Predicting malaria-infected cells with a 95% accurate neural network
Malaria is a tropical disease transmitted by mosquitos. Despite the fact that it can be cured, it is still highly prevalent worldwide (mostly in poorer countries). Malaria-infected cells can easily be identified under a light microscope. Using data from NIH, I have built a (small) convolutional neural network that can classify cells as either parasitised or uninfected with 95% accuracy.
I am working on building a python frontend to this model, so that you can see the model working in a REPL.
Unfortunately I had to train this model elsewhere as (a) repl.it does not offer GPU access (probably smart) and (b) it's easier to do this in a Jupyter notebook (which should definitely be added to repl.it).