Sound clinical and business decisions in healthcare hinge on high-quality data. Capitalizing on available data resources is a technological challenge for any industry, but healthcare faces the steepest challenges of all—no other industry encounters so much volume, variety, and velocity of information, and in no other industry are the decision-making stakes so clearly a matter of life and death.
While it’s an exciting time for predictive analytics in healthcare, it’s also important to be realistic about what data-mining, modeling and machine-learning can and cannot accomplish. Even the most sophisticated electronic health record (EHR) system can’t generate outcome improvements without first being coupled with other metadata sources, to provide needed layers of context and evidence.
Despite limitations and growing pains, the potential benefits of using predictive analytics within healthcare are enormous, particularly when health networks and payers are willing to work together. As financing systems and care delivery networks collaborate more closely, even to the point of full integration, predictive and prescriptive analytics will inform how all parties manage at-risk populations, identify candidates for clinical intervention, and provide better value and care to individual consumers.
Here are a few areas of promise:
Utilization. To provide more cost-effective care, providers must manage space and personnel resources optimally. Many hospitals already use software to help predict emergency room utilization, or to predict no-shows for specialty appointments.
These algorithms are getting more sophisticated every year. Computers will soon know where the flu is spiking, when there’s a big game or concert in town, and how those real-time trends are likely to affect emergency room admissions hour by hour.
Real-time infection control. It’s not enough to conduct analytical deep-dives after a sepsis case or a central line-associated bloodstream infection (CLABSI) report.
Today, health systems are piloting real-time analytics platforms that look for advanced warning signs of serious infections. They determine which central lines are due for maintenance, or identify patients that are at risk for sepsis by using “sniffer” algorithms to assign risk scores.
There’s a lot at stake for all parties: 250,000 CLABSIs are reported annually in the U.S., according to the CDC, while sepsis may contribute to half of all in-hospital deaths and account for $24 billion a year in hospital costs, according to recent studies. The networks that can figure out how to predict, and prevent, infections will squeeze cost out of the system and create a safer care environment for patients.
Oncology. Treating cancer effectively is like solving a puzzle, and oncological cost management is a challenge for health carriers. Because there is so much variability and cost built into cancer treatment, the prospect of using analytics to augment value-based oncological care is tantalizing.
That’s why health systems are now leveraging analytics to make clinical decisions that help chemotherapy patients avoid extra hospitalizations, and protect payers from associated expenses.
Hospitalizations can be circumvented if we are able to effectively comb through and interpret the relevant clinical and treatment data. Different factors—the type of chemotherapy agent, the patient’s age and weight, the cancer stage—feed into risk scores, and those risk scores can then be married with predictive models that can anticipate health outcomes for an individual.
That’s where the real promise of analytics lies—not just predictive medicine, but personalized, prescriptive medicine, creating a better, high-value care experience for our patients.
Cynthia Hundorfean, a Managed Healthcare Executive editorial advisor, is president and CEO of Allegheny Health Network (AHN), an integrated healthcare delivery system that serves Western Pennsylvania. AHN is part of the Highmark Health family of companies.