One of the first symptoms of early stage lung cancer is a chronic cough. However, what makes a lung cancer cough distinct from other types of coughs due to ailments such as respiratory infection or a cold?
Our guiding idea is very simple: are there detectable changes in a person’s cough or voice that we can leverage for early lung cancer detection? Can we develop a consumer friendly interface for our machine learning model?
We developed a proof of concept machine learning algorithm that is able to cluster coughs obtained from lung cancer patients, children with whooping cough from normal, everyday coughs. From this preliminary data, we were able to develop a prototype Alexa skill to classify coughs.
Canaries once served as warnings for toxic environments. We wanted to reimagine how to detect lung cancer earlier. So we built an AI to be like a canary, something that doesn’t change how you live. But is capable of warning you, before it’s too late.