patientsafety.select

• 

  Board along with box.

Inspiration

I read about the work that patient advocates have done in short-handed medical settings along with the need to having "fail-safes" when delivering medication. We knew how to use arduino hardware along with GPT, so we applied it to the issues.

What it does

Our project has two components. The AI Patient advocate is accessed through a web interface, where you can chat with it and it will provide meaningful advocacy for your medical interests as you desire. It will ask you some starter questions to gain information. Then you can ask it questions about medical information posed to you by medical professional. It's meant to be a supplementary source for advocacy. Second component is the Arduino based double check system. The medicine drawers are adjacent to an entry computer. When refilling the box, you first input the drug name, quantity, and box number, checking over all of the drugs before placing inside. When removing medication from the box, you input the drug name, and quantity on the computer, where it will light up your appropriate box. From that, you can read the label and cross reference the LCD screen with the drug you intended on using. It's meant to be a rudimentary alternative to eMAR systems trading lower overhead for needing more human attention.

How we built it

We build it using OpenAI's APIs. For the hardware portion, we used an Arduino Nano, LEDS, breadboard, Processing lang.

Challenges we ran into

We did not start early into the hackathon so we had less time than we would have liked. Consequently, we had some elements of the project that were not in full completion of our MVP goal.

Accomplishments that we're proud of

We created two inexpensive patient safety mechanisms that have potential to be expanded upon.

What we learned

We learned about implementing GPT in a fullstack project along with the medical context for building projects with hardware.

What's next for AI Patient Advocate + Medicine Dispenser Double Check System

We can test the AI Patient Advocate with patient scenarios to refine its capabilities along with enhancing the redundancy of systems for the Medicine Dispenser Double Check System.

IIV Points

In a lot of medical settings in the United States and abroad, of a variety of sizes, there exists a need for cheap ways to increase patient safety. We have a web based medium along with a physical solution to create a wide range of impact. For the hardware, compared to eMARS and other digital systems there is a lot of overhead. A requirement of internet connection, physical lockers, a database of medications. Our system requires itself and a computer. It could be improved to be more competitive, but it has potential to be more robust in terms of durability and reverification of double checks. This is a significant improvement in that its dirt cheap. The total cost of the system minus the computer which is required for all of them, is under $50. For the AI Patient advocator, it's an enabling tool that helps the patient help themselves. There aren't solutions at the moment that do what we have aimed to do.

Multiple stake-holders benefit from our project. We improve both the patient side and medical professional side of healthcare. We increase safety for patients through drug delivery along with helping their biggest advocate, themselves.

There are risk with both systems that can be resolved with further progress. For instance, the issues with human error and negligence can be solved by implementing some sort of cage or lockbox integration along with a compartmentalized dispenser. It would make the system more expensive, but is meaningful to a majority of healthcare settings. It would most likely take a few weeks in order to flesh out the patient advocator along with the drug dispensing system. They need a lot more work. Measures of success could be established through comparing long-term results for deaths and complications. These systems would work together to reduce those numbers over time.

NOTE: Two of our teammates bailed, so we don't have the code for the AI Patient Advocator. We have a design prototype for the hardware. Code is WIP.

Built With

• ai
• arduino
• gpt