Beyond Buzzwords: Applications of Artificial Intelligence in Healthcare
Healthcare — especially healthcare analytics — is a field laden with buzzwords. Factor in the innumerable list of abbreviations and acronyms that have become industry-standard vernacular, and it can sometimes seem like analysts are speaking a different language entirely.
Understanding these words in concept is one thing, but how exactly they apply in healthcare, and the value they deliver, is another. In an effort to decode the jargon, we’re tackling some of the most nebulous terms of all: artificial intelligence, along with two of its subsets, machine learning and natural language processing.
Artificial Intelligence (AI): An Umbrella Term
In his book, Deep Learning with Python, Keras creator and Google AI researcher François Chollet offers one of the simplest, yet broadest, definitions of AI: “the effort to automate intellectual tasks normally performed by humans.” AI has matured considerably over the last several decades, making these tools more affordable, and, thus, more widely accessible to researchers, clinicians and technology providers.
Just as AI has evolved, so too has the data that underlies it, notably with the introduction of electronic health records and other health informatics systems. As computer processing power continues to increase exponentially, the volume of data available for healthcare analytics and AI tools presents tremendous opportunity across many areas of healthcare.
Machine Learning: Driving Continuous Improvement
A broad and commonly cited definition of machine learning is “an application of AI that provides systems the ability to automatically learn and improve from experience without being explicitly programmed.”
Machine learning is all around us — from autonomous vehicles to those targeted online ads that have become eerily relevant based on our browsing history. These tools are driving value in several facets of healthcare. For example, deep learning algorithms are getting better and better at detecting abnormalities in diagnostic images, enabling earlier intervention and prevention of disease.
Machine learning is also helping healthcare organizations improve administrative efficiencies and combat fraud, waste and abuse. Applying machine learning to program integrity processes enables systems to automatically learn and improve from experience. This creates an ever-sharpened focus on the attributes and triggers that not only merit an enforcement event, but also lead to an effective resolution.
Natural Language Processing (NLP): Analyzing Human Language
NLP is a subset of AI and machine learning that analyzes language. It’s a big part of the reason chatbots and virtual assistants have gotten so good at what they do, and why claims administration and payment integrity processes have gotten much more accurate and efficient.
When reviewing claims for potential fraud, waste and abuse, NLP-enabled tools like smart text detection, predictive text and optical character recognition can help streamline workflows and minimize human error. NLP can also analyze different coding languages and, when integrated with machine learning algorithms, drive continuous improvement in fraud detection for smarter, more accurate decision-making.
Future-Ready Solutions Capturing Fraud, Waste and Abuse
Gainwell Technologies is turning analytics buzzwords into action to combat the multi-billion-dollar issue of healthcare fraud, waste and abuse.
To that end, we are actively investing in our data infrastructure to support a number of AI applications to improve payment accuracy and enhance healthcare workflows. This includes adhering to data governance principles and ensuring our dataset meets the five Vs of Big Data — volume, variety, velocity, veracity and value — all fundamental components of AI.
Currently, we are applying machine learning to detect anomalies in provider billing patterns. Employing a clustering approach, we can identify statistically significant differences between datasets that could indicate potential fraud. Using pattern recognition, we continuously improve our ability to identify these anomalies, providing a strong basis for analyzing future data.
Additionally, NLP is streamlining the process of reviewing medical records and claims to ensure services are accurately coded, billed and paid, while accelerating the recovery of improper payments.
Drawing on thousands of data analytics, algorithms and machine learning technologies, Gainwell’s Cost Containment and Care Quality solutions, including solutions for addressing fraud and payment integrity, are continuously refreshed for accuracy as well as the ability to identify emerging improper payment trends. Our capabilities extend across all claim types and billing issues and are driven by the expertise of over 850 clinicians and coders and more than 1,100 technologists.
Interested in learning more about meeting the data-driven challenges of today and tomorrow? Read “The Rise of Advanced Analytics in Medicaid.”