After training the machine learning model, three testing images were given to model and it accurately identified the skin lesions.
Google API used to locate nearby healthcare services to allow users to have a choice outside of Healthcare Hub.
A community-driven forum that allows everyone to ask questions and get answers from professionals in the community.
This platform also provides a free outlet for professionals to share their medical journals and articles to educate the public accordingly.
This platform can also give health journalists a platform to share their advice and findings.
The U.S. has the most expensive healthcare system of all the developed countries in the world. Additionally, the cost of healthcare is drastically unaffordable compared to the average working American income. We wanted to provide a platform for the community to help each other in every way possible and also include the development of machine learning to the community's advantage.
What it does
Healthcare Hub is a website platform with multiple capabilities:
AI Approximate Diagnosis - Users can upload a picture of a visible health issue and the AI can make an accurate conclusion. As of now, the model can identify and differentiate between 7 different skin lesion types: (bkl, akiec, mel, nv, df, vasc, bcc).
Healthcare Locator - Users can find the nearest healthcare service buildings and clinics.
Community Forum - The public can ask questions concerning any issues they have and professionals in the community can get back to them with insightful advice.
Medical Journals - Provides a free platform for medical professionals to share their findings and research, educating and informing the public.
Health Articles - Provides a free platform for health journalists to post their findings and conclusions based on social research.
How I built it
I created a Python sorting program that sorts very large directory of images into separate directories according to a CSV as reference and the numpy and pandas libraries. I also used Teachable Machine with Google to train a machine learning model to identify and differentiate the 7 different types of skin lesions.
Google Maps API.
3, 4, 5. HTML/CSS.
Challenges I ran into
Turning our ideas in a short amount of time was difficult because we had so many features we wanted to include for users but we didn't have that much experience with creating such a project. We used the Agile development method to work in sprints so that we could get our ideas to come to life. For the AI aspect, we needed a lot of data to train the model accurately, but the Teachable Machine could not hold as much training data as we would like. We decided to use as much data as we could, to make the most out of it.
Accomplishments that I'm proud of
We are proud of completing what we did in the amount of time that we had and the amount of experience we have. We are fairly new to hackathons: 3 of us being first-time hackers and 1 of us being a second-time hacker. We think our idea has a lot of potential and we're proud of our collaboration skills, considering that we all met each other about 24 hours ago.
What I learned
This is our first time executing such an idea, and we learned a lot by doing, failing, and trying again. We also learned a lot about what it takes to develop a business model that could benefit all users and contributors, and the time, effort, and knowledge needed to develop functional and usable technology. It all comes down to understanding the users needs, time management, and effective communication within a team.
What's next for Healthcare Hub
I personally hope to learn more about using Tensorflow for machine learning so that I can train models with larger amounts of training data. Then, I can come back to this idea of AI Diagnosis and hopefully develop a more powerful program. I believe the concept of AI diagnosis could be used with even more branches in the medical industry such as: CT scans, MRI scans, X-Rays, and even cellular image analysis. At the end of the day, I want to use technology as a way to move society further and help everyone.