Future of AI in Healthcare

Any prescription for improving the health of hospital systems today includes a megadose of artificial intelligence (AI). The technology is transforming how care is delivered and how the business of healthcare operates, helping improve patient access, providing more accurate clinical diagnoses, facilitating better workflows, relieving workforce shortages, and combating provider burnout.

Another important benefit is that AI can help harness the flood of healthcare data, speed up payment processing, and automate repetitive tasks, freeing workers to take on more demanding roles. The rapid development of generative AI, which uses deep-learning algorithms to parse billions of data points and then automate processes, promises even more dramatic progress.

However, advances in AI vary widely across the health system landscape, and some technological innovations are much further along than others. For hospital leaders considering how to get the most out of their AI investment dollars, it’s important to understand the evolving ecosystem of clinical and operational changes and to consider what kinds of internal and external capabilities to pursue now and in the months and years ahead.

Improving Care While Reducing Costs

One way healthcare AI is already benefiting hospital systems is by assisting medical professionals. For example, radiology is a pain point for many hospitals. More than half of radiologists say they suffer from burnout, and a recent survey found that job openings for medical imaging personnel are at a 20-year high. But help is arriving in the form of AI-aided approaches that increase efficiency and quality of care, while minimizing costs and relieving pressure on physicians and support staff.

For several years now, researchers have "trained" AI algorithms, using millions of existing images, to read X-rays, CT scans, and other images to detect disease. AI has proven adept at spotting anomalies on scans and, in some cases, is able to find possible lesions that human eyes can’t see. Yet the true value of AI in radiology, at least for now, is to help radiologists rather than replace them.

According to a Mayo Clinic study, most Food and Drug Administration-approved diagnostic applications for AI to date involve reading radiology images to help physicians detect disease, often by handling time-consuming aspects of a radiologist’s job: tracing tumors and structures, measuring fat and muscle in full-body CT scans, and helping identify aneurysms, strokes, and embolisms.

AI "co-pilots" can also search medical imaging databases for similar images that may help radiologists make more definitive diagnoses. Collaborations among established and start-up companies in the field are helping streamline radiology workflows, enabling radiologists to handle growing volumes of images without sacrificing diagnostic quality.

The technology is poised to aid clinical diagnosis in other areas as well. In the growing field of using blood tests to search for cancer, one AI-assisted application detected more than 80% of liver cancers in a large patient population. AI is also crucial in expanding the use of precision medicine, which uses advanced methods of diagnosis to find disease earlier, match therapies to the precise genetic or molecular features of a patient’s disease, and monitor patients after treatment for recurrence of disease. In addition to liquid biopsy, next-generation genomic sequencing is crucial for these efforts, and AI "alignment" algorithms can be used for the rapid comparison of a patient’s DNA sequence with a reference genome.

AI may even aid the push for more equitable care. A new study found that generative AI outperformed healthcare providers in exploring notes in patient records to determine social determinants of health—factors such as housing, transportation, community support, and financial stability—that may adversely affect care.

As AI-spurred advances make their way into mainstream care, the financial payoff for hospitals and providers will increase. An estimate from the Harvard School of Public Health suggests that using AI to make diagnoses could ultimately reduce costs by half while significantly improving health outcomes.

Transforming Operations

AI may be having its most significant early impact on the business side of healthcare. Much of the technology needed to automate operations already exists, and health systems that have put it to work for payment reconciliation and processing, staff and appointment scheduling, and other expensive, time-consuming tasks are helping them reduce the cost of collecting payments and providing positive returns on their investments. Robotic process automation (RPA), for example, can handle many jobs and encourage the redeployment of staff to more complex functions that produce higher revenue.

Automation can be particularly helpful in managing the transition from fee-for-service billing to value-based care. It can help make sense of rapidly expanding data sets and codes from the International Classification of Diseases (ICD). That allows the technology to mine and digest vast amounts of data for risk adjustment and quality improvement purposes.

In an example of how a bank is partnering with a leading tech firm to leverage data automation to help healthcare firms accelerate their operations, Big Data Healthcare, a wholly owned indirect subsidiary of Fifth Third Bank, uses intelligent data automation to reconcile healthcare payments with bank deposits. While the vast majority of health systems continue to process such claims manually, in part because almost a third of claims are still on paper, Big Data Healthcare leverages RPA and machine learning in its revenue cycle management tools to accomplish in minutes what normally may take days of labor-intensive effort.

FUSE, the company’s flagship product, automates the reconciliation of remittance detail to deposit information, producing a customized output file for the providers’ systems. Drudgery is replaced with speed, and the army of workers that used to handle payment reconciliation can be reassigned to more productive work. "With FUSE processing files, employees can be quickly redeployed to other revenue-producing areas," says Dean Puzon, President and Co-Founder of Big Data Healthcare. "Our case studies suggest that this automation results in 60% to 75% cost savings."

Accelerating and improving clinical data coding can reduce errors and control costs while ensuring reimbursements are at the appropriate level. AI can also make sure that providers have valuable information at the point of care, minimizing the amount of time they must spend searching for and recording data during patient visits.

It can also reduce the administrative burden on already overtaxed nurses. AI technology can codify documents, automate data abstraction, and search medical records for information, freeing nurses and other staff members for patient care and other higher-level work. Taken together, such technological fixes can help reduce costs while improving care—the essential elements of value-based care.

Prompting Industry Collaboration

Along with this rapid innovation, many hospitals have joined with other health systems and industry partners to explore how the technology can benefit their operations and their patients.

