The challenge of long-term lung diseases is multifaceted; and it requires a machine learned model that can intelligently identify the potential lung disease that is affordable, quick, accurate and easy to implement. Globally, more than 50% of TB cases are misdiagnosed. Over 1.7 billion people are infected and present with Tuberculosis symptoms and the disease itself annually. The screening of X-Rays of people is time consuming, especially when they are in large numbers and in cases where a compensation needs to be paid out to them or further diagnosis and treatment need to be suggested. So, in order to shorten this period of screening without reducing the accuracy levels, Dake has developed a Machine Learning model, fronted by an Edge Computing Device that can automatically screen scanned X-ray images and classify them into Tuberculosis (TB), and detect for other abnormal lung conditions. This cutting-edge innovation was made possible with the help of research partnerships of the WITS University Health Consortium and the University of British Columbia.

The model is packaged into various application logics to work online as Software as a Service (SaaS). It is also packaged and hosted in miniature computing devices such as Dake’s MeDedge, this makes the software completely portable and makes it work offline. This enables the use of this device in remote or rural environments even without any network connectivity. The device and the model are designed in such a way that it does not store any images after processing the digital X-rays, only the results are stored in the Device.