DrNose is an electronic nose that uses a person’s exhaled breath to identify patterns associated with multiple early-stage cancers and other diseases in real time. The intent is to disrupt the diagnostics industry by building a system that attempts to emulate a dog’s nose: a single device designed to provide accurate, real-time screening across a large number of diseases.
This has required a multidisciplinary team involving medical practitioners, chemical engineers and electronics experts to build the device. To generate intelligent results from the platform, the company also brought together a team of AI experts.
Traditional artificial intelligence techniques would have required huge amounts of data, massive computing resources and significant training time. That was not feasible for a device intended to provide high accuracy and specificity across a range of cancers and other diseases, especially when large datasets are not readily available for many of them. The team also wanted the system to operate in real time.
The DrNose AI team addressed these limitations with its “Intuitive AI” capability, which can work with very small datasets—10 to 50 examples rather than thousands. When datasets are small, less labelling and training are required, and there is no requirement for GPUs. This reduces the overall cost of the system and enables faster development and deployment.