Hospitals have drastically cut nursing staff over the past decade with millions more patients flowing into the healthcare system. Around 300,000 patients die from medical mistakes, many of which are caused by chronic overwork & staffing shortages. Robots have the potential to minimize busywork for nurses and re-focus hospital attention on patient care. We made MediBot to provide a cost-effective automation solution for hospitals and clinics.
What it does
We provide a fully automated intra-hospital delivery & patient care service to patients in a seamless fashion. The first part of our system is the Amazon Echo Dot + Alexa. Patients (that may be physically disabled) can call for MediBot deliveries. These delivery requests (intents) include: *Deliver Food *Deliver Water *Deliver a Tissue Sample *Deliver the Blood Sample *Call for Help *Bring Nurse *Call Nurse *Get the Doctor *What is my prescription schedule *When do I take my pills
The tasks assigned are instantly uploaded to our express server and processed to be viewed on our mobile application. In our app, we just list out the tasks that have been assigned to the bot.
After the task has been assigned (via bluetooth), the Bot begins to move along a solid colored path. In hospitals, MediBot would have a great impact. Due to the fact that colored lines are used as navigation in hospitals, MediBot would have an easy time navigating around the turns using it's photo sensor to follow hospital lines.
How we built it
The automated intra-hospital delivery & patient care service has three parts: The Alexa commands, a mobile app, and a complimentary delivery bot. The Alexa leverages the Echo SDK and Express to allow for communication within the mobile app based on vocal commands and broadcasting. Once the command has been said, we iterate multiple updating processes to allow for the commands to be shown on the mobile app created entirely in Ionic. Commands then trigger the Bot to begin to move on a solid colored path. This type of sensing and movement is possible using a BLE receiver and transceiver along with 2 360 servos, Arduino Uno, 4 light sensors, a battery pack, and a 3d lasercut caster/wheels. When all 3 parts eventually combined with live data from commands, they truly form a fully automated Voice assisted intra-hospital delivery & patient care service.
Challenges we ran into
We initially wanted to create a dashboard of multiple Bots in a hospital environment, however this was troublesome due to the limited capability React had with animations. This made us move to a mobile Ionic app. We also had to look into getting Light Sensors to work with because IR sensors were really finicky to work with. We had to learn how to use Light Sensors because change in light for IR Sensors did not change significantly when turned on. We attempted to account for these problems as much as possible, allowing us to bring all 3 parts together with a seamless easy experience.
Accomplishments that we're proud of
We were excited when we knew we had a cohesive system, from requests to visualization of these requests, to LINE SENSING with our bot, instead of just one part. It was exciting to even see UPenn Medicine professors excited about our product! We were also proud of creating something that all of us will actually see in practice (we solved a common problem and we hope to change lives significantly with this product).
What we learned
We learned how to use Light Sensors. We also learned how to efficiently sync data across Alexa, our Mobile App, and the Bot. This project was filled challenges that we had to google extensively and debug constantly to get right.
What's next for MediBot
We want to implement a machine learning layer on the bot and application to allow for robust sharing of requests across all users. With NLP features, such as keywords and common requests, we can provide patients with even more commands. We can develop ML models to factor in faster routes within a hospital to reach patients. We also want to provide this product ready out of the box to other hospitals or our own in the near future. MediBot can improve in a number of ways -- we will tackle these tasks generally in order of importance as we see fit.