SkinSense

What it does

Our solution prioritizes Global Health, Patient Outcomes, and Safety. By leveraging advanced image recognition with YOLO, SkinSense can accurately screen patient-uploaded images to differentiate various skin conditions, with a primary focus on identifying Monkeypox. Our main focus relies on whether the image is MonkeyPox or not, and then our two trained models allow us to further examine the skin condition. This empowers individuals to seek timely medical guidance, enhancing healthcare accessibility and supporting more effective public health responses. Physicians can also benefit from SkinSense by using it as a tool to quickly narrow down potential diagnoses, streamlining patient care and improving clinical efficiency.

How we built it

Our front end is developed with HTML, JavaScript, and React, making it user-friendly and accessible. On the back end, we use Python for image processing and AWS for cloud storage, ensuring our system can scale seamlessly. We rely on YOLO and Llama models for identification, providing high accuracy in real-time.

Additionally, our platform integrates Google Places and geolocation services. This way, once a condition is identified, SkinSense can guide users to the nearest healthcare provider, turning our solution into an actionable healthcare tool.

Accomplishments that we're proud of

We are a team of 9 and we're proud of how we've collaborated together in the past 48 hours. We switched between three different application pitches throughout the hackathon and were able to successfully combine all of our ideas in the end.

What's next for SkinSense

SkinSense is designed to be both scalable and practical for real-world use. Built on AWS, it can handle large datasets and manage multiple requests simultaneously. This makes it suitable for individual use and potential integration with healthcare providers or public health organizations.

We’ve also built SkinSense with future growth in mind. With the implementation of geofencing services, users can remain informed about outbreaks near them.

The future of SkinSense is in time progression analysis and generation. Our models can be robustly trained with time series sequence techniques such as transformers and recurrent neural networks to track patient photos sites over time to ensure healthy healing. We can use image generation technology to generate images of what patient sites should look like.