check out the Devpost at https://devpost.com/software/test-jrqbek
This project won the following prizes:
[Genentech] Most Beneficial to Patients
[SoundHound] Best Use of Houndify Voice AI API (1st place)
No one likes waiting around too much, especially when we feel we need immediate attention. 95% of people in hospital waiting rooms tend to get frustrated over waiting times and uncertainty. And this problem affects around 60 million people every year, just in the US. We would like to alleviate this problem and offer alternative services to relieve the stress and frustration that people experience.
We let people upload their medical history and list of symptoms before they reach the waiting rooms of hospitals. They can do this through the voice assistant feature, where in a conversation style they tell their symptoms, relating details and circumstances. They also have the option of just writing these in a standard form, if it's easier for them. Based on the symptoms and circumstances the patient receives a category label of 'mild', 'moderate' or 'critical' and is added to the virtual queue. This way the hospitals can take care of their patients more efficiently by having a fair ranking system (incl. time of arrival as well) that determines the queue and patients have a higher satisfaction level as well, because they see a transparent process without the usual uncertainty and they feel attended to. This way they can be told an estimate range of waiting time, which frees them from stress and they are also shown a progress bar to see if a doctor has reviewed his case already, insurance was contacted or any status changed. Patients are also provided with tips and educational content regarding their symptoms and pains, battling this way the abundant stream of misinformation and incorrectness that comes from the media and unreliable sources. Hospital experiences shouldn't be all negative, let's try try to change that!
We are running a Microsoft Azure server and developed the interface in React. We used the Houndify API for the voice assistance and the Azure Text Analytics API for processing. The designs were built in Figma.
Brainstorming took longer than we anticipated and had to keep our cool and not stress, but in the end we agreed on an idea that has enormous potential and it was worth it to chew on it longer. We have had a little experience with voice assistance in the past but have never user Houndify, so we spent a bit of time figuring out how to piece everything together. We were thinking of implementing multiple user input languages so that less fluent English speakers could use the app as well.
Treehacks had many interesting side events, so we're happy that we were able to piece everything together by the end. We believe that the project tackles a real and large-scale societal problem and we enjoyed creating something in the domain.
We learned a lot during the weekend about text and voice analytics and about the US healthcare system in general. Some of us flew in all the way from Sweden, for some of us this was the first hackathon attended so working together with new people with different experiences definitely proved to be exciting and valuable.