Inspiration------------------------------------------------------ In the United States, the number of specialist doctors per 100,000 people in rural areas are only around 30. In the case these doctors were sick, about to retire or even about to die, there isn't going to be someone that can provide medical assistance. Most of these citizens would need to travel to the city for medical attendance, bringing an economic imbalance. That's why rural citizens are the most affected with today's pandemic. The limitation of medical help, causes the easy propagation of the virus through the community. We believe that rural doctors should be provided with the same tools as urban doctors. Through AI we can manage to give general doctors the tools that medical specialists have to have a more accurate and fast diagnosis.

https://www.ruralhealthweb.org/about-nrha/about-rural-health-care

What it does--------------------------------------------------- Our services provide inexpensive virtual diagnostic services with the help of x-ray and deep learning, to analyze the x-ray pictures and provide a possible diagnostic for the patient. This can provide another hand to the service of the rural doctor. Through the video service and the AI tools, rural medics can have a greater reach in regards to disease they can diagnose. And in times of Covid-19, they can still diagnose patients virtually without risking their lives by seeing them in person.

How we built it----------------------------------------------- The program consists of two parts. The first one is the Deep Learning algorithm that classifies x-rays in piles of "infected by Covid-19" or "normal". This was done with tensorflow and keras and the whole thing was coded in python. We created a CNN and fed it 1000 medical images out of which half were infected with Covid-19 and the other half was not. Our results were really good, reaching accuracy rates of around 95% and losses around 0.008%. The second part of the project is the video calling system. This was done with javascript, node.js, and WebRTC. We created a web video platform that creates private video chat rooms with unique IDs so that medics can connect with their patients through unique links and therefore ensure security and confidentiality throughout the patient-medic interaction.

Challenges we ran into----------------------------------- Our first challenge was identifying what rural medics required. We wanted to provide a service that could aid anyone without regard to their economic status or nationality. We came up with this idea of decentralized healthcare in order for medics anywhere to aid patients everywhere, with tools to make their online diagnosis more precise. Our second and last challenge was avoiding overfitting in our dataset so we had to tune our parameters and do multiple training rounds with different batches from the dataset as well as checking our results on external datasets.

Accomplishments that we are proud of------------- We are proud that we created an efficient code which can help many lives in rural areas. We are proud to create a peer-to-peer video connection between two individuals anywhere with secure communication and stable audio and video. We are also proud to provide tools to medics so that they can improve their diagnosis and extract extra information from medical images.

What we learned------------------------------------------- We learned about the importance of utilizing our advanced knowledge of technology to help those people in need. We noticed that programming can bring a huge help anywhere in the world, to find different solutions to problems that have been present since the start, like poverty. We learned how hard it is to create a medical algorithm since there are lives at stake. Most importantly, we learned that in less than 2 days we can create something that has a social impact and that we are extremely excited to continue on,

What’s next-------------------------------------------------- We hope that we can bring this system to everywhere. We want to keep on adding more uses such as cancer detection. We want to provide medical assistance to undocumented immigrants who can not go to the hospital without risking their livelihood in the US. We believe that doctors from Latin America could aid these undocumented immigrants and provide them with the diagnosis they require while avoiding imprisonment. All online.

Built With

Share this project:

Updates