Chatbot-Augmented Flipped and Case-Based Learning to Enhance Linguistic and Clinical Communication in EMI Contexts

In hospitals and clinics around the world, the language of medicine is increasingly English. This shift, known as English as a Medium of Instruction, means that future doctors and nurses must not only master complex medical facts but also communicate them clearly, kindly, and accurately in English to patients who may not speak their native tongue. For students learning in countries where English is not the primary language, this creates a double challenge: they must learn the science of healing while simultaneously learning the specific words and social skills needed to explain a diagnosis or listen to a patient's fears. Traditional classrooms often struggle to bridge this gap, separating the study of grammar from the messy, real-world practice of talking to a sick person. When students lack confidence or vocabulary, the result can be hesitation, misunderstanding, and a breakdown in the trust essential for good care.

To address this, researchers at the State University of Jakarta explored a new way to teach these vital skills. They combined three distinct ideas into a single learning path. First, they used a "flipped" approach, where students learn the basics on their own before coming to class, saving face-to-face time for practice. Second, they employed "case-based" learning, which uses real-life medical scenarios to force students to apply what they know. Finally, they introduced a chatbot, a type of artificial intelligence program, to act as a patient and a tutor. This digital partner allowed students to practice speaking without the fear of being judged by a human, offering a safe space to stumble, correct themselves, and build confidence before facing real people.

The study focused on eighteen first-year medical students who were just beginning their journey in English. These learners were identified as needing significant support to communicate effectively in clinical settings. Over the course of seven weeks, the researchers guided these students through a structured program. The process began with students studying foundational medical vocabulary and communication patterns on their own. They then moved to a digital practice phase, where they interacted with a custom-designed chatbot named "Dr. OET Speaking." This tool served two purposes. In one mode, it acted as a strict tutor, guiding students step-by-step through specific medical scenarios, correcting their pronunciation, and ensuring they used the right sentence structures. In the other mode, it acted as a conversational partner, simulating a real patient so the student could practice a free-flowing conversation, asking questions and giving advice as a doctor would.

After these digital rehearsals, the students came together in the classroom. Here, they performed role-plays with classmates, recorded their interactions, and received feedback from their instructors. The goal was to see if this mix of self-study, AI practice, and human interaction could improve both their language skills and their ability to connect with patients. To measure the results, the researchers tested the students at the very beginning and again at the end of the seven weeks. They used a standard scoring system designed for medical English exams, looking at two main areas: how well the students used English grammar and vocabulary, and how effectively they communicated medical information with empathy and clarity.

The results showed a clear and significant improvement. When the students took the final test, their scores had risen substantially compared to their initial performance. The average score for their language skills jumped from roughly 12.6 to 18.7, while their ability to communicate clinically improved from about 12.8 to 19.0. These gains were not random; statistical analysis confirmed that the improvement was a direct result of the training program. The data suggests that the combination of structured preparation and repeated, low-pressure practice with the chatbot helped students organize their thoughts, choose the right words, and speak with greater confidence.

Beyond the numbers, the students shared their personal experiences through surveys, revealing why the method worked so well for them. Many reported that the chatbot's step-by-step guidance helped them organize their ideas and reduced their anxiety about making mistakes. They appreciated having a safe environment where they could practice explaining medical conditions over and over again until they felt ready. The immediate feedback from the AI allowed them to correct their pronunciation and sentence structure in real time, something that is difficult to achieve in a large classroom. Students noted that the chatbot helped them build the habit of speaking clearly and using medical terms correctly, which in turn made them feel more prepared for the actual role-plays with their peers.

However, the study also highlighted that the technology is not perfect. While the chatbot was a powerful tool, it had limitations that the students noticed. Some found that the conversations felt a bit stiff or unnatural compared to talking with a real person. The chatbot sometimes struggled to understand their voice clearly, especially if the internet connection was poor or if they spoke too quickly. There were moments when the AI gave responses that were too formal or did not quite fit the specific medical situation the student was trying to describe. These technical hiccups and the lack of genuine human emotion meant that the chatbot could not fully replace a real doctor-patient interaction. The students recognized that while the AI was excellent for practice, it could not replicate the complex emotional nuances of a real clinical encounter.

The researchers concluded that this blended approach offers a promising path forward for medical education. By using a chatbot to handle the repetitive, foundational work of language practice, teachers can free up valuable classroom time for more complex, human-centered activities like role-playing and critical thinking. The study suggests that when AI is used as a supportive partner rather than a replacement for human teachers, it can significantly boost a student's ability to speak and listen in a medical context. The success of the program lies in its structure: students first build a foundation, then practice in a safe digital space, and finally apply their skills in a real-world setting.

Despite the positive outcomes, the authors caution that this is not a complete solution. The study involved a small group of students over a relatively short period, so the long-term effects remain to be seen. Furthermore, the technical limitations of the chatbot, such as voice recognition errors and the occasional unnatural response, indicate that the technology still needs refinement. The researchers emphasize that the most effective learning happens when technology and human instruction work together. The chatbot provides the repetition and safety needed to build confidence, while human teachers provide the empathy, cultural understanding, and nuanced feedback that a machine cannot.

Ultimately, this research demonstrates that integrating artificial intelligence into medical language training can help future doctors communicate more effectively. It shows that with the right mix of preparation, practice, and feedback, students can overcome the barriers of learning a new language in a high-stakes field. The study does not claim to have solved the problem of medical communication, but it offers a clear, evidence-based method for making it easier. As medical education continues to evolve, tools like these may become essential for ensuring that every patient, regardless of where they are treated, receives care from a doctor who can listen, understand, and explain with clarity and compassion.