In all honesty, I am exhausted over the conversations regarding Artificial Intelligence (AI). It is the latest shiny new object. However, those of us that have been utilizing it are disappointed with the results. This displeasure is occurring for numerous reasons, beginning with the reality that we are early on, in the development of AI for healthcare. Additionally, we have not figured out how to integrate the information with our workflows. And we are also incredibly concerned regarding the implementation of what data drives AI within the context of health equities.
However, an additional postulation is that we are using AI for the wrong purposes. Currently, the focus centers on predictive analytics. How do we utilize all the original information to predict how a human being might act clinically in the future? We are doing this type of analysis on large data sets and translating the results to individual physiologic beings whom we don’t fully understand how they function. Moreover, environmental elements factor into the situation that are not part of the AI design. Undeniably, we are incredibly nascent in our knowledge so how can we expect AI to solve our problems by creating a prediction model?
I think instead, we need to consider changing how we regard the value of AI and its ability to aid us in executing better clinical decisions. One potential area is utilizing AI to assist and improve our diagnostic capabilities and connect us with others facing similar clinical dilemmas. “Crowdsourcing” clinical data and diagnostic pathways would be beneficial in our process of determining a navigation route between where we are today and a future diagnostic and treatment destination.
In other words, we could be utilizing AI to assist in finding better diagnostic results and successful treatment options and methods. Furthermore, AI could aid the clinician in deciding what additional data points are necessary for narrowing in on a direction that assists in needed care. Hence, the goal is for AI to be a guiding light for diagnosis and treatment versus alerting the clinician there might be potential issues in the future. Ideally, AI would be the ultimate navigation tool, not merely a predictive magic ball.
Unquestionably, what clinicians struggle with today is the cognitive load of new information and the ever-changing experiences of our colleagues. Let us pivot our implementation of AI to enhance our knowledge in real-time and utilize it to acuminate those diagnoses found to be accurate based on the constellation of data points immediately accessible. And finally, let us treat AI as an external “brain” for the clinician concerning treatments.
Diagnostic and therapeutic AI is the way forward. Let us use our newfound tools in helpful methods rather than merely informing us of negligibly beneficial information.