From burnout to better care: Tech, ‘personomics,’ and the human ROI of AI in clinical workflows

Artificial intelligence (AI) is finally moving past the theoretical hype and taking on a practical role in the delivery of health care. The industry has historically operated on a promise of digital transformation, but now, that transformation is real. AI is visible in reshaping clinical workflows, clinician time, patient engagement and systemwide return on investment (ROI).

As health care leaders with decades of experience spanning emergency medicine, family medicine and cardiology, we’ve seen this shift firsthand across the entire care continuum. The conversation for physicians has moved from whether AI will impact care to how it can be effectively deployed to be beneficial.

For primary care physicians, the goal is not to replace human judgment with an algorithm, but to use AI as a tool to manage the crushing burden of administrative tasks and restore focus to the patient. To successfully achieve this, we must understand what responsible, clinician-led AI looks like in practice.

The modern crisis: Too much information

A century ago, health care suffered from a lack of data, but today, we face the opposite problem. Between the immense volume of patient data in the electronic health record (EHR) and the constant stream of new medical literature, physicians are drowning in information with a shrinking window of time to sift through it. This forces us into an impossible choice: Do we spend our limited time with patients meticulously reviewing every detail in the chart or fully engaging with a patient to hear their story?

Without tools to properly handle the flood of data and never-ending administrative work, this frequently causes us to lose sight of the patient as a person. Whether in the high-stakes environment of the emergency room or the busy halls of a family practice, we’re often hindered by the finite amount of time available to connect with patients and find the signal in the noise. The solution is not more data, but better synthesis.

The solution: Ambient intelligence and clinical decision support

To address this challenge, two categories of AI are currently making a tangible difference in the clinical workflow: ambient intelligence and clinical decision support (CDS).

For the primary care physician, the most immediate relief comes from ambient AI’s ability to help the burnout crisis that’s driven by administrative overload. Study results show that physicians spend more time in the EHR than with patients. This impacts their ability to listen attentively and communicate with patients, often intruding on personal time after normal work hours. To help with this issue, ambient AI tools act as a silent partner, scribing conversations and automating charting in the background — freeing up time that can then be dedicated to preventive care counseling and meaningful interaction. In fact, ambient AI helped 93% of doctors give their patients their undivided attention.

Parallel to this is AI-powered clinical decision support (CDS), which provides doctors with evidence-based guidance at the point of care. For example, in the time-compressed environment of the emergency room, we strive for highly reliable systems to help improve outcomes, create efficiencies and reduce medical error. AI-driven CDS helps us achieve high reliability by adhering to the five rights of CDS: providing the right information, in the right location, at the right time, in the right format, for the right clinician.

When AI handles the heavy lifting of evidence synthesis, it allows the physician to focus on what a machine cannot: connecting compassionately at the human level and engaging in shared decision-making conversations that help inform an individual patient’s choices and care plan.

The ROI: Financial health, the human connection and ‘personomics’

In addition to financial metrics, the ROI for AI in health care should also be measured by human impact. From an economic standpoint, when physicians are empowered with the best available knowledge at the point of care, they are better equipped to deliver care strategies that maximize health, reduce harm, and ultimately prevent unnecessary hospital admissions and readmissions, which remain the most expensive event in health care.

Additionally, AI can identify subtle yet clinically meaningful patterns in large data sets that the human eye may miss, such as EKG signals that predict the development of heart failure. This helps pivot the focus of care from treating disease once it presents itself to upstream preventive care that helps preserve wellness, reduce adverse outcomes and lessen the health care expense.

However, for the individual physician, the ROI is found in what we call “personomics.” While precision medicine focuses on a patient’s biological makeup (e.g., laboratory data, genomics, proteomics), personomics accounts for the unique personal characteristics and life circumstances that influence how an individual experiences illness and reacts to treatment. In this case, AI helps bond our most advanced scientific capabilities with our deepest human capacities. By surfacing patient-tailored insights and evidence-based guidance quickly at the point of care, AI helps physicians reclaim the time they need to understand the patient as a person and build a genuine human connection.

The essential guardrails of AI

AI cannot be a “set it and forget it” technology. This is especially true in independent practices, where there’s not an IT department to troubleshoot a glitchy algorithm or a secondary administrative layer to catch errors. In these lean environments, a poorly governed AI tool often actively disrupts care. For AI tools to be successful, they must be the following:

• Transparent: Current AI models are not yet perfect enough to be considered highly reliable without human vetting. Just as most people want to know the ingredients in their food, physicians must know the information sources from which AI responses are derived. Without this transparency, a physician can’t be positive that the results are replicable and safe for use.
• Trustworthy: Responsible AI must be built on a rigorous governance process that guarantees equity and other best practices are baked in from the start. This means actively screening clinical research to ensure inclusive representation to prevent systemic bias. Additionally, it must base content on vetted medical literature to ensure all recommendations are rooted in evidence.
• Seamlessly integrated: The technology must be well integrated so that it’s not disruptive to the clinical workflow. If a tool adds multiple clicks, needs a separate login or requires a clinician to leave a program, it creates unnecessary cognitive load and wasted time. AI must blend into the workflow as an assistant so that it’s a genuine time-saving partner and not just a frustrating extra step.

From burnout to better care

The true value of AI in health care extends beyond financial and clinical benchmarks. It’s defined by time saved, clinician well-being and superior patient outcomes. Because bad care costs money, the incentives are finally aligning: What is financially beneficial is also what is best for the patient. But for AI to deliver a true ROI, we must approach it with humility and clinician-led guardrails. Only then will it transition from a disruptive buzzword to an indispensable layer of the care continuum — one that allows physicians to return their focus to the patient rather than the paperwork. This is how we restore the joy of medicine: by using technology to make health care more human.

Roy Ziegelstein, MD, MACP, is editor-in-chief and chief medical officer of DynaMed, a leading clinical decision support solution with a mission to provide the most useful evidence-based information to healthcare professionals at the point of care to improve health outcomes worldwide. Katherine Eisenberg, MD, PhD, FAAFP, is senior medical director of DynaMed. Scott Silvers, MD, FACEP, is chief clinical innovation officer of DynaMed.