5G brought a lot of advancements, and knowing these we wanted to apply our knowledge to a real-life project that would create change in a way that mattered. We brainstormed to find areas where a small improvement in latency of communications may lead. Here we narrowed down our application to communications in - Formula 1 racing, diagnostics for healthcare, financial data, media & online streaming, mixed reality, to any device tracker system, real-time gaming. Out of which after further research we decided upon diagnostics for healthcare systems where even a small improvement in latency might lead to saving someone from a scenario of life and death. AWS and Verizon's 5G provided the perfect platform for us to build upon this idea.
What it does
The monitoring of patients is still done manually by nurses. The machine beeps an alarm and expects a nurse immediately. The nurse then contacts the doctor. This is a very primitive way and our solution helps the nurses and the doctors from alarm fatigue at times like today where there are too many critical patients. It will also save critical patients' lives by bypassing critical data directly to a doctor or nurse's phone/monitor. One of the main problems we will be combating and the latency thing is helping out nurses with the Alarm Fatigue problem.
How we built it
Our project involves sending data from the patient to the AWS EC2 instance where all the data processing takes place and the results are sent to the doctor.
To do this we created an android application where the patient's side data can be simulated by logging in as a patient and the doctor side can be simulated by logging in as a doctor. The login procedure is offloaded to the AWS-Z side as it is not a latency crucial task. The AWS lambda thus only deals with the patient's data for real-time analytics. After the performing inferencing, the results are also sent to the doctor along with the real-time data. This pipeline was built for the doctor to be able to access the patient's data anytime anywhere and recommend preventative or remedial measures in case there is some anomaly.
The project was built over a short span of 15 days and it was certainly a great opportunity the team had received to apply their knowledge on a real-time 5G network.
Challenges we ran into
The biggest hurdle we faced during this project as said earlier was that none of us had ever worked on any technologies described in the technology stack. So to learn and apply the concepts was certainly a fun challenge. Another challenge was deciding what to substitute the Health Monitoring Machine in an ICU with, so that we can replicate the data and post it to the server. The other challenges faced were the generic undergraduate challenges which were managing our exams and assignments whilst we worked on the project.
What we learned
As an undergraduate team, we did not have much of an idea of how an application-server system works end to end. This project certainly proved to be challenging as each member had to go out of their comfort zone. We had to learn the basics of AWS where the office hours conducted by the organizers certainly proved to be helpful. We also had to learn how to deploy our application on an AWS server, how to create an android application to communicate with the application running on the server.
What's next for a Low-latency health monitoring system for critical patients.
As of yet our system works by simulating data from one patient, performs analyses on it, and sends notifications to the doctor accordingly. In the near future, we plan to add the functionality of monitoring multiple patients and linking patients to their nurses and doctors. Additionally, we will be leveraging the innate features of 5G such as network slicing to send huge amounts of data to the servers on dedicated slices so real-time analytics and inferences can be sent to the doctors for them to take action upon.