We aim to improve the nurses' diagnostic scope, reduce test lead time and increase accuracy while shifting the point of final diagnosis and treatment to primary healthcare facilities. Ultimately, improving public care while reducing expenditure.
The field of AI has experienced huge advancements in computer vision, we leverage this technology by converting our sound files to image like spectrograms. Using state of the art deep neural networks we are able to clinically classify bodily sounds.
Our local processing unit executes the AI model locally, allowing offline use. Connection to a cloud-based patient registry enables model updates and verification. The display guides users on how and where to record patient's bodily sounds.
Our wireless digital stethoscope captures the patients lung sounds. The device is fitted with active noise cancelling technology to accommodate use in crowded public clinic environments.
Our TB screening AI model is 90% sensitive and 84% specific. This will halve the number of TB positive patients currently left undiagnosed by the South African TB diagnoses program while halving the number of TB negative patients currently sent for lab tests unnecessarily.
MSc Eng. Mech, BEng Mechatronics
MSc Biomaterials, BTech Chem Eng.
BSc Eng. Elec.