Imagine this: if people saw their doctor every day, they would probably take their pills. Getting that extra information and insight into their health makes them assume a greater responsibility for it. Obviously, seeing your doctor isn't possible, but what if we could create a system that gives real time health updates?
What it does
We provide data analytics based on current medications and other health metrics. The main part of our system are our 'activities', which measure general or specific health, based on the medication the patient is on. Activities are somewhat like tests, that a doctor would perhaps carry out during a physical. By automating scoring on these activities, patients can monitor their own health around their doctor visits.
How I built it
We used Swift and tapped into the CareKit API to build our pill tracking capabilities. Then, we trained a model on Keras that scores pictures of clocks drawn by Alzheimer's patients as a proof of concept for the generic concept of prognosis tracking activity. These activities can be very easily built out in the future based on our code design and code we built around our Keras neural net to ingest varied forms of data.
Challenges I ran into
Negotiating Swift was tough especially as we were relative newcomers.
Accomplishments that I'm proud of
We are particularly proud of how we pivoted our idea. We started out with a smart pill bottle, but realized it was an oversaturated idea, and took our half-developed pill tracking platform and repurposed it for the data driven concept.
What I learned
We learned how to use Flask, integrate AWS services into our project, and integrate development on multiple platforms (web, iOS)
What's next for Anadose
Taking this to the next level, adding more useful activities, doing lots of work on the data science end to create more meaningful insights, and building on tracking robustness.