A Precision Medicine Recommender System for Inflammatory Bowel Disease: a pilot study using real electronic health records data from UCSF

Partner: Vivek Rudrapatna, University of California, San Francisco, Academic

Overview

Project Description

The idea of Precision Medicine is to move beyond our one-size-fits-all healthcare system and towards one where we make data-driven treatment decisions that take into account individual factors, like patient demographics, genetic background, and medical history. With the advent of Electronic Health Records data, we are now closer than we’ve ever been in terms of realizing this vision. What’s needed now is the right combination of talent and grit to develop these recommender systems and iterate to make them useful to treating physicians.

In this project, you’ll be helping build the very first recommender system for clinicians treating IBD. Together we’ll mine through EHR data, find covariates, and build a simple web app prototype to help clinicians pick a treatment for their patients based on similar patients and their outcomes as captured by the database. You’ll be working closely with a gastroenterologist/data scientist who will get you the data, get you the computing platform, and help you think through the machine learning and design specs needed to help this project succeed.

Expected Deliverable

The deliverable would be a web app (R Shiny, Python Streamlit/Flask) where we can demo the capabilities of the app using a combination of real EHR data as well as some synthetic data where necessary.

What would a successful semester look like to you?

If this app were working (no bells/whistles) by the end of the semester it would be a tremendous success – we’d write it up, show it off, iterate and eventually productize it.

Data

Models

Conclusion