A team member's grandmother was in the hospital recently, and he recalled that the hospital process was long and tedious. Nurses had to produce a chart pertaining to the patient's medical information and conditions prior to the treatment. In recent times, the charts have become electronically available, which greatly reduced the time involved in the process. The idea came through an organic thought process in which we wondered how to further reduce the time involved in the patients diagnosis and treatment. In a hospital, minutes and seconds can be valuable. The facial recognition aspect of the app can procure results within seconds of sighting the patient.
What it does
MedHUD pulls patient data inputted by nurses once and puts it in a heads up display (Unity) by using facial recognition with OpenCV, cutting document preparation time for patients from several minutes to several seconds. A Flask server takes a photo taken from the AR application and matches it against a set of images taken by our Meteor app. If there's a match, returns the rest of the patient data that the nurse input to our Meteor app. Our HUD (built with Vuforia) shows data such as height, weight, age, biological sex, and so on, with family history of certain conditions, and whatever drugs have been prescribed to the patient in the pass. Rather than several minutes for each of millions of patients, MedHUD resolves this problem by returning this data in seconds.
How I built it
Our AR environment was built in Unity, with Vuforia for the Heads Up Display sending data. Flask managed our OpenCV management and communicated with our Unity app, and Meteor.js and Mongo managed our patient data, and communicated with Flask to tsend the data that Flask sent to Unity.
Challenges I ran into
We were having a lot of trouble setting up Unity. Like, a lot.
Accomplishments that I'm proud of
Finishing this project. This was no small feat that really did come along in the last minute.
What I learned
How to build an AR app, and more importantly, why not build an AR app.
What's next for MedHUD
We'll see :) .