Healthcare experts estimate that augmented intelligence (AI) in healthcare will increase 40 percent by 2028. With AI growing at such a rapid rate, it is critical that health systems avoid the common AI pitfalls that impede success. By addressing the three “Hows” that set the foundation for meaningful AI in healthcare, organizations can benefit from AI in minutes instead of months. In his second podcast, Dr. Jason Jones, our Chief Analytics and Data Science Officer, explains how AI intersects with other digital healthcare trends, how organizations can effectively integrate AI into existing workflows, and lastly, how health entities can create easy-to-use AI for team members across all domains.
Download
Technology experts estimate that AI in healthcare will increase at a 40 percent compound annual growth rate from 2021 to 2028. However, Jason Jones, our Chief Analytics and Data Science Officer, defines AI in healthcare differently than other experts. In his first podcast installment of Owning the Future of Healthcare, Dr. Jones explains the role of augmented intelligence (AI) in healthcare—instead of the more familiar term, “artificial intelligence.”
In the second installment, Dr. Jones expands on the role of AI in healthcare’s digital transformation and answers the three “Hows” critical to AI success:
Healthcare today is more digital than ever. Clinicians, analysts, and financial leaders alike turn to technology to ease paperwork burdens and improve operational efficiency. In this digital healthcare landscape, AI plays a critical role because it fuels other digital healthcare domains, such as interoperability, personal data collection, and telemedicine or on-demand care
For example, telehealth existed long before the pandemic, however COVID-19 accelerated virtual healthcare by over 40 percent. Telemedicine allows clinicians to connect with patients via technology without an in-person meeting. While this arrangement already has many benefits—including adhering to social distancing during a pandemic and not requiring patients to miss work for appointments, AI can make telemedicine even more efficient. For example, an organization could create basic questions in the triage process to identify patients with a urinary tract infection (UTI). Based on the answers to these questions, AI can quickly identify the patients with a UTI, allowing them to get treatment sooner and avoid long wait times at the urgent care or emergency department.
Dr. Jones says there are two keys to effective AI integration: Defining the goal that AI can help the system achieve and establishing trust in AI. Without a goal, futile attempts leave organizations frustrated and sometimes lacking trust in AI. A clear goal keeps AI efforts focused and helps team members understand how to measure their success.
Trusting AI in healthcare can be tricky. Predictive models will get the answer wrong, but that doesn’t mean the team gives up on the model. Instead, when the model gets the answer wrong, a team member should review the model carefully and ask, “Could I have done better?” and “Is this helpful to me?”.
Organizations must make it clear to team members that the objective of AI isn’t to be perfect. In fact, sometimes an end user’s greatest opportunities to improve trust occurs when there’s an error and the team identifies how to improve the model.
When pursuing AI in healthcare, many organizations skip a very critical step: identifying end users in the organization and what they are trying to do. End users are typically a variety of different users, including engineers, clinicians, and financial analysts. For example, an IT individual will have a different set of needs from a data scientist, who has different needs from a clinician.
Understanding each end user, what they are looking for, and how AI can help them in their specific role is critical to setting AI up for success. Detailed end-user information can also help separate the noise from the meaning in all the data, as it gives specific guidance about which insight can help which persona.
Another benefit of internal persona research is a deeper understanding of team members’ AI knowledge. This can keep the data science team from creating complex models that require AI expertise and keep them focused on simple, easy-to-use models. Practical models allow users to quickly recognize if they make good or bad decisions without understanding the complexities of an algorithm. As such, they can focus more on implementing change based on accurate insight from the model.
With AI in healthcare growing more common, it’s time for organizations to accept and embrace the capability by applying it in their health systems. With expertise and simple predictive models, health systems can successfully implement and scale AI, delivering accurate insight to decision makers in minutes instead of weeks.
View the full podcast.
Would you like to learn more about this topic? Here are some articles we suggest:
