The Roanoker - July/August 2018 - 71

D A T A I N H E A LT H C A R E

costly hospital stays. Care can be preemptively handled in outpatient clinics,
by home health teams or with telemedicine. Predictive analytics can zero in
on patient populations not only within the hospital, but also at the zip code
level within the community.
Using funding for smarter investments in the business of health care is
one way a nonprofit like Carilion can ensure spending is done wisely and can
impact those with the greatest or most pressing needs.

READMISSIONS

CMS has set a national goal of cutting down patient readmissions. Carilion
is using big data to determine which patients are at high risk for remission
to the hospital once discharged. Teams of medical experts are developing an
algorithm, using machine learning in real time, to better flag patients with
a higher risk of readmission. Those at a higher risk of readmission will have
resources such as patient education, social workers and home health care
providers allocated, with the common goal of keeping the patient healthier,
better informed and treated within the appropriate setting.
Carilion currently has a major focus on addressing the opioid crisis. Part of
this is using modeling with third-party data for patients prescribed opioids.
More robust data means a multidisciplinary team; lead by Dr. Robert Trestman,
they can use a patient-focused approach targeting problems identified by the
data systems, including flagging at-risk behaviors, locating geographical hot
spots and enhanced reporting for prescription monitoring. Opioid data is
also compiled to assist with research funding. Patients considered at-risk for
overutilization of opioids are tracked, with actionable plans provided to their
health care provider in real time.
In 2017, the state mandated the Emergency Department Information
Exchange (EDID), a standardized system to share patient information between
hospital systems. The software is designed to proactively notify Emergency
Departments when high-utilization or special needs patients register. This
will allow better coordination of care for these patients. Patients frequenting
emergency rooms seeking controlled drugs are now tracked with a goal of
providing the information in real time and sharing it among all emergency
rooms. Given our nationwide opioid crisis, patients seeking narcotics at
different locations can be flagged with the goal of decreasing this behavior.

IMPROVING HOSPITAL CARE

Predicting which patients are more likely to fall is giving providers better
information to educate and assist patients and their caregivers. Combining data
points in real time, a patient is assigned a risk score and suggested, proactive
measures that can be taken to mitigate the risk of a fall.
Machine learning can also help pick up on subtle patient changes before
they become life-threatening. A tool which has been in place for the past year
at Carilion Roanoke Memorial Hospital provides real time early warning
alerts to staff when a patient's status is deteriorating. This surveillance sends
out easily understood, color-coded graphical charts to medical staff which are
monitored on every unit. This intelligence also adds continuity of care and
facilitates better communication during shift changes and handoffs.

COMMUNITY INVOLVEMENT

Carilion is partnering with multiple schools at Virginia Tech to use data
to better understand how to address our community's health needs. One
project involves combining EHR data, publicly available school systems data,
economic data and social determinants to develop best practice risk models
and treatment plans to target appropriate care for high risk communities. Dr.
Morgan states, "Understanding our patients' social and economic challenges
is key to improving the health status of the areas we serve." I
TheRoanoker.com

JULY/AUGUST 2018

71
http://www.TheRoanoker.com
The Roanoker - July/August 2018

Table of Contents for the Digital Edition of The Roanoker - July/August 2018
