Separating hype from reality: a buying guide for AI technology in healthcare

The number of health tech solutions has exploded in recent years. Driven by the need to reduce costs, improve patient care, and reduce staff burden, healthcare has begun to embrace digital transformation out of necessity. In this shift, the number of vendors has skyrocketed.

Executives now face a dizzying array of options when making purchasing decisions, all promising to revolutionize the industry. In a landscape cluttered with buzzwords and lofty promises, it's crucial for decision-makers to discern hype from reality, especially amidst all the buzz about artificial intelligence (AI) in healthcare.

The impact of AI technology in healthcare

The potential of AI is enormous. According to a recent paper by McKinsey and Harvard economists, AI and automation in healthcare have the potential to save $200 – $360 billion annually by reducing inefficiencies and improving care quality. 

Organizations using these intelligent solutions are already seeing a real-world impact. Rhode Island’s second-largest health system, Care New England, automated radiology authorizations and notices of admission, resulting in a projected cost savings of $644k annually. At Castell, chart review is an essential part of their value-based care business, but maintaining enough staff to review all patients’ charts for care gaps would be costly. Automating chart review saves Castell $2.8 million annually.

To unlock this value, executives should learn what these technologies do and how best to put them to use.

What is AI?

Let’s start with some basic definitions. AI refers to a computer's ability to mimic human cognitive functions, such as learning, problem-solving, and decision-making. AI's unique strengths lie in its ability to process vast amounts of data, identify patterns, and make data-driven decisions. 

Recently, large language models (LLMs) like GPT have been making headlines. Large language models are more sophisticated than their predecessors because they can process and generate human-like language. These models are particularly applicable in healthcare because they excel at interpreting and acting in response to text. Between patient medical records, claims documents, clinical documents, consent forms, waivers, and more, healthcare is a documentation-heavy industry. Applying these models, which have the ability to quickly read, interpret, and act on massive amounts of text, is a competitive differentiator.

Sample use cases of AI technology in healthcare include: 

• Predicting disease: AI can analyze patient data to predict diseases and identify patients who are at high risk.
• Driving revenue cycle efficiency: Machine learning techniques can be used to help technology learn how to select the correct payer-plan without a human in the loop, select the correct billing codes, and automatically submit claims. 
• Personalizing care: Notable’s Patient AI uses LLMs to analyze a patient’s full medical history, to offer the right kinds of support to patients at the right time. 
• Detecting imaging anomalies: AI can analyze medical images such as X-rays, CT scans, and MRIs to detect diseases, tumors, and other anomalies.

What is automation?

Automation, on the other hand, involves using technology to perform tasks without human intervention, often following predefined rules and processes. Automation focuses on streamlining repetitive tasks, reducing manual labor, and improving efficiency. 

Sample use cases of automation in healthcare include:

• Patient registration: Automation can facilitate the collection and management of patient information, such as demographics, insurance details, and medical history, simplifying the registration process and improving data accuracy.
• Patient intake: By automating the intake process, healthcare providers can efficiently gather relevant patient data, such as chief complaints, medication lists, and vital signs, reducing manual labor and improving the overall patient experience.
• Appointment scheduling: Automation streamlines the process of booking, confirming, and rescheduling appointments, minimizing manual labor and enhancing patient access to care.
• Referrals and authorizations: Automation can expedite the process of obtaining referrals and prior authorizations, ensuring that patients receive timely care and helping providers adhere to insurance requirements.
• Insurance verification: Automated systems can verify patient insurance coverage and benefits, identifying potential issues and ensuring accurate billing.

While both AI and automation offer significant benefits, it’s the combination of their strengths that can truly impact many of healthcare’s most pressing needs. 

Intelligent automation — the antidote to what ails healthcare

One common pitfall in healthcare IT has been to take processes that used to occur on paper and simply move them to a digital format, without considering what the ideal process should be. This is digitization, without automation. 

Intelligent automation combines AI and automation, leveraging machine learning and natural language processing to enhance task execution. It addresses the shortcomings of many healthcare IT vendors that simply digitize analog processes without fully automating the work. 

Systems of record versus systems of intelligence

While the EHR took an important step in digitizing patient records, it didn’t ultimately transform processes. With EHRs, team members still have to do just as much manual work, and in some cases even more. Ultimately, the EHR is a system of record. It is not a system of intelligence or a system of action. 

“And what we're actually doing when we try to use the EHR as our system of action, is we're introducing friction into the patient experience,” said Notable’s Bri Buch on the Perspectives podcast. “We're taking a tool with one core competency and trying to use it to accomplish a different goal.”

Meanwhile, intelligent solutions are able to automate processes from beginning to end without any humans in the loop, eliminating work for staff. Augmenting the system of record with a system of intelligence provides the benefits of both and enhances the experience for patients and for caregivers.

The benefits of intelligent automation — combining AI and automation

The benefit of intelligent automation becomes clear when examining the process of insurance card capture. Without automation, front desk staff typically flip through binders of information on payer-plan selection. Error rates with a fully manual method such as this tend to be high. Simply digitizing this process would allow the patient to take a picture of their insurance card and upload it, while still relying on a team member in the background to select the correct payer-plan. 

But by adding machine learning, intelligent automation can determine the correct payer-plan without any human intervention — not only getting the job done faster but ultimately increasing accuracy. For example, when Austin Regional Clinic achieved 78% touchless registrations, their denials decreased. 

Whereas digitization without automation can maintain or add to staff burden, true intelligent automation decreases the amount of work for team members. Learn how Castell deployed intelligent automation for payer care gap attestation and in doing so offloaded work to a digital workforce that would otherwise require an additional 47 full-time care coordinators to complete. 

Vendor evaluation criteria for AI technology in healthcare

While some organizations have in-house capabilities to build, train, and maintain AI, others choose to partner with highly experienced organizations that focus exclusively on this work. To make well-informed decisions, C-suite executives should consider the following criteria when evaluating healthcare IT solutions:

• True intelligence: Truly intelligent solutions fully eliminate manual work for staff. Identify intelligent solutions that are able to remember information that is already in the system, and use it to proactively take the next step without staff intervention. 
• Scalability: Ensure the solution can grow with the organization and adapt to increasing demand. The vendor should be able to deploy quickly, no matter the organization's size. For example, Montage Health increased its digital patient engagement by 200% within 6 weeks. 
• Integration: Verify compatibility with existing systems and workflows, minimizing disruptions. Integration should be seamless with any systems you use. 
• Measurable outcomes: Assess the potential for quantifiable improvements in cost, efficiency, and patient satisfaction. Conversations with the vendor should begin with estimates of the kind of value they can offer.
• Proven results: Research what other health systems have been able to achieve with the vendor. 
• Data security: Choose solutions that adhere to regulatory standards and safeguard sensitive information.
• Transparency and analytics: Determine how much more information the vendor can offer to help you assess their performance. For example, Reid Health uses pre-visit completion dashboards to make informed decisions about how to maximize engagement. 
• Vendor support: Focus on providers that offer robust customer support, onboarding, and continuous improvement. 

These evaluation criteria make it easier to choose a vendor who will deliver on their promises. 

Whether considering intelligent automation for the front end of the revenue cycle or as an end-to-end platform, the key is to focus on efficiency, capacity, and revenue to assess the value of automation.