By Jean Robillard, M.D./Guest Column
Predictive analytics – the use of data, statistical algorithms and machine-learning techniques to identify the likelihood of future outcomes based on historical and current data – has become a part of everyday life, even though we often don’t realize it.
One of the better-known examples of predictive analytics is its use by financial institutions to determine an individual’s personal credit score, but its application is quickly spreading across all economic sectors.
While analyzing data to forecast the future isn’t new to the health care arena, its usage has been mostly relegated to broad scope quantification of things like risk stratification and readmission rates. To ensure better outcomes for patients and promote healthier populations, many in the health care industry are now beginning to understand and harness its greater potential.
For example, UI Health Care is home to one of the earliest uses of predictive analytics in the operating room. Dr. John Cromwell, associate chief medical officer, director of surgical quality and safety, and clinical associate professor of surgery, is the first surgeon to extend the precision inherent in surgery past the procedure, using data analysis to help determine the course of patient care.
Through a variety of different tactics, including predictive analytics, Dr. Cromwell and his team discovered in 2015 that surgical site infections (SSI) for colon surgery patients at UI Hospitals and Clinics had dropped 74 percent in just over three years.
Data was developed and validated on about 1,600 patients, and limited to particular procedures with higher infection rates. From this data, practitioners quickly learned there is value to gain from computational modeling for surgery. They are now routinely, and successfully, bringing machine learning to bedside practice as an early warning system for SSIs—treating patients based on calculated predictions to prevent infections before they start.
Financially, with management of one case of SSI costing upward of $20,000-$30,000 per patient, the benefits of avoiding such complications are tremendous for hospitals and health systems. For the patient, an SSI means pain, disability, financial strain and restriction on activity. Ultimately, that can translate into monetary loss for the patient and the employer if an employee is unable to return to work in a timely manner due to complications that arise post-surgery.
Moving forward, predictive analytics show promise in helping to determine other interventions that could lead to significant improvements in patient care. For example, UI Health Care’s early experience has revealed that a surprising indicator of probable infection is a patient’s home zip code, because it was an indicator of health care access, socio-economic status and health literacy. This valuable information isn’t traditionally accounted for when treating patients and could possibly lead to better outcomes for discharged patients through actions such as home monitoring.
Outcomes data such as this is crucial both to building a model and to guiding the model’s success. As more data becomes available, appropriate models can be built to expand the use of predictive analytics to other specialties and clinical issues. Models also respond to process improvements and are constantly recalibrating.
Knowing this, UI Health Care continues to seek new ways to harness the potential of predictive analytics in health care by engaging with technology companies such as Dell and IBM to bolster and expand our capabilities through additional computational software and cognitive computing tools.
Additionally, like our colleagues across the country, we are beginning to look at how best to integrate the use of predictive analytics across our hospital while maintaining fidelity, transparency and improved outcomes for patients.
Predictive analytics is yet another emerging, powerful tool that, with a bit of experience, data wrangling ability and action, health care providers can harness in our ongoing efforts to provide the best possible personalized care for those we serve.
Jean E. Robillard, M.D., is vice president for medical affairs at University of Iowa, and dean of the Carver College of Medicine
