repeated HIV testing in resource-limited settings.