OncoAI

Inspiration

We wanted to tackle issues of inequality in healthcare. Breast cancer is an issue that disproportionately affects people based on race and economic standing, with factors outside of people's control affect their ability to get a diagnosis and progress towards treatment. OncoAI strives to close the gap caused by these social issues and promote a holistic diagnosis of breast cancer.

What it does

OncoAI makes the job of healthcare professionals easier by providing a tool that aggregates and analyzes patient data efficiently to provide feedback to the professional. We analyze data that would require pathologists, geneticists, radiologists, and hematologist to analyze manually. Thus, we streamline a month long process involving four different doctors into a couple minutes with just one click.

How we built it

OncoAI takes in patient data (blood tests, mammograms, whole slide images, scRNA-seq data), processes it, and puts it through AI models before using a Large-Language Model to compile the data into one comprehensive report on the patient, streamlining diagnosis and treatment. Patient records are then encrypted and stored in a database that strives to be HIPAA-compliant in handling patient data.

Challenges we ran into

Some challenges that we ran into were deploying AI models on local infrastructure, formatting model output, and encrypting our data. Additionally, we struggled to choose which frameworks we wanted to use initially, which took up some of our time.

Accomplishments that we're proud of

We're proud of having integrated a wide variety of data into a novel architectures and or successfully deploying AI models that address combinations of data that haven't been explored previously.

What we learned

We learned to decide on development platforms early on in the development lifecycle to minimize confusion and maximize productivity! Additionally, we learned a lot about both front-end and back-end development and why it's important to have constant communication and collaboration between the people working on the two, as it's difficult to adapt code later on in the process.

What's next for OncoAI

We hope to continue improving our product, especially in its scalability and by integrating our platform with existing hospital systems; fleshing out back-end development; and deploying our platform using Google Cloud. We also wish to expand the data types our model incorporates and take additional steps to reduce inherent model bias.