Interaction with an AI Chatbot in Audiology Education as a Critical Learning Agent

INTRODUCTION The challenge of providing clinical audiology students with consistent and meaningful patient interaction is well-documented in educational literature.1–3 Opportunities for patient engagement are needed for development of communication and clinical reasoning skills, essential for competent practice.3–7 Simulated patients, actors, and other students are popular teaching tools in health education, providing hands-on practice in a controlled environment.2,4 Despite their effectiveness, these methods often come with limitations, including high costs and logistical challenges.4,6 AI offers a promising solution to these issues by providing scalable and realistic clinical simulations.8–10 This study explores the use of an AI-driven virtual patient in audiology education, focusing on its role in enhancing communication and clinical reasoning skills through self-assessment, peer feedback, and reflective practice.AI Artificial Intelligence Education Technology Concept Stock Photo 2475630335 | Shutterstock.Figure 1: Example of a training conversation with the AI Chatbot.Table 1: Personal Characteristics and Clinical Features for the AI Chatbot.Table 2: Implementation of the ASPIRS Framework in the chatbot training.BACKGROUND AND INNOVATION AI applications have demonstrated potential in various health care educational settings, primarily in fostering diagnostic skills and communication abilities.8–11 The audiology discipline has progressively embraced AI technologies to supplement traditional consultation models and learning methods.12,13 Initiatives such as the development of AI-driven virtual patients and conversational agents have provided health care students with novel opportunities to practice clinical scenarios and enhance their diagnostic and patient management skills.14–16 This shift towards AI integration reflects a broader trend in health care education towards more interactive and personalised learning experiences.15,17 This project aimed to investigate methods to support Audiology students in improving their professional skills when interacting with clients. The research objectives were: To develop a realistic AI virtual patient chatbot with the integration of an audiology-specific framework for Master of Clinical Audiology students to practice their communication skills. To investigate how audiology students learn through their own reflections and peer interactions using the AI chatbot. Peer review is another crucial component in evaluating professional competence in health care education.18 Studies with medical students have shown the use of peer assessment to measure professional competence could reliably evaluate skills such as preparedness, respect, and trustworthiness, and provided a comprehensive view of students’ development.19,20 Integrating peer assessment with AI tools could enhance both formative and summative evaluations in health care education. Methodology This pilot study employed a design-based approach to create an AI-driven virtual patient using the Character.AI platform.21 This platform was selected for several reasons, including its natural language processing (NLP) model, behaviour-shaping algorithm, conversational memory of characters and strict protocols regarding inappropriate prompts. The research team defined the patient’s personality and clinical characteristics to simulate a middle-aged female patient programmed with a specific set of clinical features to create a realistic persona (See Table 1). To design the conversational flow, a modified version of the Audiology Simulated Patient Interview Rating Scale (ASPIRS) informed the evaluation of the chatbot’s quality and consistency of responses.1 This tool was specifically developed for assessing audiology students in case history taking and providing patient feedback.1 The chatbot was trained using the ASPIRS framework, excluding the nonverbal communication section. Additionally, the Calgary-Cambridge four habits model which serves as a practical method of teaching both the process of communication, as well as the effective gaining of content information was used a benchmark. (See Table 2 and Figure 1).[22] Over a five-month period, the development process involved incorporating typical patient case history-taking scripts, to guide the AI’s dialogue, ensuring that the interactions followed a structured progression. Use of common case history-taking protocols ,i.e. presenting complaint, past medical history etc, allowed the researchers to engage with the chatbot in a way that mirrored real-life clinical scenarios, helping to build conversation skills in a safe, repeatable environment.8 The use of cognitive strategies, such as organising clinical information, was embedded into the chatbot’s structure to promote reflective thinking and enhance diagnostic skills. To ascertain PAT’s readiness for student trials, criteria included human-like utterances with prompt response times, adeptness in detecting and responding to intent and contextual cues, recall of previous conversation information, and consistent use of Australian English grammar and vocabulary for optimal comprehension by native speakers was used. The study included seven participants (five female, two male), all first-year Master of Clinical Audiology students. Four participants were native English speakers, while three were second-language English speakers. Participants engaged with the chatbot throughout a single semester, followed by self-reflection and peer feedback. The participants rated themselves and their peers using a seven-point scale and provided either a “compliment or suggestion” to review each other in the following areas of the ASPIRS framework: professionalism, communication, interview skills, and content. An LMS-integrated educational tool known as “Feedback Fruits” was used to gather self-reflections and peer review. The tool allowed participants to share downloaded interactions from websites and upload anonymous text-based files for evaluations. The data collected from these interactions were analysed qualitatively against the ASPIRS framework to assess communication patterns, clinical reasoning skills, and the effectiveness of the learning process. RESULTS The AI chatbot successfully facilitated the development of