Mammogram Abnormality Attribution Toolkit

In Developing Countries,  90% of women do not have access to preventive mammography resources which translates into a 66% survival rate after 5 years of detection compared to 90% in 'Western' countries.  Although the driver of these figures are varied, we are specifically attempting to reduce the dependency on radiologists to read x-ray images and increase preventative screening exams by embedding a machine learned model on x-ray machines to give a risk assessment at the point of screening making it cheaper, faster and less resource (radiologist) dependent than the current process built using a western mentality to medical availability.  In technical terms, we are building our model using Convolutional Neural Networks coupled with an elastic web-app to store weights and biases from the most current models with two-way data flow updating the model continuously for ongoing learning, the benefit of this is a fast analysis/lean model design at the point of screening but a 'living' model as we gather more data for refining the model.  We hope you share our passion to potentially save literally tens of millions of lives per year and invite you to join us to bring more breast screening opportunities to women in developing countries.