top of page

Tools:

  • Articulate Storyline

  • Camtasia

  • Visual Studio Code

  • Google Cloud Function

  • ChatGPT Open AI API

  • Adobe Illustrator

  • Adobe Photoshop

Skills:

  • eLearning Development

  • Javascript & Node.js

  • API Key Security

  • Server-side Data Retrieval

  • Graphic Design

  • Video Editing

​Project Overview:​

In this scenario-based eLearning module that uses ChatGPT-powered feedback, clinical scientists engage with realistic scenarios that test their understanding and application of key regulatory standards. This module offers a hands-on approach to understanding compliance.

​

Challenge:

​

Biotechnology companies face the critical challenge of ensuring that their clinical research professionals fully comprehend and can effectively apply key regulatory standards on patient safety, data integrity, and reporting requirements to maintain compliance and uphold the integrity of clinical trials.

​

Solution:

​

To address this need, I designed a structured eLearning module segmented into chapters, each focusing on a distinct compliance guideline, supplemented with scenario-based questions evaluated by ChatGPT to provide customized feedback, and offering learners downloadable PDFs of each scenario and their interactions for further review.

​

Technical Highlights: 

​

  • ChatGPT Powered Feedback: Employs ChatGPT to provide instant, personalized feedback on learners' responses, enhancing understanding and retention of compliance guidelines.

​​

  • Mission and Values Video: Opens with a video that introduces the company's mission and values, grounding learners in the ethical context of their compliance training.

​​

  • Downloadable PDF of Feedback: Allows learners to download a PDF summarizing scenarios, their responses, and ChatGPT’s feedback for ongoing review and learning reinforcement.

​

Given that this is a concept-based project, I was not able to strictly follow the ADDIE model. I leveraged my creativity along with resources like Google and ChatGPT to simulate the ‘Analyze’ phase.

​

If this had been a real-world project, my first step would typically involve leading a kick-off meeting with stakeholders to align the training goals with organizational objectives. Key information I always explore during kick-off meetings and design sessions with SMEs include:

​

  • Identifying the learner profile and capabilities

​

  • Common mistakes and challenges related to the training topic

​

  • Determining the gap between current abilities and desired outcomes

​​

  • Identifying existing or potential data points that could inform the training design

​

In this case, it was devised so that the output of the analysis resulted in:

​

Audience: Employees at a biotechnology company, particularly those involved in clinical research and trials. 

​

Objective: By the end of this module, clinical research professionals (Audience) will be able to apply key principles of regulatory compliance on patient safety, data integrity, and reporting requirements (Behavior) to various clinical trial scenarios (Condition).

Being a proponent of evidence-based learning, I knew I wanted to implement scenario-based learning for this project. Pulling from Dr. Will Thalheimer's research, I implemented several components of his '7 Learning Maximisizers' model including 'Support for Decision-Making Competencies' by designing 3 scenarios that allowed learners to practice making real-world decisions.

​

Understanding the importance of feedback in helping learners build mental models determined the need to build the ChatGPT bot. I wanted the learners to receive feedback that was meaningful and applicable to their responses instead of generic, one-size-fits-all feedback. Utilizing this type of feedback is supported by the Constructivist theory. Custom feedback provides specific insights and corrections that help learners build (or construct) their understanding based on their unique responses and experiences.

​

While this project represents just a sample of a potentially more extensive eLearning module, I chose to provide an introduction and context from the start. In alignment with Affective Learning Theory, I incorporated a brief explainer video before the scenarios. This video not only offers essential context to the learners but also aims to captivate them with a 'moment of delight', particularly important for a topic that might otherwise be seen as dry and uninteresting.

​

Interested in checking out the storyboard for this project?

