AI in Healthcare: The Fascinating Truth

Yes, the technology is new — but the challenges are all too familiar.

If you’re like many people, you may be a little tired of hearing about AI or skeptical about its value. In part, that’s because it’s popping up in every conversation. It feels as if people are talking about AI simply for the sake of saying “AI”. If you feel that way, you’re partially right, because the AI-washing phenomenon is real in many industries. Healthcare, however, is one of the few industries where the promise of AI actually justifies the hype — which is why you need to stay tuned in, learn more, and be prepared to act.

According to a recent Workday survey, 98% of all CEOs¹ believe there would be some immediate business benefit from implementing AI, yet only 71% report that their organization is using it currently.² In healthcare, the value-to-usage dichotomy is even greater. Bain’s survey of health system executives reveals that 75% believe AI has reached a turning point in its ability to reshape the industry, yet only 6% have an established AI strategy.³

A 69% gap between perceived impact and deliberate action is significant. It also raises a question: is AI really a technology so much more complex that the barriers to implementation are so daunting — or is something else at play?

New Technology, Old Challenges

Here's where things get really interesting. While AI is a cutting-edge, ground-breaking level of technological advancement, the biggest barriers to implementing it are about as old-school as it comes. Consider these all-too-familiar challenges when contemplating a pivot to greater AI adoption:

Legacy IT equipment and infrastructure. Technical debt is not new. Every time a significant advance in technology comes along, its demands often outstrip the capabilities of an organization’s existing systems. AI is no different, and the need to overcome legacy-system issues is a common barrier to implementation.

Siloed data. The trend towards cross-functional collaboration was first to shine a light on the threat that data silos hold for organizational improvement and advancement. As the amount of data generated and our dependence upon it increases, data silos loom as a critical obstacle. Because AI is driven by data, any friction to full access is a serious impediment to successful implementation and usage of AI.

Technical expertise gap. Like all technical advances, there is a learning curve that leaves the labor pool lagging behind. Right now, 87% of the C-suite say they are struggling to find a workforce with sufficient AI skills.⁴ Scarcity in available professionals not only slows AI projects down, but also increases costs as companies compete for talent.

Organizational culture of resistance to change. Yes, the old chestnut of “But we’ve always done it this way” also comes into play around AI. Some organizations simply have a culture that has bred resistance to change, particularly when failure can result in human harm, as is the case with the healthcare industry. This resistance can be heightened when it comes to implementing AI, as many people harbor fears — often stoked by news media — that AI will take their jobs.

That’s not to say that implementing AI doesn’t present some unique challenges. There are uncertainties around just what the ultimate capabilities of AI will be, as well as what limits government or society may implement out of caution or fear. Other worries include concerns about mistakes, such as biases from the data used to train the model and false information, as well as problems with violating intellectual property rights.

The good news is that no matter which type of challenges you’re facing, the potential rewards are worth the risk — and there is a clear way forward.

The Good Stuff

While challenges may be holding you back right now, it’s important to remember that healthcare is one of the areas where the benefits of AI are worthy of the hype.  Let’s take a moment to ask “what if” and explore just a few of the exciting things that are possible for organizations that push through these obstacles to seize the moment.

Matching patients with doctors who will be a good fit. A recurrent, pressing problem for care seekers is finding the right provider for their unique needs. There are seemingly endless lists of providers for care seekers, but too often, those lists are out of date or not accepting new patients. There’s also a lack of differentiation beyond the standard assignment of a specialty category. Without accurate, meaningful information, this method of selection often creates less successful provider-patient relationships that result in poorer health outcomes that deliver a lower value compared to the cost of care.

By incorporating AI into the provider-search process, care seekers could receive a smaller, but personally tailored list based on relevant, influential factors such as patient history, doctor treatment patterns and philosophies, patient outcomes, care costs, and reviews of the provider.

Enhancing appointments. Today’s time-strapped care providers deal with heavy burdens around records, documentation, and data. The surface result of this is a rushed (and often unsatisfactory to the patient) experience. But that haste is just a slippery slope to triage-type care support instead of deeper exploration, which can lead to missed diagnoses or unnecessary tests and follow-up visits.

