Leveraging ChatGPT and Generative AI in Healthcare Analytics

ChatGPT, the artificial intelligence (AI) chatbot developed by OpenAI and released in November 2022, has already made headlines for its ability to write essays in the style of best-selling authors and pass the Bar Exam and U.S. Medical Licensing Examination.

Not surprisingly, experts say its impact—and the impact of similar “generative” AI tools (i.e., tools that generate new outputs based on existing data)—on healthcare data analytics may be substantial in the years ahead.

“Over the next decade, I anticipate these technologies will mature and become integral parts of the healthcare ecosystem,” says Bernard Marr, business and technology futurist at U.K.-based Bernard Marr & Co.

The future of healthcare is all about leveraging its most valuable commodity—data, says Robert Pearl, MD, healthcare author, podcast host, and clinical professor of plastic surgery at Stanford University School of Medicine in California. “ChatGPT takes vast amounts of data that already exist and analyzes it to provide new insights in objective ways,” he adds.

Big Data and Generative AI

Generative AI can help healthcare organizations identify the most significant challenges with population health and discover unique opportunities to address those challenges.

“The power of ChatGPT is the ability to summarize all kinds of data and ask anything you want,” says Harvey Castro, MD, author of ChatGPT and Healthcare: The Key to the New Future of Medicine. “You can literally have a conversation with the data.”

The healthcare industry continues to explore practical uses of AI. Not surprisingly, the US “AI in healthcare” market is expected to grow by 36 percent between 2023 and 2030 with partnerships between Epic and Microsoft leading the way, according to Grand View Research. Several healthcare organizations are already piloting this integration to draft asynchronous physician responses to patient questions in online portals.

Other hospitals continue to pilot ChatGPT directly. For example, Tenn.-based Vanderbilt University Medical Center discovered ChatGPT could support more expedited clinical decision-making. Health information professionals can leverage these discoveries and opportunities to promote data integrity.

“[Generative AI is] about finding outcomes and answers that would not otherwise be possible because the data is so complex,” says Castro.

Marr agrees, adding that as generative AI tools continue to mature, they’ll be able to:

• Predict what treatments will likely yield the best outcomes for each patient population based on medical histories, genetics, and lifestyle factors.
• Detect and prevent diseases, further enhancing patient outcomes.
• Identify inefficiencies in care delivery by suggesting optimal resource allocation and automation for certain tasks.

“With its natural language processing capabilities, ChatGPT can also interact with patients, reminding them to take medications, ask about their well-being, and collect subjective health information,” Marr says. “This data, when analyzed collectively, can provide a comprehensive view of a patient’s health over time, leading to more personalized care strategies and improved health outcomes.”

Disease surveillance is another area of opportunity. For example, if ChatGPT or similar generative AI tools had been fully operational during the initial stages of COVID-19, healthcare organizations could have swiftly analyzed large volumes of data from varied sources, Pearl says.

“These insights could have led to earlier warnings about the impending pandemic, possibly triggering faster responses,” he adds. “Additionally, it could have provided real-time analysis of scientific papers and global news to inform the development of potential treatments or model the disease’s spread. With such information, public health measures and guidelines might have been more targeted and effective, reducing the overall impact of the disease.”

ChatGPT and similar generative AI tools can also impact healthcare data analytics in other ways. For example, AI tools can expedite medical documentation and reduce documentation errors, says Marr. “This improves the quality and reliability of the data, leading to more accurate and actionable health insights,” he adds.

Similarly, by leveraging generative AI tools in clinical trial recruitment, healthcare organizations can improve the diversity and accuracy of healthcare analytics. “AI can efficiently identify appropriate trial participants based on diverse variables, resulting in a broader participant base,” says Marr. “This helps minimize bias in data and offers a more comprehensive view of trial outcomes, thus enhancing the reliability of health analytics and leading to more effective treatments and healthcare strategies.”

Addressing Data Challenges

What are some of the biggest challenges of leveraging generative AI for healthcare analytics? Data privacy and security, says Castro. Healthcare organizations must ensure that any data they share with generative tools doesn’t violate the Health Insurance Portability and Accountability Act of 1996 (HIPAA). This may require an intermediary to scrub and deidentify the data, he adds. Another option is to install ChatGPT locally so protected health information or intellectual property isn’t disclosed and used to train the algorithm.

Another challenge is ensuring unbiased data, says Marr. “AI systems learn from existing data, and if this data is biased, it can lead to skewed results and reinforce health disparities. Overcoming this challenge requires robust regulatory frameworks, careful data management, and continuous efforts to address and reduce bias in AI systems.”

Another caveat is that AI can also be prone to hallucination, says Castro. “It doesn’t know what it doesn’t know. It’s just putting words and algorithms together based on the probability that something is right. You don’t know whether it’s correct,” he says, adding that’s why it’s important to use caution and implement policies on the use of AI.

Generative AI tools as we know them today are just the beginning, says Pearl. “ChatGPT is doubling in power every six months to a year. In five years, it’s going to be 30 times more powerful than it is today,” he says. “In 10 years, it will be 1,000 times more powerful. What exists today is like a toy. In next-generation tools, it’s estimated there will be a trillion parameters [values used to control the model’s behavior], which is interestingly the approximate number of connections in the human brain. It’s going to be a massive magnitude of interconnected data and information.”

As more data becomes available, generative AI will become even more powerful, says Castro. And it isn’t only healthcare organizations that are the beneficiaries. For example, in an era of price and healthcare quality data transparency, healthcare consumers will benefit, too.

“A consumer could ask, ‘I have $10,000 for a knee replacement. Where can I go to get it done?’ ChatGPT uses the power of summarization to go into the data and provide an answer,” he adds.

Experts say health information professionals should ask these questions about their role in generative AI:

• What are the emerging generative AI tools, and what are the potential implications on healthcare data analytics for my organization?
• How can I improve my skills in the areas of machine learning, data analytics, and privacy and security?
• Will my organization start to leverage generative AI now? If so, what data will we analyze? What are the most pressing questions to which we seek answers?
• Does our electronic health record vendor have a plan to incorporate generative AI? If so, how and under what timeline?

“People should be very optimistic and positive about generative AI,” Castro says. “It will completely revolutionize how we practice medicine and empower patients. It will also lead to more consistent, higher-quality care at a lower cost.”

Health information professionals must be ready and willing to embrace inevitable change to their jobs, Pearl says. “Healthcare jobs will definitely be lost, but new ones will be created,” he adds. “The most important part is that patients will benefit, clinical quality will improve, and patient convenience will increase dramatically. Healthcare is one of the only industries that has not undergone disruption. That is about to change.”

Lisa A. Eramo, MA, is a Rhode Island-based healthcare freelance writer and editor.