Natural disasters happen all the time and when there’s no power or communication, it’s hard to get help. This is what inspired our Respond application, the need to find and inform about people who need help. Coming from Houston’s recent hurricane Harvey, and then the ongoing Hurricane Florence, one of our team members knows what natural disasters are like. We designed Respond to meet everyone's needs in order to better reach people and provide help.

What it does

Respond helps first responders survey the severity and distribution of those in need of help after a natural disaster. Using drones, Respond can use audio and visual input to 1) rank severities of different patient situations, 2) present this information and audio messages to first responders, and 3) create a variety of options for the first responder to make sure everyone is helped.

How we built it

We first used the RevSpeech speech to text api to translate audio files into sentences. We then used sentiment analysis to break down these sentences into their principle emotional components: sadness, fear, anger, joy, and disgust, and overall sentiment. We extracted these features from several manually created training examples to train a neural network capable to assigning urgency levels to sentences.

Challenges we ran into

On the backend, we ran into challenges when creating a quick and accurate model for machine learning. First responders need the information about severity before all else, so they can know which patients to prioritize; thus, speed was a huge priority to us when creating the neural network.

On the design side, we took a lot of time to empathize with first responders and figure out exactly what information they would need and when they would need it. Ideating and creating a way for first responders to take in large sums of detailed information on mobile screens was definitely a challenge, and was one we overcame by placing ourselves in the first responders’ boots.

What's next for Respond

We hope to "respond" to judge and peer critique and be able to improve the app on future iterations. Respond has the potential to save first responders a lot of time and brain power when facing not only logistical, but also moral dilemmas. Next time an emergency official gets a call about someone who might need help, someone else will always Respond.

Built With

Share this project: