Both chronic pain disorders and opioid misuse are on the rise, and the two are even more related than you might think -- over 60% of people who misused prescription opioids did so for the purpose of pain relief. Despite the adoption of PDMPs (Prescription Drug Monitoring Programs) in 49 states, the US still faces a growing public health crisis -- opioid misuse was responsible for more deaths than cars and guns combined in the last year -- and lacks the high-resolution data needed to implement new solutions.
While we were initially motivated to build Medley as an effort to address this problem, we quickly encountered another (and more personal) motivation. As one of our members has a chronic pain condition (albeit not one that requires opioids), we quickly realized that there is also a need for a medication and symptom tracking device on the patient side -- oftentimes giving patients access to their own health data and medication frequency data can enable them to better guide their own care.
What it does
Medley interacts with users on the basis of a personal RFID card, just like your TreeHacks badge. To talk to Medley, the user presses its button and will then be prompted to scan their ID card. Medley is then able to answer a number of requests, such as to dispense the user’s medication or contact their care provider. If the user has exceeded their recommended dosage for the current period, Medley will suggest a number of other treatment options added by the care provider or the patient themselves (for instance, using a TENS unit to alleviate migraine pain) and ask the patient to record their pain symptoms and intensity.
How we built it
This project required a combination of mechanical design, manufacturing, electronics, on-board programming, and integration with cloud services/our user website. Medley is built on a Raspberry Pi, with the raspiaudio mic and speaker system, and integrates an RFID card reader and motor drive system which makes use of Hall sensors to accurately actuate the device. On the software side, Medley uses Python to make calls to the Houndify API for audio and text, then makes calls to our Microsoft Azure SQL server. Our website uses the data to generate patient and doctor dashboards.
Challenges we ran into
Medley was an extremely technically challenging project, and one of the biggest challenges our team faced was the lack of documentation associated with entering uncharted territory. Some of our integrations had to be twisted a bit out of shape to fit together, and many tragic hours spent just trying to figure out the correct audio stream encoding. Of course, it wouldn’t be a hackathon project without overscoping and then panic as the deadline draws nearer, but because our project uses mechanical design, electronics, on-board code, and a cloud database/website, narrowing our scope was a challenge in itself.
Accomplishments that we're proud of
Getting the whole thing into a workable state by the deadline was a major accomplishment -- the first moment we finally integrated everything together was a massive relief.
What we learned
Among many things: The complexity and difficulty of implementing mechanical systems How to adjust mechatronics design parameters Usage of Azure SQL and WordPress for dynamic user pages Use of the Houndify API and custom commands Raspberry Pi audio streams
What's next for Medley
One feature we would have liked more time to implement is better database reporting and analytics. We envision Medley’s database as a patient- and doctor-usable extension of the existing state PDMPs, and would be able to leverage patterns in the data to flag abnormal behavior. Currently, a care provider might be overwhelmed by the amount of data potentially available, but adding a model to detect trends and unusual events would assist with this problem.