AI allows providers to flip this script. Incorporating AI capabilities into pre-appointment prep for patients can turn the pages of information patients tediously complete ahead of time into select, relevant data that can quickly and easily augment a care visit. AI can also improve nurse advice services as well as provide physicians with patient-preferred language for answers to their questions. In a recent research study published in the JAMA Internal Medicine journal, researchers concluded that “a chatbot generated quality and empathetic responses to patient questions posed in an online forum. Further exploration of this technology is warranted in clinical settings, such as using chatbot to draft responses that physicians could then edit.”⁵

Improving care orchestration. AI has the potential for even greater value outside the single-visit interactions of patient and provider. Right now, care events that require multiple visits, providers, and/or treatments over a period of time require patients to manage complexity in addition to their own health needs. In the face of jammed practice schedules, they must orchestrate appointments with the proper provider in the right order and deal with a multitude of billing issues. These patients can often be found carrying a sheaf of folders or a binder in an attempt to keep their care more coordinated — an added burden to those already coping with cancers or other chronic illness, serious injuries, debilitating conditions, or difficult-to-diagnose maladies. At the same time, providers may be dealing with stale records that impede proper care. 

This problem is exacerbated by the reality that patients grappling with complex health conditions find themselves burdened with the responsibility of managing and coordinating their health data across various providers. This fragmentation not only complicates their care journey but also poses significant challenges for healthcare providers striving to deliver coordinated and effective treatment. The lack of data interoperability in the healthcare sector underscores the urgent need for a unified approach to patient health records, ensuring that all caregivers have access to a comprehensive, up-to-date view of a patient's medical history, regardless of where the care was provided.

If this sounds like a recipe for poor experiences and outcomes, it is. Patients often become overwhelmed and simply disengage. According to a study published in the National Library of Medicine, 42%⁶ of people never make the first appointment, and 70% of those that do, don’t make the specialist follow-up. Other issues include misaligned or mis-timed treatment and handoffs, poor adherence to treatment, and ultimately, a lower value received compared to the actual cost of care.

Putting AI into these workflows can contribute to better, simpler, and more successful care coordination for complex or chronic care events. This is possible through context-aware appointment scheduling and planning, individually tailored treatment plan roadmaps, and analyses of ongoing results plus awareness and suggestions for how that should change the trajectory of care. In these complex cases, AI can also provide for responsive change management when plans are disrupted — giving immediate, data-driven guidance on what the new options or next best steps are when an appointment is missed or lab results cannot be obtained by a certain time.

Influencing the trend towards outcome-based care. Taking an even broader view, use of AI has the ability to elevate its impact beyond the practitioner level to influence trends in healthcare. For example, many healthcare systems — including that in the U.S — remain stuck in a fee-for-service model, which focuses on interventions to treat symptomatic conditions rather than promotion of overall health. This orientation unfortunately reduces the focus on prevention and leads to both higher costs and poorer overall outcomes.

AI can help support the much-needed shift towards outcome-based care in two key ways. First, it can quickly and accurately identify at-risk individuals and populations earlier than is possible with current-gen technology and processes. AI can also identify and indicate wellness and prevention measures that offer the highest likelihood of success in a given situation or patient population.

Successful AI Implementation

Healthcare organizations that are motivated by the benefits of AI, but stymied by the process of incorporating it, do have viable options. It can help to view the adoption of AI as yet another step in the continuous march of digital transformation. Since most companies have been engaged in this process already, they can apply the benefits of proven practices to their AI iterations. The easiest and most efficient way to do that is to partner with outside experts. Here are three critical elements to look for if you’re considering taking this route.

1. A commitment to collaboration: Every healthcare organization will have different needs for both technology and process when it comes to incorporating AI into their business. That’s why it’s especially important to partner with experts who insist on a collaborative approach. This should be infused throughout the entire project, from evaluation to planning to implementation and beyond. Successful execution requires that everyone — internal and external — have an informed stake in the project.

2. The expertise and ability to affect both technological and cultural change: Some firms are well-versed in the latest technology options. Others are more process-oriented when it comes to making changes. Because you will likely be dealing with issues on both these fronts, look for a partner who can handle both with equal aplomb.

3. Support to establish self-sufficiency: Many consultants are happy to keep you on the hook long term for support and any future changes or additions. But one way to beat the technical skills shortage and keep costs down in the long run is to partner with an organization who believes in “teaching a person to fish instead of just giving them a fish.” In other words, you want to work with experts who happily can and will train your own internal staff to support your AI program for themselves in the long run.

The Time Is Now

The AI edge, as with many innovations, will go to those who seize the opportunity early. While the technology is still evolving, the benefits of adoption are clear. For those who need help moving forward, CI&T is here for you. We approach AI with the same strategy that has successfully supported so many digital transformation clients for nearly three decades — helping them become more collaborative, integrated, and outcome focused. If you’re ready to start your AI journey, CI&T is ready to help. Visit our website today to learn more.




Citations

---