Prime Global logo
Who is Prime Prime capabilities See our ISMPP US 2024 posters Case studies of augmented reality posters Case study of interactive poster
Home

Effective education and integration of generative artificial intelligence into medical communications: use of a fictional-disease exercise to increase confidence

Amanda Webb, Valerie Moss, and Safeer Mughal
Prime, London, UK

Table of contents

Conclusions

  • This exercise improved team sentiment and the understanding of AI within a medical communications context, highlighting its potential as a first step in AI integration.
  • Sustained efforts, including formal training and dedicated support, are crucial to building confidence and ensuring effective and compliant utilization of AI within the team.
Top

Introduction

  • Generative artificial intelligence (genAI) has the potential to improve the efficiency and quality of healthcare information and is being rapidly integrated into medical communications.
    • As a result, adequate education and training are necessary.
  • Here, we evaluated the impact of a group training exercise using genAI tools to determine team sentiment and preparedness for AI integration in medical communications.
Top

Methods

  • In a team meeting setting, 37 participants were divided into small groups of ~4 people to complete an AI-based group exercise.
    • Groups had 20 minutes to create a fictional disease awareness storyboard using AI tools, including ChatGPT 3.5, Bing Copilot text and image generation, Perchance AI photogenerator, and MURF.AI.
  • A follow-up survey was distributed to all participants to evaluate any changes in confidence and comfort when using AI tools, as well as any potential concerns or perceived needs for AI implementation. A full description of the AI task and the survey questions can be accessed here.
Top

Results

  • Of 37 participants, 14 completed the survey (38% response rate).
  • The mean (standard deviation) rating of the overall experience with the AI exercise, where 1=poor and 10=excellent, was 7.4 (1.4; Figure 1).
  • No participants rated their experience less than 5 out of 10.
  • Overall, 10 (71%) respondents said that the AI exercise increased their understanding of the capabilities and potential applications of AI in medical communications (Figure 2).
Figure 1.
Overall experience with the AI exercise
Diagram showing the Overall experience with the AI exercise

Respondents (n=14) were asked how they would rate their overall experience with the AI-driven exercise (1=poor; 10=excellent)

Figure 2.
Perception of whether the AI exercise increased understanding of the capabilities and potential applications of AI
Diagram showing the Perception of whether the AI exercise increased understanding of the capabilities and potential applications of AI
  • Prior to the AI exercise, 10 (71%) respondents had a neutral feeling towards use of AI in medical communications (Figure 3A).
  • After participation, 8 (57%) reported feeling somewhat more positive towards AI , 5 reported feeling neutral, and 1 reported feeling somewhat negative (Figure 3B).
Figure 3A.
Overall feelings about the use of genAI before participation in the AI exercise
Diagram showing the Overall feelings about the use of genAI before participation in the AI exercise
Figure 3B.
Overall feelings about the use of genAI after participation in the AI exercise
Diagram showing the Overall feelings about the use of genAI after participation in the AI exercise
  • Most respondents felt that AI was helpful to speed up the process and build on their own ideas (Figure 4).
  • The most reported reason for AI being unhelpful when completing the task in a short timeframe was that inaccurate or irrelevant information was provided (Figure 5).
Figure 4.
Perceptions on how AI can help when completing a task within a short timeframe
Diagram showing the Perceptions on how AI can help when completing a task within a short timeframe

Respondents (n=14) were asked how they found AI helpful when completing the task within the short timeframe. They could check all that applied.

Figure 5.
Perceptions on how AI was considered unhelpful when completing a task within a short timeframe
Diagram showing the Perceptions on how AI was considered unhelpful when completing a task within a short timeframe

Respondents (n=14) were asked how they found AI unhelpful when completing the task within the short timeframe. They could check all that applied. A response was not provided by 4 (29%) respondents.

  • When asked about their biggest concerns with AI use in medical communications, 11 (79%) respondents were concerned about accuracy and reliability, and 9 (64%) were concerned about compliance and client perception when considering the use of AI.
    • Four were concerned about the potential for job displacement and security.
  • Respondents had the opportunity to provide free-text responses indicating what would help them to integrate AI into their everyday practices.
    • Responses included AI platform guidance, clear ethical and compliance guidelines, and additional training opportunities.
Acknowledgements

We would like to thank Hannah Boyd for medical writing assistance and the Prime Production and Editor teams for their support with developing this poster.

Disclosures

Amanda Webb, Valerie Moss, and Safeer Mughal are employees of Prime, London, UK.

Supplementary Materials