The development portion of this project included several key aspects:

​

​

Explainer Video:

I designed the explainer video in Camtasia using copyright-free videos from various sources online. Because the videos were sourced from different places and not from a set with the same aesthetics, I applied a blue color grading to all of the video assets. This created cohesion among the videos. The video features several text pop-ups highlighting key points of the narration, emphasizing the company's mission and values. Consistent branding, fade transitions, and an intro/outro wrapper are used throughout the video.

​

Select this link to view the storyboard for the explainer video.

​

​

ChatGPT Bot:

A significant part of the project involved coding and integrating a ChatGPT AI bot, a task that challenged me to create a dynamic, interactive element that could provide context-aware feedback in real-time. 

Learners interact with the chatbot, inputting their responses to scenarios. The chatbot evaluates their answers against regulatory guidelines, giving immediate, personalized feedback to reinforce learning.

​

Continuous improvement and growth as a learning development professional and staying on top of learning trends are of utmost importance to me, so I must give credit to Discover eLearning. They offer a course on how to build a ChatGPT API application using the OpenAI platform. Their course teaches how to develop a chatbot that can provide tailored feedback to a learner. I utilized the knowledge I learned in their course to help me develop the chatbot in this eLearning module.

 

However, their course teaches you only how to develop one on your local server. I took the knowledge I gained from my project, 'Navigate Compliance' and set up an environment on Google Cloud, deployed my function while securely storing my OpenAI API key allowing users to experience the module from anywhere.

​

Examples of ChatGPT-powered feedback:

​

​

​

​

​

​

​

​

​

​

​

​

​

Branding:

Because this was a concept project for a fictitious company, I developed a branding guide to help define the visual aesthetic of the eLearning module. This included designing a logo, selecting a color palette, and determining font type, size, and weight for the title, subtitle, and body text. Following sound visual design principles, I limited font type to 1, utilizing font size to denote hierarchy. The logo is clean and maintains legibility across different sizes, ensuring effective presentation whether it's scaled down for small applications or enlarged for larger displays. Branding is kept consistent among all elements.

​

Want to see the branding guide I developed for this project?

​

​

UX & Accessibility: 

It was important to me that this module be as accessible as possible because accessibility is for everyone. Customized tab order was specified for each interactive element on the screen as well as alt-text for images. Closed captions were included in the video.

 

One issue I ran into while testing different screen readers with the content was how each accessed and read the information. I tested both NVDA and VoiceOver with mixed results. Both screen readers struggled to read the prompt text on each slide. To solve this problem, I labeled each slide's title with the prompt text. While lengthy, this was a suitable workaround. 

​

Another issue that still needs to be solved is that the learner cannot move forward and backward through the course freely. Ideally, following UX design and accessibility principles, I would have preferred for the module to include home and back buttons. However, upon testing this possibility, it quickly became clear that if the learner accessed these buttons after interacting with the chatbot, it would not reset the API request and allow the learner to make another 'call' to the ChatGPT API. I suspect with some tweaking of the Javascript and Node.js, it might be possible. Further testing is in the works to solve this problem. 

Screenshot 2024-01-14 at 9.42.59 AM.png
Screenshot 2024-01-14 at 9.42.20 AM.png
Screenshot 2024-01-14 at 9.44.15 AM.png

​In developing this eLearning module, I drew upon lessons learned from my previous work on the pharmaceutical transport project, particularly in the area of technical integration and security. My experience with integrating real-time data APIs into an eLearning platform was further honed. Securing API keys, using cloud functions and environment variables, was a critical step.Looking forward, for real-world application, especially in regulatory compliance, it would be important to utilize the latest version of ChatGPT, such as ChatGPT-4. Using the most up-to-date model would be essential for accuracy and reliability, especially with such a sensitive topic.

 

Another crucial implication is weighing the scalability and cost of implementing a ChatGPT-powered learning experience across a large organization. It would be imperative to utilize the most up-to-date version of ChatGPT, which currently is not free, and then every time an API request is made that also costs money. And then also making sure the course is accessible to all is imperative before rolling it out large scale. 

©2024 by REN Learning Design

bottom of page