Virtual Patients for remote learning in General Practice

Problem

The COVID-19 aftermath triggered a shift towards hybrid learning in medical education, with a significant portion now in e-learning environments. Increased medical school admissions intensified competition for patient interactions, making patient consultation skills vital for student development. This chatbot addresses evolving educational needs.

Approach

To bridge this gap, we created a Virtual Patient (VP) web app prototype (accessible at https://patientinterview.pythonanywhere.com/). Students interact via text inputs, and responses are displayed conversationally. The system underwent training by the project lead and students, with additional input from a generic AI model for unforeseen queries.

Findings

The VP's standout feature is its "training" mechanism, enabling the app to learn and provide accurate responses based on user input. Users can train the system with new responses when needed. Users engaged with up to 20 consecutive questions, all met with pertinent replies. This engagement is attributed to the app's immersive narrative construction. Conversational data has been extracted which allows route analysis providing in depth information on the structure and flow of students' interactions with the virtual patient, this data will be analysed and presented More case scenarios are in development. This educational tool is built entirely on free software, enhancing accessibility and sustainability.

Consequences

Developing and implementing a Virtual Patient (VP) with free software demonstrates the potential for remote learning This innovation introduces an interactive dimension, bridging didactic methods and real-world clinical experiences. It also enables the integration of existing teaching resources like OSCE simulated patient scripts. Future prospects include VP validation for assessments, offering versatile tools for evaluating students' clinical competencies.