The Truth about AI in Healthcare

Posted on June 18, 2018 I Written By

The following is a guest blog post by Gary Palgon, VP Healthcare and Life Sciences Solutions at Liaison Technologies.

Those who watched the television show, “The Good Doctor,” in its first season got to see how a young autistic surgeon who has savant syndrome faced challenges in his everyday life as he learns to connect with people in his world. His extraordinary medical skill and intuition not only saves patients’ lives but also creates bridges with co-workers.

During each show, there is at least one scene in which the young doctor “visualizes” the inner workings of the patient’s body – evaluating and analyzing the cause of the medical condition.

Although all physicians can describe what happens to cause illness, the speed, detail and clarity of the young surgeon’s ability to gather information, predict reactions to treatments and identify the protocol that will produce the best outcome greatly surpasses his colleagues’ abilities.

Yes, this is a television show, but artificial intelligence promises the same capabilities that will disrupt all of our preconceived notions about healthcare on both the clinical and the operational sides of the industry.

Doctors rely on their medical training as well as their personal experience with hundreds of patients, but AI can allow clinicians to tap into the experience of hundreds of doctors’ experiences with thousands of patients. Even if physicians had personal experience with thousands of patients, the human mind can’t process all of the data effectively.

How can AI improve patient outcomes as well as the bottom line?

We’re already seeing the initial benefits of AI in many areas of the hospital. A report by Accenture identifies the top three uses of AI in healthcare as robot-assisted surgery, virtual nursing assistants and administrative workflow assistance. These three AI applications alone represent a potential estimated annual benefit of $78 billion for the healthcare industry by 2026.

The benefits of AI include improved precision in surgery, decreased length of stay, reduction in unnecessary hospital visits through remote assessment of patient conditions, and time-saving capabilities such as voice-to-text transcription. According to Accenture, these improvements represent a work time savings of 17 percent for physicians and 51 percent for registered nurses – at a critical time when there is no end in sight for the shortages of both nurses and doctors.

In a recent webinar discussing the role of AI in healthcare, John Lynn, founder of HealthcareScene.com, described other ways that AI can improve diagnosis, treatment and patient safety. These areas include dosage error detection, treatment plan design, determination of medication adherence, medical imaging, tailored prescription medicine and automated documentation.

One of the challenges to fully leveraging the insights and capabilities of AI is the volume of information accumulated in electronic medical records that is unstructured data. Translating this information into a format that can be used by clinical providers as well as financial and administrative staff to optimize treatment plans as well as workflows is possible with natural language processing – a branch of AI that enables technology to interpret speech and text and determine which information is critical.

The most often cited fear about a reliance on AI in healthcare is the opportunity to make mistakes. Of course, humans make mistakes as well. We must remember that AI’s ability to tap into a much wider pool of information to make decisions or recommend options will result in a more deeply-informed decision – if the data is good.

The proliferation of legacy systems, continually added applications and multiple EMRs in a health system increases the risk of data that cannot be accessed or cannot be shared in real-time to aid clinicians or an AI-supported program. Ensuring that data is aggregated into a central location, harmonized, transformed into a usable format and cleaned to provide high quality data is necessary to support reliable AI performance.

While AI might be able to handle the data aggregation and harmonization tasks in the future, we are not there yet. This is not, however, a reason to delay the use of AI in hospitals and other organizations across the healthcare spectrum.

Healthcare organizations can partner with companies that specialize in the aggregation of data from disparate sources to make the information available to all users. Increasing access to data throughout the organization is beneficial to health systems – even before they implement AI tools.

Although making data available to all of the organization’s providers, staff and vendors as needed may seem onerous, it is possible to do so without adding to the hospital’s IT staff burden or the capital improvement budget. The complexities of translating structured and unstructured data, multiple formats and a myriad of data sources can be balanced with data security concerns with the use of a team that focuses on these issues each day.

While most AI capabilities in use today are algorithms that reflect current best practices or research that are programmed by healthcare providers or researchers, this will change. In the future, AI will expand beyond algorithms, and the technology will be able to learn and make new connections among a wider set of data points than today’s more narrowly focused algorithms.

Whether or not your organization is implementing AI, considering AI or just watching its development, I encourage everyone to start by evaluating the data that will be used to “run” AI tools. Taking steps now to ensure clean, easy-to-access data will not only benefit clinical and operational tasks now but will also position the organization to more quickly adopt AI.

About Gary Palgon
Gary Palgon is vice president of healthcare and life sciences solutions at Liaison Technologies, a proud sponsor of Healthcare Scene. In this role, Gary leverages more than two decades of product management, sales, and marketing experience to develop and expand Liaison’s data-inspired solutions for the healthcare and life sciences verticals. Gary’s unique blend of expertise bridges the gap between the technical and business aspects of healthcare, data security, and electronic commerce. As a respected thought leader in the healthcare IT industry, Gary has had numerous articles published, is a frequent speaker at conferences, and often serves as a knowledgeable resource for analysts and journalists. Gary holds a Bachelor of Science degree in Computer and Information Sciences from the University of Florida.