Artificial intelligence in virtual standardized patients: Combining natural language understanding and rule based dialogue management to improve conversational fidelity.

INTRODUCTION: Advances in natural language understanding have facilitated the development of Virtual Standardized Patients (VSPs) that may soon rival human patients in conversational ability. We describe herein the development of an artificial intelligence (AI) system for VSPs enabling students to practice their history taking skills. METHODS: Our system consists of (1) Automated Speech Recognition (ASR), (2) hybrid AI for question identification, (3) classifier to choose between the two systems, and (4) automated speech generation. We analyzed the accuracy of the ASR, the two AI systems, the classifier, and student feedback with 620 first year medical students from 2018 to 2021. RESULTS: System accuracy improved from ∼75% in 2018 to ∼90% in 2021 as refinements in algorithms and additional training data were utilized. Student feedback was positive, and most students felt that practicing with the VSPs was a worthwhile experience. CONCLUSION: We have developed a novel hybrid dialogue system that enables artificially intelligent VSPs to correctly answer student questions at levels comparable with human SPs. This system allows trainees to practice and refine their history-taking skills before interacting with human patients.

Using virtual standardized patients to accurately assess information gathering skills in medical students.

Introduction: Practicing a medical history using standardized patients is an essential component of medical school curricula. Recent advances in technology now allow for newer approaches for practicing and assessing communication skills. We describe herein a virtual standardized patient (VSP) system that allows students to practice their history taking skills and receive immediate feedback. Methods: Our VSPs consist of artificially intelligent, emotionally responsive 3D characters which communicate with students using natural language. The system categorizes the input questions according to specific domains and summarizes the encounter. Automated assessment by the computer was compared to manual assessment by trained raters to assess accuracy of the grading system. Results: Twenty dialogs chosen randomly from 102 total encounters were analyzed by three human and one computer rater. Overall scores calculated by the computer were not different than those provided by the human raters, and overall accuracy of the computer system was 87%, compared with 90% for human raters. Inter-rater reliability was high across 19 of 21 categories. Conclusions: We have developed a virtual standardized patient system that can understand, respond, categorize, and assess student performance in gathering information during a typical medical history, thus enabling students to practice their history-taking skills and receive immediate feedback.

Evaluation of high-fidelity and virtual reality simulation platforms for assessing fourth-year medical students' encounters with patients in need of urgent or emergent care.

BACKGROUND: Medical students in the U.S. must demonstrate urgent and emergent care competence before graduation. Urgent and emergent care competence involves recognizing, evaluating and initiating management of an unstable patient. High-fidelity (HF) simulation can improve urgent and emergent care skills, but because it is resource intense, alternative methods are needed. STUDY OBJECTIVE: Our primary purpose was to use program evaluations to compare medical student experiences with HF and virtual reality (VR) simulations as assessment platforms for urgent and emergent care skills. METHODS: During their emergency medicine clerkship, students at The Ohio State University College of Medicine must demonstrate on HF manikins, competence in recognizing and initiating care of a patient requiring urgent or emergent care. Students evaluated these simulations on a five-point quality scale and answered open-ended questions about simulation strengths and weaknesses. Faculty provided feedback on student competence in delivering urgent or emergent care. In 2022, we introduced VR as an alternative assessment platform. We used Wilcoxon Signed Ranks and Boxplots to compare ratings of HF to VR and McNemar Test to compare competence ratings. Comments were analyzed with summative content analysis or thematic coding. RESULTS: We received at least one evaluation survey from 160 of 216 (74.1%) emergency medicine clerkship students. We were able to match 125 of 216 (57.9%) evaluation surveys for students who completed both. Average ratings of HF simulations were 4.6 of 5, while ratings of VR simulations were slightly lower at 4.4. Comments suggested that feedback from both simulation platforms was valued. Students described VR as novel, immersive, and good preparation for clinical practice. Constructive criticism identified the need for additional practice in the VR environment. Student performance between platforms was significantly different with 91.7% of students achieving competence in HF, but only 65.5% in VR (p≤.001, odds-ratio = 5.75). CONCLUSION: VR simulation functions similarly to HF for formative assessment of urgent and emergent care competence. However, using VR simulation for summative assessment of urgent and emergent care competence must be considered with caution because students require considerable practice and acclimation to the virtual environment.

Medical students found value in using virtual reality simulation as a platform for practice and feedback in a formative assessment arrangement.Students described the virtual reality simulation as immersive and good preparation for clinical practice.Technical difficulties were common and the student learning curve for acclimating and learning how to function in the virtual environment was noteworthy.

First Responder Virtual Reality Simulator to train and assess emergency personnel for mass casualty response.

