Malaria remains a devastating global crisis, claiming over 608,000 lives annually, primarily children in sub-Saharan Africa. Current diagnostics are unreliable and incapable of measuring infection severity, leading to antimalarial overuse and increased drug resistance. We aimed to provide low-cost, automated, and accurate diagnostics in resource-limited settings by overcoming the accuracy limitations of Rapid Diagnostic Tests and the accessibility barriers of traditional microscopy.
We developed the Hem.AI system, a compact battery-powered device that combined microfluidic sample preparation with AI-driven diagnostics. Our device automates the entire workflow by utilizing a computer-driven imaging array and a Convolutional Neural Network to identify Plasmodium parasites. This approach eliminates the need for trained pathologists and reduces the time-to-result from two to four hours to under ten minutes.