Robert Hester, PhD. and colleagues are developing a software simulation program for healthcare education, Osler One that enhances medical knowledge, professionalism, practice-based learning and improvement, ethics, and systems-based practice.
Clinical medicine involves caring for both chronic and acute conditions. Medical simulations routinely address acute conditions such as hemorrhage, heart attack, and pneumothorax, by focusing on crisis response and team building exercises. These simulations help trainees learn the correct practice for well understood mechanistic disorders, but they neglect patient management issues arising during long-term care. We are unaware of any products in the medical education simulation industry that addresses managing chronic health conditions.
Long-term care software simulations must address three goals: 1) long-term care and responsibility, 2) patient-centered problem solving skills, not fixed responses, and 3) provider-provider communication skills. The following describes a software project under development that addresses these goals. Once developed, modeled, and refined, the educational program can be packaged and utilized by any medical training program including undergraduate and graduate medical training.
Osler One, our educational program, will allow a longitudinal care experience for students, nurses, residents, and other health care providers. The system will operate in real-time so that students may follow patients over months to years. Osler One will utilize simulated patients with common chronic illnesses such as hypertension, diabetes, progressive heart failure, and renal disease, along with more acute disorders like hemorrhage and pneumothorax to train students to provide safe and efficient care. Integrating acute crises relevant to the simulated patient’s underlying disorder will make the experience more realistic. We will also implement socio-economic complications into treatment to make the experience consistent with the real world. As an example, irregular prescription use due to apathy, forgetfulness, or economic insufficiency can complicate long-term patient management. Recognizing noncompliance is critical for health care professionals. Further, the simulation will include patient migration, requiring students to practice transfers and hand-offs. These transfers and hand-offs could potentially be between multiple academic institutions. A visual representation of the student interaction with Osler One is shown in Figure 1.
Osler One will assess medical knowledge, professionalism, practice-based learning and improvement, ethics, and systems-based practice. Assessment will be accomplished by monitoring patient health outcomes, responding to patient needs, utilizing resources efficiently, and identifying system dysfunctions, which impede the care of patients.
Osler One will require four parts for implementation: 1) a physiological engine that provides realistic simulation of chronic conditions over time, 2) a controller program to develop “patients” for the end users, 3) an EHR front end, and 4) an assessment suite. Together, these parts ensure a self-contained, robust environment for vigorously challenging medical students.
Osler One will use the pathophysiology engine HumMod (http://hummod.org), a time-dependent physiologic simulation engine developed to simulate normal and pathophysiologic processes across different organ systems. HumMod resulted from over 40 years of development in the Department of Physiology at the University of Mississippi Medical Center.1-2 Drs. Guyton, Coleman, and Granger published the initial mathematical model of integrated human physiology in 1972. Dr. Coleman continued this development, later with Drs. Hester, Summers, and Pruett.
HumMod contains a series of mathematical equations and >7,500 variables and parameters that describe cardiovascular, renal, respiratory, endocrine, metabolism, and neural physiology; it can simulate chronic diseases, such as hypertension, congestive heart failure, and numerous endocrine disorders (e.g., diabetes). The HumMod software is efficient; it can simulate a month in less than five minutes. Figure 2 provides the results of a one-month HumMod simulation of a progressive heart failure. Osler One will utilize the currently available server-based HumMod.
The controller will generate patients and interface between the HumMod engine and the user interface, typically an electronic health record (EHR). It will provide random patients, including historical and physical data for the relevant pathological condition, for each student; this will better mimic reality. Real-life patient responses to pathology vary. For example, drug treatments can have different responses across a population. Equation-based physiological simulations are traditionally deterministic, with simulation outcomes always being the same unless initial conditions change. To overcome this issue, we developed a novel technique to generate robust populations of individuals with specific disorders, whereby the individuals may have different underlying physiology for the same pathology.
To ensure each patient experience will be different for each trainee, our process will furnish unique individuals to each student without relying on a template for each disease state. Matching patients to an integrative physiological model rather than a script provides trainees with an open-ended tool designed for inquisitive exploration rather than directed linearity. As such, it represents a disruptive breakthrough in the medical education paradigm.
Trainees will interface with patients in two ways: 1) through an EHR and 2) through real-time communication paradigms (e.g., text, email, and calendar) to alert the student to acute changes in patient status. We will use existing EHRs to provide the most realistic simulation of patient care possible. When students enter a request for patient values, such as a CBC, the controller will request the data from HumMod and the controller will then populate the EHR. Conversely, when the student prescribes a treatment the controller will provide that treatment information to HumMod, which will then simulate the results of the treatment. The HumMod engine can provide continual data; however, the controller will only provide data to the EHR when requested by the student, with delays mimicking a hospital’s data collection pace.
An expert system informed by evidentiary standards will perform the assessment. The tool will be a Bayesian probability family, whose prior probability is obtained by inverting the probabilities of specific symptoms appearing with specific diseases. Physicians managing patients within the system will adjust the probabilities. The probability approach allows a series of acceptable actions to be determined at each point in time, along with a score for each action based on the amount of uncertainty that action eliminates. The result is a repeatable measure of a student’s treatment of a patient, and a specific explanation of each action that lowered their score. When debriefing the student, this feature will stimulate recall and develop specific teaching points. Furthermore, the existence of this tool reduces the requirement for inefficient one-on-one time with physicians, reducing the cost of training while increasing effectiveness.
Osler One operates in real-time to provide a more genuine training experience. This means that when a patient takes a diuretic, the result isn’t seen for a week or more. Real-time simulation provides a more realistic training experience for the student. Interventions take time in real life; they should take time in training too. Similarly, blood work, labs, and every other aspect of clinical care will take place with a delay.
Patients will be ‘seen and evaluated’ by appointment. Since Osler One is real-time, trainees may need appointment reminders to prepare for the patient encounter. Low wait-times make for happier patients, which improves outcomes. In addition, students will need to respond to the urgent or emergent problems for a select number of their patients. Response time and intervention will be tracked via the program. Finally, a select number of patients for each student will have significant physiologic changes requiring intervention. These changes will result from chronic disease processes, superimposed major physiologic stressors, and system dysfunction. Instructors will be able to introduce specific perturbations to reinforce lessons. Students will be able to assess their patients and implement care plans.
Physician communication ranks as one of the highest, if not the highest, source of preventable medical mistakes. The largest source of communication errors is in transfer of care. Osler One will allow multiple modalities of transfer of care, including hand-offs, transfers and walk-in (people on vacation that become ill). In these cases, the medical record plus direct communication must suffice to pass all relevant information to the new caregiver. This type of transfer is a skill like any other, and must be taught and trained.
The simplest case is that of hand-offs. For example, an "on-call" physician signs-out new patients to other caregivers, essentially functioning as an information conduit. Similarly, if a patient “moves” or changes physician, a summary of care must be made to smooth the transition. For each case, appointments between caregivers can be made within the system and a phone call used for the communication. This may or may not be viewed by the local simulation center as part of the training process.
The other modality to consider is that of a traveling patient who becomes ill or is injured. Since the nearest physician won’t be their primary physician, the patient would present with little information. The treating physician can contact the primary physician to gather pertinent information; however, the response depends on the primary physician, although an appointment or phone conversation may not be feasible. In this case, the record and a brief note may have to suffice.
Osler One’s design will allow individuals to practice team management by working with the same patient, allowing teamwork between physician-trainees and nurse-trainees, for example, to take place seamlessly. Furthermore, hand-offs and presentations could be practiced.
Long-term care means some patients die. Whether from illness, trauma, or error, HumMod patients will die. Physicians lose patients when the patient dies. Deaths may trigger accountability measures too. Conversely, patients may have children. These children will be cared for by the parents’ physician. This may change if we implement specialties. For now, all Osler One physicians will be primary care physicians.
For undergraduate medical education, we propose that Osler One be integrated as a required longitudinal project throughout the second to fourth year. During the second year, students will meet their Osler One patients. During the first half of the third year and coinciding with the traditional transition to a clinical curriculum, students will review and utilize appropriate practice guidelines for patient care. During the second half of the third year, students will manage their patients’ complications, which coincides with students receiving increased responsibility for patient care in the clinical setting. During the fourth year, students assume full ownership for patient care and must utilize best practices and self-assess patient care. At project completion, each student will receive a summative performance assessment covering application of medical knowledge, professionalism, practice-based learning and improvement, and systems-based practice.
In summary, Osler One will provide a unique training opportunity for medical students, nursing students, residents, and other healthcare professionals. It will offer directed learning, while avoiding the linearity of directed acute simulations. It will assess medical knowledge, professionalism, practice-based learning and improvement, ethics, and systems-based practice by monitoring patient health outcomes, student responsiveness to patients’ needs, resource utilization, and system dysfunctions that impede patient care.
About the Authors
Dr. Robert Hester is a Professor of Physiology at the University of Mississippi Medical Center (UMMC) in Jackson, MS. He has 30 years of expertise in medical education. He has directed the development of HumMod for the last 10 years. He is also owner of HC Simulation, LLC which has licensed HumMod from UMMC.
Dr. William Pruett is an Instructor in Physiology at UMMC. He is a mathematician that specializes in producing and analyzing population models using machine learning methods. He developed multiple submodels in the HumMod system and has played an active role in the development of Osler One.
Mr. Leland Husband is a fourth year medical student at UMMC with over a decade’s worth of experience in project management and software support and development. For the past five years, he has served as a project manager on HumMod-related activities. He has been involved in the development of HumMod and the primary development and design of Osler One.
All authors blog about integrating medical simulation into medical training at all levels at http://failureisinstructive.com
- Guyton AC, Coleman TG, Granger HJ (1972). Circulation: overall regulation. Annu Rev Physiol 34, 13-46.
- Hester R, Brown A, Husband L, Iliescu R, Pruett WA, Summers RL and Coleman T (2011). HumMod: A modeling environment for the simulation of integrative human physiology. Physio. 2:12. doi: 10.3389/fphys.2011.00012.