Now more than ever, healthcare organizations are tasked with delivering better patient care. But for many reasons, including worker shortages, healthcare systems are having to improve efficiency and effectiveness with the same resources. A growing trend is to introduce an Internet of Things (IoT), artificial intelligence (AI), or machine learning (ML) solution to streamline staff workloads and address other common challenges. But these solutions can be expensive and risky if they fail to deliver. The many organizations that have decided the reward is worth pursuing are adopting creative solutions to mitigate the potential risk.
For Americans living in rural areas accessing high quality healthcare is a challenge and nowhere is this truer than in Appalachia. Many in Appalachia face barriers to seeking preventative healthcare and the population also contends with some of the nation’s highest prevalence rates for lung cancer, diabetes, and heart disease, higher infant mortality and lower life expectancy. These health issues are compounded by limited access to care courtesy of the lack of transportation and distance to healthcare centers, and by poverty and unemployment rates well above the national average. Even when patients have access to care, appointment cancellations and no-shows are common.
To support their constituents in accessing care and support the continued presence of high-quality medical services in Appalachia, the Appalachian Regional Healthcare System (ARH) decided to implement an ML-powered solution to improve patient care, population health, and the hospitals’ operations. The high frequency of missed appointments raises concern for patient health, especially with the region’s prevalence of chronic issues. It also raises revenue concerns. ARH is a not-for-profit organization, and Medicare and Medicaid are its primary payers. The revenue lost to an abandoned appointment is more than just an inconvenience—it affects ARH’s ability to care for other patients and drive initiatives to benefit the health and wellbeing of residents.
To address the issue of no-shows, ARH needed a way to identify risk factors for a patient who was likely to miss or cancel their appointment and identify potential solutions. ARH opted to pursue a machine learning (ML) solution to make use of the data that had already been collected. The solution had to be simple enough that their existing data science team, whose skill sets range from finance and accounting decision support to SQL development, could use and apply it without formal training.
ML models serve two main functions: to find patterns in historical data and to predict future results based on those patterns. In the case of missed appointments, ARH could use an ML model to examine the information on patients who had previously missed appointments, such as what health conditions they had, or demographic information like age and sex. They could then use any patterns identified by the model to identify similar patients. Once the model is established, it can also be used to plan for various “what-if” scenarios by adjusting aspects of the patient population, supply chain, or other factors to see predicted outcomes.
The team chose Qlik’s AutoML and began creating models to analyze potential barriers to attending visits. One notable factor revealed by the model was travel distance. Patients who had to travel a significant distance to their appointments were more likely to cancel. Having identified this as a driver, ARH staff were able to call patients with upcoming appointments and suggest that they visit a closer clinic instead.
By addressing the signals of appointment cancellations identified by the ML model, ARH reduced cancellations and no-show rates at their pilot clinic from 20 percent to 15 percent over a three-month period. The decision to pursue an automated solution also proved successful, as their existing data science team was able to utilize ML’s full capabilities without significant investment in hiring or education. ARH plans to expand the strategy to other clinics in its network to provide more effective patient communication, with the goal of promoting a local attitude of openness toward and trust of preventative medicine.