LymeML

Inspiration

Research suggests that over 14% of the world's population has been affected or is currently being affected by Lyme disease, a consequential illness that requires an extensive diagnosis and the need for access to a doctor. However, with billions of individuals living in remote areas without access to a medical professional, it seems that little can be done to provide clinical assistance. But, with the use of novel machine learning algorithms that implement convolutional neural networks, an effective clinical diagnosis is available to all. It is important to effectively diagnose Lyme disease since a timely and accurate response is needed to prevent complications in the brain and heart infections. With LymeML, countless people will have access to an effective telemedicine solution.

What it does

The app requires users to input an image that runs through a convolutional neural network to diagnose Lyme disease. Next, a symptoms checker is required to ensure that the user is undergoing the symptoms of this illness. By simultaneously using data from both the image classification algorithm and symptoms checker, this application returns an effective clinical diagnosis.

How we built it

I built this app in MecSimCalc by creating an Upload File button for the user to input an image while also including a checkbox to ensure that the user has given us consent to use the image in the model. We used libraries such as NumPy and TensorFlow to diagnose the disease and return an effective and accurate result instantly to the user. To ensure that our app does not solely rely on the machine learning model, a symptoms checker is also used to simultaneously confirm the case.

Challenges we ran into

Since I used a machine learning model for image classification, there is not 100% accuracy that the rash is one produced by Lyme disease. For this very unlikely case, sometimes the "submit" button needs to be pressed more than once to return the accurate result. In order to decrease the chances of this issue impacting others, a symptoms checker is also run on the app.

Accomplishments that we're proud of

I am proud that I was able to create an application that can impact millions to billions of individuals around the world and create a solution to a global issue I was inspired to solve. Using MecSimCalc technology, the widespread and contemporary solution will provide beneficial results to many. I am also proud that I integrated machine learning into the solution as it will provide for an immediate and effective through implementing it in this app.

What we learned

Through the journey of creating this app and the challenges I faced along the way, what I learned throughout the journey is most important to me. First, I learned how to integrate machine learning in a mobile and web application as a solution to Lyme disease around the world. Additionally, I tested my ingenuity and creativeness while coming up with methods of creating the app, which is evident through the symptoms checker and image classification simultaneous evaluation. I also learned unique Python techniques that I was unaware of while coding this application. Overall, I am proud of the extent of my learning through creating LymeML.

What's next for LymeML

In the future, I want to publicize LymeML so it is well-known and accessible to countless more people. Additionally, LymeML will include a more accurate machine learning model and more symptoms will be evaluated to diagnose Lyme disease. I will improve the user interface of this app to make it more appealing and likely, in the future, provide machine learning solutions to other diseases that many face in remote conditions. I would also like to make this app available on iOS or Google Play for others to test out.

Built With

• back-end
• cnns
• front-end
• machine-learning
• mecsimcalc
• mobile-app
• neural-networks
• python