The Permanente Medical Group

Jing MacKenzie

With over two decades of dedicated experience, Jing MacKenzie stands at the forefront of healthcare innovation, weaving together her expertise as a physician, management consultant, and healthcare IT specialist. Her strategic acumen has been instrumental in driving forward initiatives in strategic planning, market assessment, vendor evaluation and clinical transformation across a spectrum of healthcare entities. MacKenzie’s involvement in evaluating industry startups for business cases and funding recommendations underscores her commitment to fostering growth and advancement within the healthcare landscape.

Through this article, MacKenzie delves into the transformative potential of artificial intelligence (AI) in healthcare and sheds light on the challenges of data quality, workflow integration and privacy that come along with the adoption of AI technology.

Dr. Smith walked in the busy corridors of an integrated health system that’s going through value-based care transition and faced a problem that was not new to her. She always struggled to reconcile the needs of her profession with her role as a seasoned primary care doctor who always wanted to provide top-notch healthcare services to all her patients. It was not easy for her because, with full appointment schedule, she also needed to complete the charts in EHR, respond to a never-ending long list of secure messages from patients that never seem disappear from her inbox as well as reach out to the patients due for preventive screens and management. Time pressed on as usual, yet she still yearned for a better way to accomplish all this work without feeling burned out. 

"With over two decades of dedicated experience, Jing MacKenzie stands at the forefront of healthcare innovation"

Dr. Smith is fictional, but her story repeats across the country every day.

During the chaos, the idea of a revolutionary solution became a buzzword—artificial intelligence, or AI as it is commonly known. What we were offered with AI was quicker diagnosis times and more accurate results; customized treatments which led eventually to better patient outcomes. Nevertheless, what came out during the incorporation process indicated that there would be multiple hurdles hindering us from attaining success while implementing AI.

The effectiveness of AI depends on the quality of data it relies on. De-biased and accurate data is in determining how well and true AI functions. Healthcare systems have a range of systems and data stores collectively constructing patients’ medical histories. When data is not good or is discriminative, which sometimes is the case due to the lack of upfront data validation processes or representation biases, it could lead the AI system to incorrect or unjust outcomes.

In the quest to unlock the potential of AI-powered diagnostics, healthcare professionals often embark on this journey with mixed feelings of excitement and caution. A stumbling block is that a lot of these systems lack transparency and explanation, which holds users back from full engagement. It is like navigating uncharted waters when you cannot understand why AI recommended this versus that, removing the practitioners’ capability to fully participate in the validation and decision-making process, thus hindering wider adoption. 

Implementation of artificial intelligence in the existing workflow can also interrupt established workflows and data collection. Use AI-based ambient listening as an example. Over the past two decades, the management of healthcare quality and efficiency has benefited from standardized documentation enabled by EHR, which feeds downstream analytics for population management and performance improvement. While AI-based ambient listening technologies free healthcare providers from note-taking and allow them to spend more time with patients, often missing is the next step of transforming free-text notes generated into structured and canonicalized data and therefore, disrupted is the established data flow to support downstream operations.

Privacy is important when it comes to safeguarding patient’s data, particularly about the use of AI technologies. As AI relies on vast amounts of data to tease out underlining patterns and insights, preserving confidentiality while still drawing the utility of data requires strong measures in place to prevent unauthorized access and disclosure. 

The rapid growth of technologies often outpaces regulations, and AI is a prime example of this challenge. Organizations adopting AI must stay informed about the rapidly evolving regulatory landscape and develop a compliance plan outlining approaches to, among other things, data privacy, security, bias mitigation and explaining AI-recommended decisions.

Adopting AI holds the promise of being revolutionary and disruptive. However, we undoubtedly live in a hype that has not yet fully understood the essentials and practical applications. Understanding both the promises and pitfalls helps us navigate more effectively as we continue to learn and integrate AI as a potential multiplier in the future of care delivery.