As mass casualty incidents continue to escalate in the United States, we must improve frontline responder performance to increase the odds of victim survival. In this article, we describe the First Responder Virtual Reality Simulator, a high-fidelity, fully immersive, automated, programmable virtual reality (VR) simulation designed to train frontline responders to treat and triage victims of mass casualty incidents. First responder trainees don a wireless VR head-mounted display linked to a compatible desktop computer. Trainees see and hear autonomous, interactive victims who are programmed to simulate individuals with injuries consistent with an explosion in an underground space. Armed with a virtual medical kit, responders are tasked with triaging and treating the victims on the scene. The VR environment can be made more challenging by increasing the environmental chaos, adding patients, or increasing the acuity of patient injuries. The VR platform tracks and records their performance as they navigate the disaster scene. Output from the system provides feedback to participants on their performance. Eventually, we hope that the First Responder system will serve both as an effective replacement for expensive conventional training methods as well as a safe and efficient platform for research on current triage protocols.

Virtual Antenatal Encounter and Standardized Simulation Assessment (VANESSA): Pilot Study.

BACKGROUND: Prenatal counseling at the limits of newborn viability involves sensitive interactions between neonatal providers and families. Empathetic discussions are currently learned through practice in times of high stress. Decision aids may help improve provider communication but have not been universally adopted. Virtual standardized patients are increasingly recognized as a modality for education, but prenatal counseling simulations have not been described. To be valuable as a tool, a virtual patient would need to accurately portray emotions and elicit a realistic response from the provider. OBJECTIVE: To determine if neonatal providers can accurately identify a standardized virtual prenatal patient's emotional states and examine the frequency of empathic responses to statements made by the patient. METHODS: A panel of Neonatologists, Simulation Specialists, and Ethicists developed a dialogue and identified empathic responses. Virtual Antenatal Encounter and Standardized Simulation Assessment (VANESSA), a screen-based simulation of a woman at 23 weeks gestation, was capable of displaying anger, fear, sadness, and happiness through animations. Twenty-four neonatal providers, including a subgroup with an ethics interest, were asked to identify VANESSA's emotions 28 times, respond to statements, and answer open-ended questions. The emotions were displayed in different formats: without dialogue, with text dialogue, and with audio dialogue. Participants completed a post-encounter survey describing demographics and experience. Data were reported using descriptive statistics. Qualitative data from open ended questions (eg, "What would you do?") were examined using thematic analysis. RESULTS: Half of our participants had over 10 years of clinical experience. Most participants reported using medical research (18/23, 78%) and mortality calculators (17/23, 74%). Only the ethics-interested subgroup (10/23, 43%) listed counseling literature (7/10, 70%). Of 672 attempts, participants accurately identified VANESSA's emotions 77.8% (523/672) of the time, and most (14/23, 61%) reported that they were confident in identifying these emotions. The ethics interest group was more likely to choose empathic responses (P=.002). Participants rated VANESSA as easy to use (22/23, 96%) and reported that she had realistic dialogue (15/23, 65%). CONCLUSIONS: This pilot study shows that a prenatal counseling simulation is feasible and can yield useful data on prenatal counseling communication. Our participants showed a high rate of emotion recognition and empathy in their responses.

Developing a Conversational Virtual Standardized Patient to Enable Students to Practice History-Taking Skills.

INTRODUCTION: Although traditional virtual patient simulations are designed to teach and assess clinical reasoning skills, few employ conversational dialogue with the patients. The virtual standardized patients (VSPs) described herein represent standardized patients that students interview using natural language. Students take histories and develop differential diagnoses of the VSPs as much as they would with standardized or actual patients. The student-VSP interactions are recorded, creating a comprehensive record of questions and the order in which they were asked, which can be analyzed to assess information-gathering skills. Students document the encounter in an electronic medical record created for the VSPs. METHODS: The VSP was developed by integrating a dialogue management system (ChatScript) with emotionally responsive 3D characters created in a high-fidelity game engine (Unity). The system was tested with medical students at the Ohio State University College of Medicine. Students are able to take a history of a VSP, develop a differential diagnosis, and document the encounter in the electronic medical record. RESULTS: Accuracy of the VSP responses ranged from 79% to 86%, depending on the complexity of the case, type of history obtained, and skill of the student. Students were able to accurately develop an appropriate differential diagnosis on the basis of the information provided by the patient during the encounter. CONCLUSIONS: The VSP enables students to practice their history-taking skills before encounters with standardized or actual patients. Future developments will focus on creating an assessment module that will automatically analyze VSP sessions and provide immediate student feedback.