Inspiration
When we started this project, we were interested in researching cancerous diseases. We were trying to find gaps in current research, which is when we stumbled into multiple myeloma. Multiple myeloma - a cancerous disease - takes the lives of over 12,000 people annually, creating the need for a software-assisted approach to aid in its diagnosis.
What it does
Our software combines serum protein electrophoresis and immunofixation tests and analyzes these results to provide the user with a coherent and understandable diagnosis of multiple myeloma. This software helps create significant advancements in the diagnosis of this disease, with revolutionary time efficiency and accuracy, to help save over $8,000,000 annually.
How we built it
To build this software, we first broke down an expert's workflow into 3 separate parts. This involves taking note of the patient's protein levels, analyzing the serum protein electrophoresis and immunofixation graphs, and giving a diagnosis. Each of these parts had their respective algorithm, which was developed in Python using a variety of libraries to help speed up the workflow.
In addition, a neural network was developed to identify abnormalities to further aid in the diagnosis of multiple myeloma.
Challenges we ran into
Some challenges we ran into were collecting data for training the neural network, understanding the expert's workflow, and understanding the lack of previous knowledge on this subject.
Accomplishments that we're proud of
We are proud of making this novel software applicable and beneficial to real-life scenarios to help the lives of hundreds of thousands of people each year by diagnosing multiple myeloma.
What we learned
We learned the process behind electrophoresis scans, blood drawing, and how things work in a clinical environment. We learned about the scientific process that could be applied to real-life scenarios, and we also learned how beneficial AI and software-assisted approaches are to analyzing cancerous diseases. We also learned a lot about the programming that goes into developing these algorithms and AI software.
What's next for SPIF: A Novel Software for Diagnosing Multiple Myeloma
Next steps for this project include:
- Further developing this software to be usable in laboratories
- Implementing this software in conjunction with existing software in laboratories
- Gathering more data to help make this software coherent and accurate across all laboratories to make it generalizable
Log in or sign up for Devpost to join the conversation.