CoronaVoice

Voice Inflection Spectrogram
Graph of Past Symptoms

Inspiration

As COVID-19 continues to grow rapidly around the globe, the stock of testing supplies begins to dwindle, and hospitals are overloaded with patients. CoronaVoice was created to ease the load on health care workers, while also making it possible to detect potential cases of COVID-19 earlier and more efficiently.

What it does

With CoronaVoice, our ultimate aim is for doctors to be able to remotely monitor patient’s vital signs, and to flag them for testing if necessary. Our website uses a voice inflection test to detect if there are significant differences in a patient’s voice, indicating potential sickness/irregularity in health. The impact on healthcare workers and the spread of the virus will be significant; healthcare workers will be able to test a streamlined patient pool, and COVID-19 can be detected earlier, limiting the spread. In the workplace, this technology can be implemented as a health check for returning employees, protecting everybody's safety by preventing ill individuals from entering the building.
Along with this, CoronaVoice allows users to post daily logs of important medical data, including their body temperature, heart rate, and other symptoms relevant to the Coronavirus. These logs are instantly displayed in a series of graphs for a quick view of how your sickness is progressing. If you ever find yourself in the ER, the data gives doctors an accurate and fast patient history, lessening a burden from their normal patient processing.

How we built it

We built CoronaVoice on a Node.js server hosted by Google App Engine. Registration information, symptom history, and baseline audio recordings are saved in a MongoDB database. For the symptom logging page, Chart JS was used to construct a visual representation of temperatures and heartrate over time, and the command line tool 'Sox' was used to generate spectrograms on the classifier page.

Challenges we ran into

Many of us were unfamiliar to MongoDB, Google App Engine, and many of the other packages we used to make this project. It was a challenge to learn how to use these tools and implement them on the webpage in this short amount of time.

Accomplishments that we're proud of

We're proud of creating a fully-functional CoronaVoice webpage within the span of this hackathon.

What we learned

We've become well-versed in using Google App Engine along with Node.js, HTML, CSS, and MongoDB. It was a rewarding experience to have learned how to successfully integrate all of these different technologies.

What's next for CoronaVoice

We aim to conduct more researching regarding the inflection threshold and to make our flagging system more accurate. In addition, we would like to add a visual feature, so our system can potentially detect visual cues of COVID-19, flagging people even earlier.