Iodine Software is charging ahead to integrate generative AI technology into its software to make a "quantum leap" in automation and predictive analytics capabilities.
The Austin, Texas-based company, a major player in healthcare artificial intelligence, is collaborating with OpenAI to integrate generative AI and large language models, including GPT-4, across the breadth of its solutions for clinical administration and revenue cycle management.
Iodine is a large clinical AI engine used by more than 900 health systems and 80,000 physicians. Its clinical AI engine improves clinical documentation to help health systems capture more revenue while also allowing providers to improve clinical care.
The company, which was founded in 2010, experienced rapid growth over the last decade by applying its machine learning engine to solve midcycle revenue leakage—a problem that costs healthcare systems billions in lost revenue due to resource-intensive, highly manual clinical documentation workflows.
By infusing generative AI into its AwareCDI product suite, Iodine sees the potential to improve the accuracy and predictive capabilities of its existing technology. The aim is to also develop clinical automation tools that will further stem revenue cycle leakage by ensuring documentation accuracy.
The collaboration with Microsoft-backed OpenAI enables Iodine to work with a "fellow pioneer in the field of artificial intelligence," according to William Chan, Iodine CEO and co-founder.
"Strategic, fast-paced yet cautious innovation has always been our guiding principle. The evolution towards generative AI is a natural next step for us," he said.
While there seems to be a general frenzy around generative AI in the broader tech community and within healthcare, Iodine has a deep-rooted history in clinical AI tech and has growing ambitions to use the power of AI to automate complex clinical tasks and generate insights for doctors and for health system operations.
"We have been using AI all this time to detect inaccuracies in clinical documentation. When you have inaccurate documentation that can impact patient care and it can also certainly impact the reimbursement that the hospital gets back. We look at this partnership with OpenAI as getting to that next level of what we think we will be able to achieve with this technology," Chan said in an interview.
Many organizations are experimenting with generative AI language models, like GPT-4 or ChatGPT, to generate text, such as drafting a response to an insurance claim denial. Chan sees the potential to use the technology for summarization to ease the paperwork burden for clinicians.
"Entering text or data into systems like ChatGPT and say, 'Give me a summary of this.' I think there is a very applicable use of that within healthcare," he said. "There is a lot of information that a care provider has to understand in a very short amount of time. We believe that technology like ChatGPT can help summarize that very briefly, succinctly and accurately so that the physician can move forward to focus on the care treatment plan instead of having to read through reams and reams of virtual paperwork."
Technologies being developed by companies like OpenAI also can boost the capabilities of predictive analytics in healthcare operations.
"We could predict whether a patient has encephalopathy or heart failure. We can better predict whether the documentation is accurate or not. That could be another quantum leap in terms of the accuracy and predictiveness of our existing models. We are seeing that ability to use that technology to kind of take that next leap, and that's why we're excited about this partnership," Chan said.
Many health tech companies are jumping in to incorporate generative AI as to not get left behind in the AI race.
According to Chan, Iodine Software has capabilities and assets that put it a step ahead in the push for AI innovation in healthcare.
The efficacy of AI is dependent on the data on which the language models are trained.
Chan says his company has a massive data set, what he calls "unmatched" in the industry, that contains more than 27% of all U.S. inpatient data, giving the company an opportunity to transform how it supports hospitals across a variety of functions. This data set is a powerful asset for training machine learning and large language models, Chan noted.
"We believe we can build better models and better prompts into ChatGPT because of the breadth and depth of the data that we have," he noted.
While Chan says he is optimistic about the potential impact of generative AI in healthcare, he also acknowledges that the tech is still in a nascent stage. There is a need to be cautious due to the significant impact on patient care, physician trust and patient reimbursements, he said.
"Aware of its limitations, such as data hallucinations, we are committed to a responsible approach, ensuring we balance progress with prudence," he said.
By incorporating the latest innovations in large language models, Iodine Software can ensure more specific patient documentation and coding efficiency, increase prediction accuracy and improve its predictive analytics. The company believes these advancements will help streamline the physician documentation experience, allowing doctors to focus more on patient care, and help hospitals and health systems capture more earned revenue from the care they provide.
While Iodine's machine learning approach initially focused on solving things like clinical documentation improvement, Iodine has recently picked up other tech companies to build out its AI platform and broaden its market reach.
In May 2021, Iodine snapped up Artifact Health to get its AI technology directly into the hands of more than 80,000 providers across the country. A few months later, the company bought competitor ChartWise Medical Systems.
About 18 months ago, private equity firm Advent International bought a significant stake in Iodine Software, valuing the company at more than $1 billion. The private equity deal reinforced Iodine's growth and provided the company with "firepower" to expand its technology to other strategic areas of care delivery, Chan said in an interview at the time of the deal.