PREDICTING 30 DAY READMISSIONS
project name

PREDICTING 30 DAY READMISSIONS

description

RESULT – Logistic Regression Model Uncovers Patients with 2 Times Higher Likelihood of Readmission

Predicting 30 day Readmissions
Using Remote Patient Monitoring data

Remote Patient Monitoring
CONTEXT

Tele health platform and services vendor focuse on real-time patient management solutions
Client’s remote patient monitoring program utilizes a variety of technologies to gather biometric data as well information on medication adherence and on self-reported symptoms and behaviors via a remote patient monitoring system.


CHALLENGES


The client was looking for a BI and Predictive modeling solution to analyze biometric, symptom and behavior data to uncover patterns in their data to help predict likelihood of 30 day readmissions.

TOOLS & METHODS
Quantiphi delivered a solution to transform and load a data set of several million readings into an in-memory columnar database and developed intuitive BI dashboards for action ready analysis using Yellowfin BI.

Quantiphi built a predictive model in R that indicated that when patients answered a certain subset of questions in a particular way, they were two times more likely to be readmitted vs. the population average.

Model allows the client to focus on the patients with the greatest risk of readmission and bring down cost of care while improving health outcomes.


RESULT – Logistic Regression Model uncovers patients with 2 times higher likelihood of Readmission

You may also be interested in

contact_small.png