Artificial intelligence-driven mobile interpretation of a semi-quantitative cryptococcal antigen lateral flow assay.

Area of Science:

• Medical Diagnostics
• Artificial Intelligence in Healthcare
• Infectious Disease Research

Background:

• Cryptococcosis is a significant global health threat requiring rapid diagnostics.
• Point-of-care tests (POCTs) like the cryptococcal antigen semi-quantitative lateral flow assay (CrAgSQ LFA) show promise but suffer from subjective interpretation.
• There is a critical need for objective and reliable methods to interpret CrAgSQ LFA results.

Purpose of the Study:

• To develop and validate an Artificial Intelligence (AI) digital platform for interpreting CrAgSQ LFAs.
• To assess the AI platform's performance in semi-quantitative interpretation of LFAs.
• To explore the AI's capability in directly quantifying cryptococcal antigen (CrAg) concentrations from LFA images.

Main Methods:

• An AI algorithm was developed to automate the interpretation of 318 CrAgSQ LFAs.
• A dataset of 1272 images was generated using two smartphones to capture LFAs with known CrAg concentrations (0-5000 ng/ml).
• The relationship between test line intensity on LFAs and CrAg concentrations was analyzed.

Main Results:

• The AI algorithm demonstrated superior sensitivity and fewer discrepancies compared to visual interpretation (p < 0.0001).
• The AI system accurately predicted CrAg concentrations from LFA images with a Pearson correlation coefficient of 0.85.
• The developed platform offers objective and quantifiable results for CrAgSQ LFAs.

Conclusions:

• AI-driven interpretation of LFAs offers standardized, reliable, and efficient POCT results for cryptococcosis diagnosis.
• The technology's adaptability suggests potential applications beyond cryptococcosis diagnostics for various LFAs.
• This AI platform has the potential to revolutionize infectious disease diagnostics at the point of care.