October 26, 2015 | Written by: Jeff Goodhue
Categorized: Real-time analytics
Share this post:
As big data abounds from sources all around us, the healthcare industry is poised to act on this data and deliver better service to patients, caregivers, and practitioners. In the United States, there is a dramatic increase in the meaningful use of EHR technology (see HealthIT.gov) along with the need to move from Stage 1: capture and sharing to Stage 2: advanced clinical processes and soon to Stage 3: improved outcomes. To improve outcomes, healthcare must: leverage big data, act in context and avoid delay.
Many healthcare processes require business process management and decision management capabilities. Let’s look at three healthcare interactions and see how leveraging data will improve their outcomes.
Patient appointment reassignment
Mobile connected devices have redefined the way we work, interact and live, so why not redefine the way we receive and provide healthcare? Consider a patient, Sandra, leaving home for her primary care appointment. Based on preference and location, the notification service sends her mobile app a push or SMS reminder. Sandra arrives at the office and has opted in for mobile location access so as she parks her car the office senses the device using Wi-Fi and/or Bluetooth Low Energy (BLE) and sends a patient arrival event which is correlated with events for other patients (a physician starts an appointment and a patient leaves an exam room) to determine room and physician availability. Automatically, rules determine the estimated wait time to see if Sandra’s physician exceeds the threshold set by either Sandra or the physician’s office (45 minutes?) and an offer is sent to Sandra’s mobile device to see an alternate physician or wait as scheduled. She considers the respective wait times, checks her calendar and picks the alternate so she can make her lunch meeting on time. Sandra selects the alternate on her mobile device, the system schedules her new appointment and the clinic schedule is recalculated, all before she steps through the door.
Caregiver request classification and escalation
When IBM Watson first made the game show stage, it astounded audiences with what would later be termed cognitive computing. Now provided as micro services, Watson can be leveraged to bring the power of cognitive computing to applications like the IBM Watson Natural Language Classifier (and many more) that provide services for machine-learning, natural language interfaces. A caregiver, Abu, works with at least 15 patients a week. During his daily rounds, he makes appointments for them to see primary physicians or specialists. Abu only needs a few moments to fill in the request on his tablet and submit it to schedule the appointment.
Using IBM Watson Natural Language Classifier, he can enter his statement or question in natural form, and the service will classify it appropriately; for example by department (neurology or cardiology), to submit for scheduling. From that point, events are correlated together from patient history, and if a pattern of degrading health is detected, then action is taken immediately by launching a case in IBM Business Process Manager to track the interactions and orchestrate them to completion. This all began with a few words written by Abu on his tablet.
Incident coordinator response
Analytics is to data what decisions are to events: a value driver. Bruno is an incident coordinator for a regional healthcare network and leverages multiple data sources to determine the best resource schedule, both human (residents, nurses, physicians) and capital (medical supplies, scanning equipment, etc.). Over the last few years, the volume of data Bruno utilizes to make decisions has exploded: not only weather, traffic and scheduled events, but also twitter trends, nearby hospital admission rates and more. Some data is retained for historical time horizons of minutes while others are retained for months. The data can be easily aggregated for simple conditional logic, such as the number of admissions within 50 miles within the last hour. Finally, real-time event information can be coupled with off-line, weighted predictive models for likelihood of incidents. Let’s see how Bruno puts this all together:
• an increase in rain to the south for the past two days,
• continued humidity for the next two days,
• a spike in flu admission rates in hospitals to the south,
• Twitter trends in the local hospital area with flu discussions.
The above leads to the system notifying Bruno to perform additional analysis and recommend redeployment of medical supplies and residents to avoid a more disruptive and expensive rush order. As appointments begin to spike at the local hospital, physicians are also rescheduled to meet demand.
Dust off your event data
Healthcare is only one industry, and the above are only three examples. As data continues to grow at exponential rates and we access over 1 trillion connected devices in 2015 and beyond, consider utilizing and correlating your data in new and interesting ways to create insight for your customers. Leverage mobile and Internet of Things, invoke cloud-based micro services for cognitive computing, start cases and processes to ensure traceability and utilize analytics inside and outside your infrastructure to predict situations and improve operational decisions.
Download our whitepaper to learn more about how IBM’s business process management and decision management capabilities are helping healthcare companies transform operational decision making.