An interest in the growing market-led drive towards oncological study and tailoring of chemotherapy for leukemia patients.
What it does
Tracks blood oxygen saturation levels, subjective fatigue, dosing, and administration timing of re-targeted dosing parameters for Venclexta. These data allow for the construction of global and individual predictive RNN LSTMs to derive tomorrow's targeted dosing needs from today's reactions.
How I built it
Constructed an Android app to (a) ensure security of data availability to the patient throughout the PMOS lifetime, (b) push daily notifications to collect subjective fatigue input in relation to the drug dosing regimen, and (c) utilize photoplesymography (PPG) sensory input to interpolate blood oxygen saturation levels.
Challenges I ran into
Data availability was sparse. Very few individuals have app development history dealing with PPG sensory input and thus much of this was innovative. Constructing a tri-parameter LSTM without reference to human genomics and other clinically-gathered data disallowed high accuracy in the construction of the RNN for better prediction of tomorrow's dosing needs. More data sharing between clinic and publically available baseline datasets..
Accomplishments that I'm proud of
We are proud of our ability to bring various skills to the table and construct a cohesive story from different forms. Our individual interests led to great group collaboration and our approach schema worked well to capitalize on our personal and group strengths.
What I learned
PPG sensory data is difficult to capture via current technologies. Leukemia patient outcomes are directly correlated to blood oxygen saturation level.
What's next for V-Connect
The connection of our predictive model to clinical genomics will allow for a push towards a duopa-like experience with individualized, targeted dosing regimens which will alleviate fatigue and lead to better persistence throughout the leukemia maintenance phases of Venclexta regimens.