Healthcare artificial intelligence (AI) recently made headlines when The Wall Street Journal reported that IBM intends to sell Watson Health. The company’s Jeopardy!-winning AI software uses natural language processing (NLP) for clinical decision support and includes “real-world, longitudinal clinical, operation and financial data, and analytic tools.”
However, AI’s potential for healthcare is not limited to clinical support. The technology may also help enhance revenue cycle management, improve social determinants of health, and combat racial disparities.
How AI Can Streamline and Simplify Revenue Cycle Management (RCM)
A study from Change Healthcare concluded that AI “will transform the way doctors, hospitals, and healthcare systems identify, collect, and manage their revenue cycle over the next three years as healthcare organizations evolve from day-to-day use to strategic integration within their systems.” Nearly all (98 percent) surveyed healthcare leaders anticipate using AI for RCM purposes in the near future, and 65 percent are already applying AI-driven RCM practices.
“Providers that close the gaps revealed by this research will be well-positioned to reap financial, operational, and clinical gains from the technology—including improving the end-to-end revenue cycle, claims accuracy, denial reduction, clinical insights, level-of-care prediction, and more,” said Luyuan Fang, Ph.D., Chief AI Officer at Change Healthcare.
AI can enhance revenue integrity, using advanced software or robotic process automation (RPA) that “proactively identifies and detects anomalies, then swiftly corrects any errors,” according to Carleigh Moore, Revenue Optimization Manager at Privia Health.
AI-Driven Insights for Social Determinants of Health
Data shows that, at 61 percent, clinical decision support is the most common use area for AI. AI can analyze the social determinants of health using external sources, such as the U.S. Census, as well as NLP to identify social determinants in unstructured data inside an electronic health record (EHR). Smart software, such as Google’s suite of AI tools, can rapidly sift through data sitting in “clinical notes, patient-reported data, secure e-mail exchanges, patient portal messages and other places,” Healthcare IT News reported. This advanced process takes place in the background, generating insights while care teams are freed to focus on delivering patient care. Mining and structuring patient data can “rapidly identify facts, relationships, and assertions” to form a comprehensive patient profile of factors that may account for 20 percent of health outcomes.
“A user can quickly create a query to extract key concepts and relationships from unstructured patient data to identify issues such as social isolation, transport problems and cultural factors that may impact health and outcomes,” Elizabeth Marshall, MD, Director of Clinical Analytics at Linguamatics, said. “The data can then be used with analytic tools, such as machine learning algorithms, predictive analytics and risk stratification models.”
Combating Racial Disparities in Healthcare
Race and ethnicity are among the most influential social determinants of health. Research shows racial disparities in cardiovascular health and telehealth utilization, and many patients report discrimination. Critics have also found racial disparities in healthcare technology. The Food and Drug Administration (FDA) recently warned that pulse oximeters are less accurate for people with darker skin.
This bias can extend to AI. As Scientific American noted, “We’ve long known that AI algorithms that were trained with data that do not represent the whole population often perform worse for underrepresented groups.” It is crucial that developers account and test for bias when programming and coding. When done correctly, AI can actually reduce unconscious bias, according to a recent study published in Nature Medicine. Researchers found that AI could predict patients’ pain, a highly subjective measurement that, as such, is at risk for bias, which in turn can lead to misdiagnosis or mistreatment.
What Stands in the Way of AI Adoption?
“Financial, security, and privacy concerns block AI adoption and dampen success factors,” according to researchers from Change Healthcare. A lack of transparency and understanding of AI’s “black box” design may also hinder advancement. Among providers, 56 percent reported “liability, risk, and privacy concerns.”
However, COVID-19 has highlighted AI’s value. One AI model was able to predict the length of COVID-19 hospitalization, and some experts believe AI may help combat the virus’s mutations. As the pandemic has illustrated telehealth’s vast, previously untapped potential, so too might the pandemic advance AI. As the FDA’s landmark clearance for an AI-driven ICU predictive tool and recent action plan show progress, it is crucial the industry consider how AI tools affect physician workflows. Physician input into clinical informatics is essential to ensure AI tools work for — not against — care teams and patients. If approached properly and purposefully, AI can make physicians’ lives easier, streamline workflows, and improve health outcomes.