From Amazon Alexa to automated customer service chatbots, artificial intelligence has made strides into industries such as retail, banking, and automation. With the emergence of ChatGPT making this technology even more accessible, there has never been more buzz around the potential of AI, particularly in industries such as healthcare. In fact, new research revealed that broader adoption of AI in healthcare could save the U.S. up to $360 billion annually.
The reason this number is so staggering is because of the current inefficiencies impacting the healthcare system. Although healthcare is intended to be standardized, it is not, with processes such as clinical documentation ranging across facilities. Not only is this documentation not standardized, but it is also time-consuming for providers, who are spending an average of 4.5 hours a day logging information into electronic health records. Leveraging AI tools can help to automate and standardize clinical processes, streamline workflows, manage patient logistics, and most importantly, allow providers to spend more time with patients. It also can result in immense savings for healthcare organizations, especially when it comes to fee-for-service care.
Barriers to Success
Despite its potential to transform healthcare, there are still barriers to this type of technology becoming more widely implemented across the industry. Like other new technologies, the trajectory of artificial intelligence and machine learning (ML) has followed the traditional hype cycle. Innovators started by trying to apply AI to some of the more complex problems, and when thinking about a complex problem in healthcare, the first thing that comes to mind is typically a complex diagnosis. Not only is this use case extremely high risk when it comes to patient outcomes, but innovators also need to navigate stiffer regulatory requirements. Likewise, it is hard to validate the results from these models, which are being done in small sample sizes.
Current Implementations
Instead of applying AI and ML to the more complex problems, innovators are learning from previous mistakes and are working to apply this technology to the more basic challenges facing healthcare. This is the most successful way to navigate the innovation space and curtail some of the barriers to wider adoption. Some examples of successful AI implementations in healthcare to date include:
Tissue Analytics: One area where AI and ML technology is being implemented effectively is wound care. Through state-of-the-art machine learning and computer vision algorithms, non-specialists can measure wounds in a systematic manner through a secure, HIPAA compliant smartphone app. Models can then predict how well a wound will heal and the estimated timeframe, or if the wound is at risk for deterioration. This technology seamlessly integrates into the EHR to save on documentation time, improve clinical workflows, and improve patient outcomes.
Appointment Scheduling: Another area where AI is successfully being leveraged is scheduling. Artificial intelligence can be used within scheduling platforms to predict how likely it is that a patient will miss their appointment. This works to maximize clinical productivity and ensure revenue optimization – two key elements in a fee-for-service arena.
Prediction of Health Risk: Algorithms can be developed based on de-identified patient data within electronic health records to predict the likelihood of health episodes, such as whether a diabetes patient is at risk for an amputation or a patient in long-term care facility is at risk for pressure injuries, falls or rehospitalizations. Being able to identify these types of risks can not only improve patient outcomes but also result in cost savings for healthcare organizations from avoidance of penalties from Medicare and health insurance companies.
Looking Ahead
Despite some barriers to adoption, wider implementation of AI technology will be instrumental for shaping the future of healthcare delivery. In addition to the financial benefits of this technology, utilizing artificial intelligence and machine learning to augment the quality of care for patients and the efficiency of operations for providers will result in improved patient and provider satisfaction for a better healthcare system for us all.
Josh Budman
Joshua Budman is a technologist specializing in machine learning, computer vision and the development of integrated healthcare applications. He holds a B.S. in biomedical engineering from Johns Hopkins University and received his MSE in biomedical engineering from the Johns Hopkins Center for Bioengineering Innovation and Design. Josh currently serves as the SVP of Research & Incubation for Net Health. Previously, Josh served as the CTO of the digital health startup, Tissue Analytics, where he helped integrate his company's AI-based digital imaging product with many of the major electronic medical record systems. Tissue Analytics was acquired by Net Health in the spring of 2020.
