Three examples of how AI is going to change medicine

AI is changing medicine: ©The other house - stock.adobe.com

Artificial intelligence in the health care sector has exploded in the last year, as companies seek to revolutionize the way medical professionals approach diagnosis, treatment, patient care, and even billing. The convergence of advanced algorithms, big data analytics, and machine learning techniques has empowered health care companies to extract insights from vast datasets, that can lead to more accurate diagnostics, personalized treatment plans, and enhanced patient outcomes. AI's ability to rapidly process and analyze massive amounts of medical information, coupled with its capacity to identify patterns and trends, has positioned it as a transformative force in health care. Proponents of AI say it promises not only to streamline clinical workflows but also to significantly improve the overall quality and efficiency of patient care.

As these lofty promises start to turn into real-world applications and are rolled out to health systems, providers are getting their first look at how AI may change how they practice medicine and how patients receive care.

Here are three recent examples of AI applications and how they might change the roles of physicians, nurses, and medical billers.

AI in primary care

Sentara Health, which serves 1.2 million patients and has 30,000 employees, created a partnership with RhythmX AI to boost primary care through the use of predictive and generative AI technologies.

According to both companies, the partnership seeks to create hyper-personalized primary care for both clinicians and patients. The AI platform will provide Sentara primary care clinicians with direct access to clinical actions derived from relevant clinical and payor guidelines. This includes earlier disease detection, comprehensive EHR data analyses, and documentation support.

The platform is designed to present pertinent information to clinicians at the point of care, streamlining their workflow and enabling them to engage more fully with patients during visits. The goal is to empower primary care clinicians by eliminating the need to spend valuable time searching for information, allowing them to focus on delivering optimal patient care.

The platform is in its early stages, but at maturity, Sentara sees a future where all members of the primary care team, including physicians, advanced practice providers, and RNs, utilize the same intelligence platform. This unified approach will provide clinical actions based on relevant guidelines, fostering true team-based high-quality care, according to the company. Patients accessing Sentara primary care services, whether through walk-in clinics, urgent care facilities, or traditional primary care offices, will receive personalized care plans to ensure the highest quality of care.

Jordan Asher, MD, MS, executive vice president and chief clinical officer for Sentara, expressed enthusiasm for the partnership's vision. "Our vision is to infuse personalized clinical intelligence into every visit and treatment so that physicians can deliver more personalized care while returning the joy of practice," he said in a statement, adding that the collaborative effort focuses on addressing not only the physiologic dynamics of clinical conditions but also the broader aspects of patients' lives, including social determinants of health, mental health, and lifestyle factors.

"We are going after a big, shared vision that truly reshapes healthcare," said RhythmX AI CEO and founder Deepthi Bathina, a former Humana executive, in a statement. "There will never be enough physicians or staff to provide care using traditional models of care delivery, and demand will continue to outstrip capacity unless the models of care evolve.”

AI instead of a chronic care nurse?

Nursing is also seeing the effects of AI, and in the future, some nursing functions could be handled by an autonomous digital nurse named Florence.

Generated Health and the College of William & Mary have joined forces to develop next-generation AI, aiming to revolutionize the automation of care coordination. The focal point of this initiative is the creation of synthetic patient data to train an autonomous digital nurse using generative AI and reinforcement learning.

Reinforcement learning is a type of machine learning where in this case, the digital nurse, learns to make decisions by interacting with patients and clinical workflows. The digital nurse receives feedback in the form of positive or negative rewards based on the actions it takes with the objective of learning strategies that maximize the cumulative reward over time. This approach has been successfully adopted in training popular AI models like ChatGPT, according to the college.

Ingolv Urnes, CEO of Generated Health, said the company’s mission is to deploy one million digital nurses to enhance health care accessibility while reducing costs. With clinical evidence from 200,000 patients, Generated Health has already demonstrated the efficacy of combining clinical protocols and AI for better health outcomes.

For example, in the management of hypertension and medication titration, Florence was able to sustainably reduce average systolic blood pressure by 15%, reduce physician and pharmacist time by over 75%, and reduce administrative work ten-fold. Florence has also proven valuable in managing patients (typically with chronic conditions) pre- and post-procedure or surgery; in cardiac surgery an independent study concluded that Florence reduced hospital readmissions by 67% compared to the control group, according to Generated Health.

Dr. Haipeng Chen, assistant professor of data sciences at William & Mary, emphasized the project's focus on AI for social good, addressing challenges in health care access and staffing shortages. The team plans to leverage Generated Health's dataset to develop two models: a diffusion model generating synthetic patient data and an AI model for Florence, integrating large language model and reinforcement learning.

The overarching goal is to create a digital nurse capable of dynamic decision-making with minimal clinical rules. In a statement, Urnes emphasized the importance of engaging and providing feedback to patients, underlining the necessity for personalized care plans based on individual disease progression.

The research project will utilize Generated Health's dataset covering 25 million clinical conversations with over 200,000 patients in the US, UK, and Australia. Florence has demonstrated success in managing various chronic conditions, showcasing significant reductions in blood pressure, healthcare professional time, and administrative workload, according to the company.

To expedite the development of an autonomous AI nurse, the teams are creating an AI diffusion model to simulate patient engagement with Florence. This model, trained on de-identified real-world data, enables the generation of synthetic patient data for high-volume simulations, optimizing Florence's responses using reinforcement learning.

Synthetic patient data provides a safer and privacy-compliant alternative, allowing researchers to share data without privacy concerns. Florence will be trained through interactions with synthetic patients, mimicking real-world scenarios. The trained AI agent will then engage with real patients, continuously improving through feedback from these interactions.

AI in billing and coding applications

Medical billing is the lifeblood of any practice. Even slight delays can threaten the practice’s cash flow that it needs to survive, and with medical coders hard to find among a wider workforce shortage, it should be no surprise that AI companies see the space as the perfect place to apply their expertise.

A Medical Group Management Association poll conducted among medical group leaders found that the role of medical coders is the most challenging to fill, with 34% citing difficulties.

In another example of collaboration between an AI company and a health care company, Maverick Medical AI and ImagineSoftware, a revenue cycle management company, announced they are teaming up to provide an autonomous medical coding platform to achieve higher direct-to-bill rates.

The digitization of medical records, coupled with staff shortages, has created a pressing need to automate the coding process using AI. Maverick's coding solution is powered by generative AI, and combines domain knowledge with machine learning. The platform autonomously analyzes clinical notes and reports, accurately generating reimbursement codes (ICD-10, CPT) in real time, resulting in an 85% direct-to-bill rate.

ImagineSoftware processes more than $37 billion in annual charge volume, and it will offer the platform as a complementary product.

Yossi Shahak, CEO and founder of Maverick Medical AI, said in a statement: “The health care industry is facing severe challenges, and they are all interconnected and ultimately negatively impact the quality of patient care. As a leader in the RCM space, ImagineSoftware is well aware of the challenges that health care organizations face, and this is one of the reasons they decided to partner with us. Together, we will establish a new standard for revenue cycle management.”

More AI in health care is inevitable

While these are just three examples of the many collaborations and investments that are out there, physicians should expect to see AI rapidly infiltrating every aspect of their job. A Markets and Markets Research report estimates that the AI health care market in 2024 to be around $21 billion. That number is expected to grow to $148 billion by 2029, a growth rate of 604% in just five years.

With that kind of exponential growth, every person in health care can probably expect to see changes in how they do their job in that timeframe.