Precision education in medicine: A necessary transformation to better prepare physicians to meet the needs of their patients

Across the continuum of emergency medicine (EM) education, physicians strive to continuously develop their skills while navigating multiple demands. To achieve the aim of learning amidst exponential growth in medical knowledge and increasingly complex medical care, a new system of accessible, personalized, and continuous learning is needed. In this commentary, we describe the model of precision medical education (PME),¹ which includes using data and technology to transform lifelong learning by improving data inputs, personalization, and efficiency.

Innovation creates transformation in medical education. In other spheres, the arc of innovation empowers users and builds value.² For example, Amazon shifted purchasing power from local stores to consumers.³ Netflix transferred power of choice to viewers, creating an industry for asynchronous content. Google shifted power of information from the few to many.⁴ Generative AI (artificial and augmented intelligence) similarly has shifted the ability to gain and apply knowledge from experts to the people. While it takes time to fully realize their potential, these innovations largely meet the needs of consumers and society by shifting the locus of control to the end users. We believe medical education should create similar transformational shifts for learning to bring the locus of control to individual—student, emergency medicine (EM) resident, and practicing physician in the arc of lifelong learning.

The goal of training is to produce an EM physician workforce capable of delivering high-quality care to patients and communities. Explosive growth in medical knowledge and remarkable procedural advances have underscored physicians’ need for continuous and effective lifelong learning. The need to make this learning simple and accessible so that it weaves within existing workflows is also an imperative. Yet how do physicians maintain and advance their knowledge?⁵ The gap between the need and process of learning, including resources, time, and methods, can contribute to the challenges of ongoing learning, contributing to burnout and moral distress as physicians struggle to keep up.

Medical education for medical students, residents, and practicing physicians has not evolved sufficiently with the pace of change in technology and remains encumbered by inflexibility, inefficiency, and inequity. This gap enhances the struggle to meet the current and future needs of physicians.⁶ There is little emphasis on the process of lifelong learning or maintaining competency in the rapidly expanding universe of medical knowledge and new procedures.

Because of resource constraints and the need to deliver training at scale, structured medical education (undergraduate medical education, graduate medical education, and continuing professional development [CPD]) are experienced as “one size fits all.” The same curriculum is delivered to all regardless of context, individual knowledge, or patient care needs. Additionally, CPD orients to hours to complete the session or to obtain recertification versus knowledge or skills needed. It becomes performance of a “check the box” approach with substantial frustration without advancing patient-relevant lifelong learning. The system, trainees, and practicing physicians would all benefit from increasing the effectiveness of learning by increasing personalization, efficiency, and power for the individual physician.

PME is a system using data and technology to transform lifelong learning through improved personalization, learning efficiency, learner/physician agency, feedback, and ultimately, to improve patient outcomes. Building on the model described by Triola and Burk-Rafel,^{1, 7} Desai et al.¹ proposed PME involving a process cycle (Figure 1). This cycle builds on the quality improvement cycle of plan–do–study (check)–act and the Master Adaptive Learner model with plan–learn–assess–adjust.⁸ The model can function at the individual, program, or organizational and system level. For the individual physician, PME starts with data input sources—proactive data aggregation will be driven by data gathered from a variety of sources including the electronic health record, patient care outcomes, patient panel data, population data, health care quality measures, physician educational activities, and clinical practice patterns. Identifying disparities in care based on patient characteristics is also important to analyze and include as an input source. For EM residents, data inputs might include assessments, return visits, radiology, antibiotics, and narcotic ordering patterns.

From the analysis of data, we develop insights and understanding of the knowledge and skills gaps. Informatics will enable personalization by leveraging these multisource inputs into meaningful individualized feedback. Insights might include gaps between individual practice and changing practice guidelines or training gaps (e.g., never performing cricothyrotomy or never diagnosing or treating thyroid storm).

Actions and interventions follow—combining these data with individual characteristics and preferences, just-in-time educational programming from high-quality sources such as published research and guidelines, and benchmarking information, will help guide physicians’ development. Ideally coaching will help interpret feedback and increase participation and personalization. For example, by review of resident patient exposures (gap), specific training focused on these gaps can be implemented.

Outcomes including assessment of learning, physician performance, patient outcomes, and evaluation of interventions will provide outcomes for individual and program-level feedback. All information then feeds back into the cycle to provide data as inputs for additional insights, learning, and improvement. Further, issues of disparities of health care can be observed and addressed through interventions in physicians’ education. This cycle of PME system will promote an adaptive learning⁷ culture and help address the stress of practice and lifelong learning.

Figures 2 envisions a future context where the system of PME is developed and assisting physicians with lifelong learning and improved patient care. While some of this illustration is in the future, pilots are already being developed in PME and we anticipate that generative AI technologies will accelerate these efforts. Schaye and team⁹ are developing natural language processing to assess and provide feedback to residents on clinical reasoning. This system will review each resident's note and assess it for clinical reasoning, providing that information on a resident dashboard where the resident can link to the specific note in the patient chart and thereby seeing where they might improve. The dashboard will also provide summary data to record improvements in clinical reasoning documentation and will be used as outcomes to further improvement.

