Main Responsibilities Of A Data Scientist At Cohere Health
Shailja, a Data Scientist at Cohere Health, builds machine learning algorithms that automate healthcare prior authorizations, significantly reducing the three-to-five-week waiting period for patients needing complex treatments. This involves "analytics and data" analysis to identify opportunities for automation, and directly creating the predictive models that automatically approve or flag cases for manual review, ensuring no patient is denied care.
Machine Learning, Healthcare, Data Analysis, Predictive Modeling, Algorithm Development
Advizer Information
Name
Job Title
Company
Undergrad
Grad Programs
Majors
Industries
Job Functions
Traits
Shailja Somani
Data Scientist
Cohere Health
Johns Hopkins University, 2020
Currently pursuing my MS in Applied Data Science at the University of San Diego (part-time online while working full-time)
Psychology
Technology
Data and Analytics
Greek Life Member, LGBTQ
Video Highlights
1. Shailja works on automating prior authorizations in healthcare using machine learning, reducing the time it takes for patients to get approval for treatments.
2. She develops machine learning algorithms to predict whether a patient should be approved for a procedure, aiming to automate approvals while ensuring no patient is wrongly denied care.
3. Her work involves data analytics to identify opportunities for improvement and optimize the algorithms used for prior authorization automation.
Transcript
What are your main responsibilities within your current role?
My company automates prior authorizations in healthcare. When a patient needs a long or complex treatment, like major surgery or extended physical therapy, their doctor submits a prior authorization request to the insurance provider.
Currently, insurance providers take three to five weeks to manually review the paperwork, including lab results and patient history. Nurses and doctors read this information to decide if the patient is covered.
We use machine learning algorithms to determine if a patient can be covered. If a patient can be approved, we automatically approve it. If not, we pend the request.
We never want a machine to deny care. If our system can't approve a request, it's sent through the normal process with nurses and doctors. Many patients receive automatic approval for necessary treatments like surgery, significantly cutting down the waiting time.
My daily tasks involve analytics and data to identify opportunities and gather information on different procedures and doctor's offices. I also work on predictive modeling to create algorithms for automatic patient approval.