Medispectra

Identifies GI cancer from data acquired from a non-invasive internal scan and tells you if the patient has cancer or any other medical condition in the GI tract.

Comment

Inspiration

There have been over 2 million cancer diagnoses in the United States in 2024 alone. This number has been increasing over time over the years due to more screening tests being performed. While these advancements have been beneficial, there are still many flaws to them and America's healthcare system as a whole. To undergo a diagnostic procedure, such as a colonoscopy, costs thousands of dollars. Additionally, many diagnoses are done by the time the cancer has persisted for a long time, and the patient's life is put further at risk.

What it does

Our project involves a hypothetical camera pill, which a patient would ingest. As the pill travels through the gastrointestinal tract, it takes a scan and sends thousands of photos back to an artificial intelligence model. This model has been trained to classify images based on the medical condition they depict. It tells the patient if they have cancer or not, and it can identify other gastrointestinal problems such as inflammation or precursors of cancer. Our hope is that this will be used by small healthcare companies and pharmacies to provide less expensive diagnostic procedures to patients who cannot afford the conventional options.

How we built it

We first built a simple UI for Medispectra on Figma and began to develop a ML model using SageMaker. On FlutterFlow, we made the actual app.

Challenges we ran into

Initially, it was difficult to find a dataset large enough to run such a program on. It was also challenging to train the AI model to label images with different illnesses or patient statuses.

Accomplishments that we're proud of

We are proud that we were able to build something that seemed so complex in such a short amount of time. This product is something that can help patients nationwide and potentially globally. We feel that we have taken a step toward creating more affordable and accessible healthcare.

What we learned

Creating and testing a ML model were some of the new skills we learned.

What's next for Medispectra

Our future goal for Medispectra is that it can be implemented for identifying more types of cancer and health problems in other parts of the body. This will help patients get faster diagnoses at a low cost which would greatly benefit our country's healthcare system. We also hope that Medispectra can be implemented as a healthcare platform in third-world countries, since one of our main goals with it is to make healthcare more affordable. Many patients in third world countries go unnoticed due to lack of accessible medical help.

Built With

Share this project:

Updates