Studies show that around 40% of patients make a mistake with their medication each year and 25% of these errors are confusing different medications and therefore not taking the right one. Elderly, polymedicated, and multipathological people are at highest risk. Among patients who committed an error, 5% had to go to hospital and 3.9% required a new treatment. (Reference: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5858379/)
What it does
The tool takes voice input from Alexa ("Alexa, is this my pill?") then takes a picture of the pill using a camera connected to a Raspberry Pi. Requests are sent to a machine learning classifier which identifies this pill. This is cross-corresponded with the patients' medication prescriptions (taken via an API from TPP healthcare's patient records) and the time of day, to advise the patient whether it is the right pill to take. It confirms whether the patient takes the pill, and records it for their doctor.
How we built it
We used some technologies like PyTorch, Alexa skills, Raspberry Pi, AWS Lambda and Google Cloud.
Challenges we ran into
Integrating everything and finding some of the hardware
Accomplishments that we're proud of
Making a working voice app that has a positive impact in the personal healthcare
What we learned
How to integrate so many technology
What's next for Pill_It_Up?
Functions to automatically schedule reminders
Information about side effects
Additionally features aimed at elderly patients (reminder of actions such as locking back door, turning off oven, storing shopping lists)
Functions for contacting relatives