Several EM teams are using EPIC metadata to inform residents and EM programs. EPIC provides “signal data,” which is metadata of how the electronic health record (EHR) is used by providers. For example, reports can be created to document how much time and where time is spent (notes, orders, tracking, disposition, and chart review). Similarly, graphs can show how a resident places most of their orders (individual orders or order sets) and smartphrase/macro usage compared to others in the department. These inputs provide data for providers to help them understand their process of work and efficiencies. From the insights, they can adapt their approach and monitor changes. Similar will need to reorderly, Warm et all are providing residents the outcomes of their patients.¹⁰ Schauer builidng on this are exploring the relationships of patient outcome measures and the EPIC metadata (user use patterns) such as efficiency, inbox metrics, and workflow. While this work is with internal medicine residents, for EM residents, PME would allow residents and faculty to link efficiency metrics EHR usage, to patient throughput and importantly with patient outcomes. EHR-based nudges are offering timely recommendations and just-in-time resources germane to a patient the resident cared for that day or could receive content-specific resources based on gaps in their in-training examination performance or the clinical reasoning documented in their patient care notes.¹¹ Woodworth and team¹² are building a platform for anesthesia residents that aligns competency development, core knowledge, and patient exposure with learning resources to address gaps.

Haptics and wearable devices are other sources of data/inputs.¹³ Some programs use real-time location service trackers to collect resident location and how much time they spend in the patients’ room or in the work room.¹² In surgery there is a burgeoning use of haptics to assessment pressure, force, and hand movements and relate these to patient outcomes.¹³ While developing in surgery, we can see how this approach will be useful for EM procedural skills. For example, Phadnis et al.¹⁴ report on the use of EM haptic simulators for training of lateral canthotomy and thoracostomy. These novel methods of collecting data will provide additional inputs for precision education and will be able to provide outcomes as well.

These projects intentionally use data to provide feedback to physicians to adapt their learning and practice and the innovations leverage PME to assist physicians to improve patient care. PME can also operate on the program and organizational level. For example, American Board of Medical Specialties just funded a project to provide automated mapping of visit diagnoses to specialty board clinical domains for enhanced assessment, certification, and precision education that include EM mapping of clinical visits to the EM model of care.¹⁵ Kern et al.¹⁶ explored the relationship between number of patients seen and in-training examination scores. Hoxha et al. investigated medical students’ types of clinical encounters in the ED in relationship to Clerkship Directors in Emergency Medicine recommendations.¹⁷ Similar mapping was done for residents’ patients ICD-10 codes.¹⁸ These data provide macro and meso level information to tailor education at the program or specialty level. EM trainees, physicians, and education leaders can use these examples to identify sources of data and other PME approaches in their own contexts.

To accelerate this transformation in lifelong learning through PME, we must take advantage of the acute inflection point of growth in technology and analytics. Most EM programs do not have the necessary data infrastructures, and the transition cost to such systems can be imposing, especially for lower-resourced systems. Technological advances may outpace the effective integration of innovation in the learning environment. As the previous section demonstrated, there are pockets of PME in EM, but the issue is the difficulty scaling due to cost, informatics resources, and technology. As PME innovations are developed they need to be replicable, scalable, and accessible and will require significant resources.

Furthermore, a culture promoting a growth mindset and competency-based patient-centered education, as well as increasing emphasis on adaptive learning¹² and coaching, is needed to enable PME. The predominant culture frames learning in a deficit orientation, stunting a growth mindset. An entrenched normative orientation of assessment, coupled with a highly competitive selection process for each next phase of training, inhibits the focus on developmental needs. Branzetti and colleagues^19-21 have several papers exploring master adaptive learner and adaptive expertise that can help EM promote growth mindset. Leveraging this work, as well as coaching, the use of Individualized learning plans, and other shared resources, will help support PME in EM.²² Paying attention to the affective and emotional responses to PME will help effectiveness.

Finally, additional research will be needed on the effectiveness of a PME system, including specific outcome measures related to learning and performance improvements as well as how and why impacts are or are not occurring. For example, there are concerns that AI may incorporate existing bias, in the outputs. Thus, as PME is developed, exploration using a realist evaluation lens of … what works for whom, when, and in what circumstances, so as to identify how PME is working and possible biases that are created. The goal is to empower current and future work that is ongoing to address current educational system shortcomings through PME.

In conclusion, PME facilitates moving the power of learning to physicians to maximize the value and effectiveness of education. There are several next steps. First, leaders of health systems and educational programs should invest in systems of PME collaboratively and strategically to leverage resources and impact change more effectively. Second, the EM educational community should explore and expand current PME pilots. Third, data exist in medical education; using these data to drive the PME cycle with insights, action, outcomes supported by coaching, master adaptive learning, and growth mindset will start us on the way. With this transformation, the medical education system will become more personalized, efficient, fair, and effective—and ultimately allow physicians to care for patients, families, and communities more capably.

The authors declare no conflicts of interest.