Medical Coding Automation

CorroHealth Breaks Barriers with GenAI and Human-Like Reasoning

How PULSE Coding Automation Technology™ helps hospitals achieve 7X productivity at scale with more accuracy and shorter client onboarding time than competitors. 

Medical coding has long been plagued by one massive obstacle: the sheer variability and complexity of clinical documentation. From nuanced procedural descriptions to context-specific diagnoses, autonomous medical coding platforms have historically struggled to fully understand medical records. This challenge resulted in slower operations, increased errors, and higher costs. That is until PULSE Coding Automation Technology™ entered the market.  

In the summer of 2024, CorroHealth demolished this barrier for the industry and its clients by upgrading PULSE, demonstrating its commitment to solving complex financial and operational challenges for hospital clients. But how did they achieve this breakthrough? 

With the help of researchers at The University of Texas at Dallas (UTD), one of the top technology universities in the U.S., the CorroHealth team integrated the latest generative AI (GenAI), large language models (LLMs), and reasoning engines into PULSE. The advancements delivered unprecedented accuracy, faster client onboarding, and up to 7X productivity at scale. 

The Problem: Variability in Clinical Documentation  

Medical coding relies heavily on context. Many codes depend on details about conditions, such as how an injury occurs or what affects a diagnosis. Compounding the challenge is the industry’s lack of standardized documentation practices, resulting in highly variable medical records.  

Older coding engines equipped only with natural language processing (NLP) models grapple with these variables. They frequently miss critical nuances around language and diagnoses, leading to inefficiencies and inaccuracies.  

While PULSE performed very well in the majority of coding applications, it often required the development of additional post-process algorithms in software to compensate for aspects the NLP missed, resulting in internal programming hours and extended client implementation times. “We knew there was a better way,” said CorroHealth Executive Vice President and Chief Technology Officer Ravi Narayanan.  

The Breakthrough: Leveraging UTD’s Expertise to Solve Complex Problems 

That “better way” came through a partnership with UTD’s Center for Applied AI and Machine Learning (CAIML)  directed by Dr. Doug DeGroot and Dr. Gopal Gupta. Partnering with UTD’s CAIML provided the necessary impetus for CorroHealth to leverage GenAI alongside CAIML’s automated commonsense reasoning technology and enhance the performance of PULSE.  

“We asked Dr. Gupta and his team to leverage their expertise and help us solve this challenge,” Narayanan said. “They created a solution for some use cases and taught our team how to implement GenAI and commonsense reasoning engines into our AI architecture.” This approach ensured the advancements aligned with the company’s broader technological ecosystem.  

By incorporating the latest AI models, PULSE can process and interpret complex clinical documentation with unmatched precision. “We used large language models to extract knowledge and use commonsense reasoning engines to draw precise conclusions like humans,” said Dr. Gupta. “This allows the AI to read and understand all the variations in clinical documentation and output with more accuracy and consistency across even the most complex medical records.” “Use of large language models alone will not produce high accuracy due to hallucinations inherent in them,” he added.  

Book A Demo Today 

These innovative AI models make PULSE one of the most advanced and accurate medical coding platforms on the market and an indispensable solution for your hospital’s coding operations. Discover how PULSE transforms medical coding and financial outcomes in this case study, or book a demo by filling out the form below. 

Book a Demo

About Center for AI and Machine Learning 

 The Center for Applied AI and Machine Learning (CAIML) is an industry-facing center at UT Dallas Department of Computer Science whose goal is to apply AI and machine learning technologies to solve problems that the industry is facing. AI is having an outsized impact on our society and economy, and CAIML researchers are ready to help companies understand and exploit the AI landscape. CAIML has worked with companies around the world in solving their problems using AI. Most of these projects have resulted in intellectual property that is owned by the company. CAIML also provides training in AI and Machine Learning to companies. CAIML is home to more than a dozen researchers with the whole range of expertise in AI, ranging from machine learning, deep learning, computer vision, automated reasoning, to natural language processing, constraint optimization, to statistical relational learning. The Center is directed by Professors Doug DeGroot and Gopal Gupta.