Overcoming Barriers to Integrating AI Into Health Care Processes

Jason Wille, CTO at Aeroflow Health, has over 25 years of technology experience in various industries, including health care. In this interview, he shares his expert insights on the benefits and challenges to providers, administrators, and patients of integrating AI into the health care processes. 

What factors do you believe contribute to the limited adoption of AI tools in the health care sector, despite the increasing interest and likelihood of adoption reported by executives?

As in most industries, there is a lot of excitement around AI and the possibilities it can bring to the workplace and patients. That being said, from my perspective, data privacy, security, and regulatory hurdles present the biggest challenges to AI adoption in health care. Due to the sensitivity of patient data, it is critical to ensure privacy and security, so it makes sense to be optimistic about the potential of AI in health care and also proceed cautiously.

In addition to the above challenges, implementing AI solutions takes a whole new set of talent and resources and is a large, long-term investment. This can present challenges to smaller health care organizations unable to set aside the appropriate budget.

How do you think the complexity of integrating AI into existing health care systems affects its adoption rate? Are there specific challenges that organizations face in this process?

AI integration in existing health care systems is a very large challenge. The wide range of health systems and data formats creates a rather large barrier for adoption. A large piece of any AI project requires data standardization and data cleaning. In fact, as much as 80% of project time is spent just prepping, cleaning, and standardizing data. Due to many legacy and disparate health care systems and this elongated timeframe to prepare a proper data set and standardize data, it is often difficult to find the budget and needed resources while also supporting daily operations and current business technology needs.  
 
Can you identify any barriers, whether they be technological, regulatory, or cultural, that hinder the widespread implementation of AI tools in health care? How can these barriers be addressed? 

There are a number of barriers, but in addition to what has already been discussed, we also need to consider the accuracy of AI and the impacts AI has on health care professionals and patients. Most of us have learned while experimenting with generative AI that the answers provided are not always accurate. The same can be true of any AI model, especially while it is initially being deployed and tuned. With this in mind, we need to adopt processes to ensure we have human controls in place to validate AI accuracy and ensure desired patient outcomes.

In addition to AI accuracy, it is critical that health care professionals and patients are ready to embrace the change related to AI in health care. It will be important to find initial AI use cases that improve patient experience and health care professional work life. It will take some clear AI wins to build confidence across the industry, and with those wins adoption will spread.

In what ways do you envision AI tools improving the health care sector? Can you provide examples of potential applications and their impact on patient care, operational efficiency, or cost reduction?

AI has the potential to revolutionize the health care sector, especially with patient care and operational efficiency. There have already been big advancements in using AI to scan medical images to assist radiologists in detecting anomalies and making diagnoses. We can expect this area to advance very rapidly as more and more medical imaging data is input into AI models. 

From an operational efficiency standpoint, analysis of vast quantities of unstructured document data is a big challenge for most health care providers. With the advances of AI large language models, I anticipate improved document processing to be an AI entry point for many organizations. If AI can assist with quickly scanning medical documentation to retrieve pertinent, in-context patient information needed for patient care, it will have a large impact on operational efficiencies. Not only will this reduce health care labor costs but also it will improve patient care and experience.
 
What are some potential risks or concerns associated with the widespread adoption of AI tools in health care, and how can health care professionals mitigate these risks while maximizing the benefits?

One of the biggest risks is ensuring the security of patient health information. Prior to using any AI tool, it is extremely important to understand how patient health information will be processed, stored, and secured. It is important to work closely with your information security team to use only approved AI tools and understand the limitations of that usage. In addition, it’s also important to train staff on AI limitations as risks to ensure protection of patient health information.
 
Collaboration between health care professionals and technology experts is crucial for successful implementation of AI tools. How do you suggest fostering effective collaboration and communication between these two groups?

In my experience, a successful technology implementation and transformation starts with common goals and objectives across business teams and departments. It allows for differences to be put aside and shared common goals for everyone to rally around. Frequently, the technology implementation is much easier than the change management required to get that technology implemented and functioning in production. With that in mind, it’s important to build out a project and communications plan and look for champions across the organization. With clearly communicated common goals and broad-based support across the organization it’s impressive how technology can enable an organization.