key communication and clinical reasoning skills among participants, serving as an agent for reflection and learner-led inquiry. Peer evaluations exhibited generally elevated scores for professionalism and communication with the AI patient (ranging between 6-7), yet lower ratings were observed for interview skills and the substantive content of interactions (ranging from 4-6). In comparison to peer assessments, self-assessments generally yielded lower scores indicating students were more critical with their own performance. It is important to note that there were no pre-post score comparisons, as students only provided scores after interacting with the chatbot, which limits the ability to measure any changes in skills over time. The students reported that the virtual patient provided a realistic simulation of patient interactions and an authentic practice environment for communication and diagnostic inquiry without the pressure of live clinical scenarios. Importantly, the use of the ASPIRS framework guided students to identify specific areas for improvement, such as interview structure, and patient rapport-building. For instance, a participant commented to self: “My interview was conducted professionally with good follow-up questions; however, some questions could have been worded better and explored further.” On the other hand, one participant provided feedback to another on the specific professional areas to focus. “Your interview was conducted in a professional manner with excellent follow-up questions to get accurate information. One question you could have followed up more on was asking how long the dizzy spells last for, as this may help with obtaining a diagnosis.” While the sample size was small (n=7), the results indicate that the chatbot served as an effective tool for fostering learner-led inquiry and reflective practice. Students demonstrated increased awareness of their communication gaps, which they subsequently addressed through continued practice with the chatbot. DISCUSSION The findings of this study highlight the potential of AI-driven patients to bridge the gap between limited real patient interactions and the need for students to practice essential clinical skills. The virtual patient not only provided an avenue for repeated practice but also fostered self-reflection and peer feedback. The integration of structured frameworks such as the ASPIRS model allowed students to align their practice with professional standards, enhancing their competency development. One of the key insights from this study was the value of unlimited practice opportunities. Traditional clinical training often limits students’ exposure to real patients. In contrast, the AI chatbot offered a scalable solution that provided students with continuous access to patient simulations. This ability to engage with the AI at their own pace was seen as a major benefit by the participants. Another critical finding was the chatbot’s role in promoting learner autonomy. Students appreciated the opportunity to engage in self-directed learning and recognised that the feedback they received from peers helped them refine their communication skills. The role of metacognition in guiding this process was also noted, as students were able to assess their performance and take actionable steps to improve their skills. The integration of clinical reasoning strategies and metacognitive techniques into the AI design was a core component of this study. By using common characteristics of case-history conversations, the chatbot’s design encouraged students to organise clinical information systematically, mirroring the cognitive processes required in real patient interactions. By reflecting on their responses and evaluating their performance, students were able to engage in higher-order thinking, which enhanced their diagnostic skills. This approach also aligned with research on cognitive scaffolding, where learners are provided with structured support to help them build expertise over time. These results demonstrate that the AI chatbot successfully facilitated the development of communication and clinical reasoning skills, though certain areas, such as interview skills, require further attention. The lower scores in areas such as interviewing and substantive content could be attributed to the inherent limitations of AI in replicating the complexity and nuances of human interactions. Effective interviewing and substantive content in clinical conversations requires dynamic responses based on context and nonverbal cues. These elements that are difficult for AI to assess and replicate. As a result, students may have received less peer feedback in these areas, impacting their performance scores in both interviewing techniques and the depth of content they engaged with during the interactions. Future improvements in AI design, particularly in the integration of more nuanced, adaptive responses, may address these challenges and further enhance the realism of the learning experience. Additionally, the sample size was small, limiting the ability to generalize of the findings. Future studies could include larger cohorts to better assess the impact of AI chatbot training across different learner demographics. This study did not examine long-term outcomes, such as how AI chatbot practice influences real-world clinical performance. Longitudinal research could provide valuable insights into the lasting effects of AI-assisted learning. As with any AI research, potential biases in AI design are also acknowledged. CONCLUSION The integration of AI in audiology education holds significant promise for improving communication and clinical reasoning skills. Future research could explore the use of multimodal AI systems that incorporate voice and video alongside text-based interactions. This would more closely mimic real patient interactions, including nonverbal cues, which are critical in health care communication. Furthermore, adaptive AI models that personalise responses based on student performance could offer more targeted learning experiences. As AI technology continues to evolve, its potential to revolutionise clinical education across disciplines grows, offering more personalised, scalable, and effective training opportunities for future health care professionals.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.