5 tips to implement artificial intelligence in health care organizations successfully

© Toowongsa - stock.adobe.com

Given the growing awareness of how powerful and useful artificial intelligence (AI) can be, health care organizations have good reason to try to leverage its vast capabilities. But figuring out how exactly to use this cutting-edge technology to address primary care-specific needs is no small feat.

Meanwhile, the development and adoption of AI aren’t occurring in a vacuum.

Primary care itself is changing in the United States, especially because of the growing adoption of value-based care (VBC). This shift pushes primary care physicians and other employees to deal with patient data in new (and often time-consuming) ways, especially because of the importance of hierarchical condition category coding and risk adjustment. And many health care organizations operate in both value-based and fee-for-service models simultaneously — creating even more logistical and administrative challenges.

In this health care environment, it’s important to implement advanced technologies in ways that reduce the administrative burden on primary care providers and help them offer patients the best care possible. As your organization implements AI, there are five key steps to take in order to make sure that technology simplifies the transition to VBC for your primary care providers.

Step 1: Make sure your adoption of AI has a specific objective

Ronen Lavi
_{© Navina}

First of all, it’s crucial to have a clearly defined objective that will drive your implementation of AI.

Because AI’s potential is so wide-ranging, it’s important to have a specific problem that you want to solve. If you aim to bring the power of AI to the world of primary care, then your main objective should also be specific to primary care.

In addition to being clear and specific, your objective should be concrete enough for your success to be measurable. For example, you might want to reduce the amount of time primary care providers spend on documentation and other administrative tasks. Not only will choosing a measurable objective help you sharpen the focus of your health care organization’s use of AI, but it will enable you to use hard data to track your success and identify areas especially ripe for improvement.

Step 2: Choose tech that is well suited to meet your objective

Because primary care has very different technological needs from other fields (and even from other areas of health care), it’s important to choose an AI-powered solution that meets those needs. Clinical tools built exclusively on generic large language models will not be able to address complex clinical use cases adequately. To get the required accuracy and precision for a clinical setting, health care organizations need health care-specific AI tools, which can either be homegrown from the ground up or undergo significant fine-tuning.

When considering an AI solution, it’s also important to look into the vendor and make sure you’re a good match. Pay attention to the vendor’s track record, references, how long it will take to deploy the technology fully, and how disruptive its rollout will be to your organization. And you’ll want to find out whether medical professionals played a central role in designing the product, applying their real-world experience to make sure it meets the day-to-day needs of primary care providers.

In addition, you should think about the degree to which a given vendor is willing to work with you rather than just offering an out-of-the-box, one-size-fits-all solution. This could include working with your organization to design an AI solution that meets your particular needs, offering comprehensive customer support, and helping train your organization’s clinicians and other team members to use this technology.

As you shop around for an AI-powered tool, it’s worth getting input from your clinicians. By letting them participate in choosing the technology that they’ll use, you can get valuable insight into what it takes for a solution to help them concretely. Involving them in this process can also help you encourage them to use the technology once it’s deployed.

Step 3: Integrate your AI smoothly with your existing technology

To make your organization’s work easier and not harder, an AI solution should supplement your existing technology rather than requiring you to replace it. It’s especially important for health care-specific AI to work well with electronic health records (EHRs).

EHRs are a central part of health care organizations’ existing technology, and integrating deeply with them can make a big difference in an AI-powered solution’s ability to provide useful information to clinicians efficiently. It will be especially helpful to have your proprietary data feed your AI solution, so that the AI can derive insights from organizational knowledge rather than just from publicly available information found online — improving the accuracy of its conclusions. Your solution should also provide clinical documentation for any conclusion that it reaches, giving you hard evidence to help ensure that its machine learning-driven analysis is accurate.

In addition, it’s critical for an AI-powered solution to cause minimal disruption (or, preferably, none at all) to patient care during its deployment. As much as AI can help primary care providers, patients are primarily interested in getting the best health care possible right now.

Step 4: Get your clinicians on board (and onboarded) effectively

To implement AI successfully, you’ll need to take concrete steps to maximize its adoption among your organization’s primary care providers. Keep in mind: Clinicians can be skeptical of new technologies, and some might fear being replaced by AI. In addition, some of them might have concerns about data security.

To address their hesitations and worries, they’ll need to see the value of your AI. Not only will your organization need to calm any fears they might have, but you’ll want to get to a point at which they’re excited about the benefits that AI will offer both them and their patients. To make that happen, you’ll need to communicate with them effectively, listening and addressing their concerns in detail.

Training is also critical in maximizing the adoption of AI within your organization. For example, providing clinical evidence for any conclusions reached by the AI engine can help address clinicians’ skepticism about its accuracy, so it’s important to make sure they know how to access all relevant documentation within your AI-powered system. When making plans for training, make sure to consider how tech-savvy your primary care providers are.

In addition, some clinicians might need training relating to VBC and risk adjustment. And if there is a language barrier that risks hampering their use of your AI solution, you should make sure your training addresses their linguistic needs.

Step 5: Use analytics to make sure your AI is used widely and well

To maximize and optimize the usage of your AI-powered platform, you need insight into when, where, how and by whom it is (and isn’t) being used within your organization.

If your AI-powered platform has robust analytics, you can use data to identify and address shortcomings in the use of that technology throughout your organization. Not only is it important to consider these analytics capabilities when shopping for an AI solution, but it’s crucial to actually use them effectively.

That brings us back to the primary objective driving your organization’s implementation of AI. If you’re approaching AI strategically, then your technology is there to achieve a specific outcome — and so are your analytics. To help you make the most of your AI, it’s critical to examine how widely and effectively that technology is being used and then use those insights to achieve quantifiable improvements. That requires you to consider which metrics will provide you with the most valuable information, offering insight both into the degree of engagement with your AI tool and into the impact that engagement has on your organization’s key performance indicators.

Adapting to a changing landscape

Taking all of these five essential steps is not just a matter of tapping into the capabilities of artificial intelligence but of using that technology to meet the evolving needs of primary care physicians and their patients.

As AI continues to develop and its usage within primary care continues to grow, it’s worth keeping in mind the intersection between those technological trends and broader changes in the U.S. health care system. First and foremost, that means looking at AI within the context of the shift toward VBC — a transition that brings with it significant increases in administrative work (among other difficulties).

In this challenging situation, today’s AI can make a powerful difference for primary care providers. By methodically taking the five steps laid out above, health care organizations can help those providers adapt to the demands of VBC — reducing their administrative burden and empowering them to offer the best care possible.

Ronen Lavi is the CEO and co-founder of Navina, a physician-first AI platform that helps health care organizations succeed in value-based care. Ronen's career spans more than two decades in the Israel Defense Forces (IDF), where he received the prestigious National Security Award in 2018. Since then, Ronen has been channeling his expertise of leveraging complex, multi-modal data for real-time decision-making into solving the most complex data challenges in primary care, with a broader vision of revolutionizing the health care sector with AI.