One notable collaboration, the Health AI Partnership, has brought together Duke Health, Hackensack Meridian Health, Jefferson Health, Kaiser Permanente, Mayo Clinic, Michigan Medicine, New York-Presbyterian, UC Berkeley, and UC San Francisco, among others, in an effort to standardize industry best practices for leveraging AI software and distributing key insights.

The partnership’s goal is to formulate strategies and guidance for the use of medical AI to maximize effectiveness while minimizing risks. Findings are available online to help health systems make decisions about AI procurement and integration. The partnership is open to new members, and hospitals are invited to join or share their own findings.

Other health systems have formed networks with payers and banks to collaborate in securely sharing and using clinical and claims data. The objective is to streamline healthcare administration by overcoming operational problems that include fragmented data and laborious, manual processes for handling claims and patient information while maintaining patient privacy. Typically using blockchain technology to help secure the exchange of data, these networks are also investigating how to use "permissioned" AI, a decentralized network enabling permissioned access to data sets from network participants. When integrated with decentralized blockchain networks, it can create secure, cloud-hosted environments in which payers, providers, and others can collaborate and transact directly with each other.

Working with technology company partners can also facilitate the adoption of AI systems and platforms. In a five-year partnership with Microsoft, Duke Health is exploring how to use cloud technology to simplify and modernize IT operations. A planned Duke Health AI Innovation Lab and Center of Excellence will pursue AI applications that can help streamline clinical care, promote health equity, and advance medical research and education.

Other Microsoft partners, including NYU Langone, UC San Diego Health, Sutter Health, University of Wisconsin Health, Baptist Health Jacksonville, and Stanford Health Care, are testing Microsoft’s Azure AI platform with GPT-4, a HIPAA-compliant large language model that can be used to write prompts, generate patient-friendly explanations, suggest improvements in care plans, and flag potential safety issues. In one experiment, reported in the New England Journal of Medicine, GPT-4 correctly diagnosed 52.7% of complex test cases; in comparison, only 36% of medical journal readers made the right diagnoses.

The goal of a partnership between AI firm Annalise.ai and Mass General Brigham is to develop and deploy diagnostic-focused AI to improve medical imaging accuracy and workflow.

Other health systems are using investments in AI startups to accelerate their own progress.

Kaiser Permanente, Lifepoint, Mayo Clinic, and CVS Health Ventures have helped fund Abridge, an AI-powered clinical documentation company. Abridge software converts real-time doctor-patient conversations into structured clinical notes that can be integrated into patients’ electronic health records. Relieving physicians of the need to enter notes into the record can save an estimated two to three hours a day while helping them focus on examining and listening to their patients.

Investments in AI Education Also Pay Dividends

In determining how to use AI to meet short- and long-term challenges, many hospitals and health systems are investing in education, leadership, and AI tools. In one groundbreaking effort, the Joe R. and Teresa Lozano Long School of Medicine at the University of Texas Health Science Center at San Antonio (UT Health San Antonio) and the University College at the University of Texas at San Antonio (UTSA) have launched the first U.S. degree program to combine medicine and artificial intelligence. The five-year program will confer both an M.D. from UT Health San Antonio and a Master of Science degree in AI from UTSA, helping build a pipeline of multidisciplinary healthcare providers trained to improve diagnostic and treatment outcomes.

Other health systems, including Mayo Clinic, Duke Health, and Northwestern Medicine, are increasing their focus on AI by creating positions for chief AI officers or other technology leaders who are focused on increasing those systems’ investments in AI software and digital tools. Almost four out of five respondents to a recent Bain & Company/KLAS Research survey said they have increased spending on AI in the past year, and 75% expected to spend even more in the year ahead. More than half noted that procuring new software was a top goal, and AI is also a priority. Although just 6% said they have a strategy for generative AI already in place, half of the tech leaders said they were actively developing such a plan.

Opportunities for Healthcare Leaders

As AI applications for healthcare increase almost by the day, health system leaders must not only understand the evolving landscape of potential solutions but also decide whether to invest based on a clear cost-benefit analysis. An article in the Harvard Business Review, for example, examined the three areas in which AI can produce benefits: automating business processes, producing insights through data analysis, and communicating with customers and employees. Of the three, the article said that using AI for RPA to automate processes is the cheapest and easiest to implement and can produce the quickest return on investment (ROI).

While the labor cost savings achieved through RPA versus the cost of implementing and running the technology is a measurable metric, calculating all of AI’s ROI isn’t always a straightforward process. Some of the benefits, such as improved customer engagement, might be intangible and only become evident after the passage of time, according to one analysis.

Often, a health system’s choices may depend on scale, with smaller organizations preferring to focus on established AI applications with a track record of quality improvement and cost savings. Leaders of those systems could decide to watch what larger systems are deploying and then adopt emerging best practices.

Joining a collaboration could help systems learn from other organizations’ experiences, while partnering with financial institutions could give them access to time-saving AI automation tools that have already demonstrated their value. But at a time when advances in AI seem certain to continue to transform both the clinical and business sides of healthcare, staying abreast of changes and adopting AI-enabled solutions is becoming business as usual.

Fifth Third Bank has healthcare industry experts who understand the unique needs of healthcare firms—from physician practices to large hospital systems, as well as revenue cycle management providers, medical billing, pharmacy, and durable medical equipment, to name a few. Contact your Fifth Third relationship manager to connect with a healthcare expert.