Photobank - Fotolia
Five pitfalls of implementing an AI initiative in healthcare
As more hospitals turn to artificial intelligence to improve patient care and daily operations, missteps like overlooking security and compliance can cause an AI project to fail.
Artificial intelligence has certainly taken center stage in the last 12 months, especially with ample examples of its success in different industries outside of healthcare. However, some hospitals and health plans have highlighted their own successes with AI initiatives around patient care and satisfaction improvements, as well as improved staff efficiencies. Given the number of available options to choose from that range from AI as customer service bots to detecting abnormalities in medical images, there is plenty of AI to go around.
Instead of hospitals embarking on the latest or hottest AI trend that is advertised in front of them, some in health IT recommend hospitals consider the lessons learned from early adopters to ensure their AI initiative is successful. Here is a list of common pitfalls that hospitals should keep in mind as they begin their next AI initiative in healthcare.
Embarking on too many AI projects at once
AI comes in so many assorted flavors that can benefit different departments within a hospital. A hospital must be aware that an AI initiative in healthcare based around medical imaging classification, scheduling or virtual assistants in patient rooms requires numerous resources. To increase the success rate of AI projects, IT and other stakeholders must focus on a small set of AI-based initiatives and see them through to ensure they are fully utilized and have met their objectives.
Focusing on the technology, not the outcome
While the technology used in AI can be intriguing to many in the technical field, the focus should remain on delivering the specific outcomes. IT and the members involved in the AI project must avoid getting distracted by the software tools and clearly define the success measures needed for the project. For example, if improving patient satisfaction is the goal, measuring success might include surveying patients before and after an appointment.
Making the AI initiative in healthcare just another IT project
IT typically leads the selection and implementation processes for technology initiatives. Projects related to infrastructure and other data center components may not require the participation of other departments within a hospital. However, when it comes to deciding on an AI platform, IT should not be the sole decision-maker. Input from other departments that will interact with the technology can help determine if the AI tool addresses their needs or those of patients.
Limiting the data AI can access
With the exception of a few AI-based tools like digital voice assistants or autonomous robots that deliver medications and clinical supplies within a hospital, most AI tools consume tremendous amounts of data to help it learn and improve the accuracy of decision-making. Some of the AI platforms available for clinical analysis require direct access to patient's charts and are able to analyze millions of patient's charts that may include labs, images, medical history and other relevant data.
Deploying AI technology without adequate compliance and security review
There are several concerns around privacy and security when it comes to several of the AI services in the marketplace. Depending on the AI technology itself and whether data is being transmitted out to a centralized processing server outside the hospital, security and privacy can't be overlooked. Tools like digital voice assistants can add value if they are rolled out within a hospital setting, but because of the way they capture voice data and process it outside of the hospital environment, as well as a general lack of HIPAA compliance, these solutions simply can't and won't be compliant in their current configuration.
Similar to any other project a hospital is engaging in, an AI initiative in healthcare will require the collaboration of several departments to ensure it is successfully implemented. As more hospitals implement AI projects, more lessons learned can further be shared among IT departments and other specialties. But for now, many are still discovering what AI can do for them.
