As a health care industry executive who's been deeply involved with artificial intelligence (AI) since its early days, the emergence of ChatGPT is a pivotal moment in the evolution of the industry.
For years, AI has been the great but unrealized promise of industry. It’s been through every aspect of the hype cycle. But with ChatGPT, almost overnight AI went from jargon to viral. It is a thrilling moment, as it creates an opportunity for quicker adoption of AI in other sectors as well. When looking at health care, AI creates new opportunities both in general, and more specifically for the realization of value-based health care.
The challenges physicians are facing today
In many ways, value-based health care is a “back to the future” moment, helping family physicians return to the essence of primary medicine: it rewards them for providing high quality, proactive, empathetic care to their patients.
To be successful in the value-based model, physicians must be able to quickly, thoroughly, and effectively execute on three key actions:
- Take into account their patients’ full history
- Identify preventive care opportunities
- Fully capture the patient's health status on an ongoing basis
Problematically, the software most physicians are using predates the movement to value-based care. As a result, identifying and documenting the real risks and required preventive actions requires wading through a multitude of different screens and functions within the current EHR platforms.
This means being able to assess and act on exponentially growing amounts of patient data at precisely the moment physicians need it takes up a huge amount of time both before and after the encounter, detracting from the patient visit. So, demanding is the burden that this work often extends late into the day, at the expense of family time, and with the burn-out consequences we are all too familiar with.
AI’s role
In simplest terms, AI refers to machines that are designed to think and act like humans. Their intelligence is based on algorithms and models that enable them to learn how to make predictions and decisions from data. The more data they’re fed, the “smarter” they get, becoming more accurate with time. With advancements in technology and cloud computing, AI has become cheaper to develop and run, making it more accessible and useful even outside the technological fields from which it originated.
In health care contexts, AI can be used broadly and effectively, even now but certainly as machine training improves. That includes everything from suggesting diagnoses based on symptoms, to predicting a patient’s improving or worsening outcomes, to reducing the administrative burden on physicians. Within the broad field of AI, there are four subfields that can be especially beneficial to physicians in their value-based care journey.
- Contextual AI systems are trained to look at the whole picture when making suggestions, in the manner of human perception. When it comes to delivering on value-based care, this can mean prompting the physician to ask specific questions beyond clinical markers, to include those based on social determinants of health or family history, and—based on all those inputs—surfacing critical information at the right time. The result is a reduced burden on physicians who no longer need to sift through large amounts of data to find the information they need.
- Generative AI is used to generate new and original content or data based on existing frameworks, patterns, inputs, and examples. Within health care, these models can revolutionize the way patients and clinicians interact and access information, by providing up-to-date medical information, supporting clinical decision-making, and aiding in medical documentation. In a value-based strategy, generative AI can surface new insights from a myriad of structured and unstructured sources and bring light to complexities that may not be readily apparent to a physician going through a patient’s chart.
- Explainable AI gives users the reasoning behind the decision or prediction, ultimately leading to more informed, efficient decision-making. Perhaps the most time-consuming task for physicians in value-based care is ensuring patient conditions and complexities are properly documented. Yet with the glut of data they’re faced with every day, physicians often miss diagnoses. AI can make diagnosis suggestions based on the patient’s history, but without explainable AI, physicians have no way of knowing whether to trust those suggestions. Explainable AI, which points to the source and shows why the suggestion was made, allows physicians to make informed decisions with the clinical context at their fingertips.
- Predictive AI models use past data and analytics to predict future outcomes. In health care, this can mean predicting worsening conditions for a hospitalized patient, anticipating risk of future ER admissions, or using lab trends to predict increases in chronic condition complexity. In value-based care, predictive AI can be used to improve clinical outcomes by creating opportunities for proactive, lower cost care. Predictive AI will let us get and stay ahead, which will make preventive care broadly and affordably available.
The path forward
We can only achieve higher quality, lower cost health care by augmenting the physician-patient relationship with AI at the point of care. AI engineered specifically for clinical teams, by those with insight into both software and medicine, can move mountains.
Having worked on AI platforms for decades, I know there is a lot of work ahead to better tailor them for health care. Statistical models, especially those generated by machine learning based on population health data, require caution so that bias and inaccuracy do not lead to misleading and incorrect outcomes. Many systemic inequities are so embedded in our existing data that they need to be identified and corrected for. But that said, there is only so far that algorithms can take us.
There is no replacement for training and human judgment, but purpose-built AI developed with and for primary care can augment physicians’ ability to provide and be reimbursed for care. At Navina, our goal is to support the value-based process by using AI to make sense of chaotic patient data, in turn reducing the burden on physicians and improving clinical outcomes.
About Navina
Navina is the physician-first platform for your value-based journey. Using AI to turn chaotic data into an organized patient portrait, Navina gives coders, pre-visit planning teams, and clinicians a better understanding of their patients’ health status, and provides actionable risk adjustment and quality insights at the point-of-care. The result is an improved physician experience, better quality of care, and stronger value-based outcomes.
About the author
Ronen Lavi, co-founder and CEO, Navina, spent 24 years in the Israel intelligence corps, where he established and led the AI Lab of Israel’s Military Intelligence.
The Lab collaborated with leading tech companies and academia to develop cross-functional platforms that provide insight into areas challenged by diverse and complex data. In 2018, he was awarded the National Security Award. Lavi is now employing his skills and talents to revolutionize primary care—and through that, health care in general.