Designer - Chatbot for Children with Autism

January 2024 - May 2024

Overview

This project entailed the capstone assignment for ITEC534 - Advanced Human-Computer Interaction taught by Dr. Dezhi Wu. This project was completed alongside two other students, Wyatt Belamy and Lucas Bowers under Dr. Wu’s supervision. Dr. Wu invited our class of 3 to create a theraputic chatbot for children with autism due to her experience volunteering at the South Carolina Center for Assistive Technology and Educational Research (SC-CATER).

University of South Carolina College of Education Logo.

Project Aims

The primary objective of this project is to develop a chatbot tailored to the unique needs of children diagnosed with ASD. The chatbot will serve as a virtual companion that engages users in interactive social scenarios, providing guidance, feedback, and reinforcement in real-time. Through this chatbot, we aim to assist in their navigation through common social scenarios while also providing some entertainment to users.

Specifically, our project aims to achieve the following:

  1. Design and implement a user-friendly interface that is visually appealing and accessible to individuals with ASD.
  2. Develop a conversational AI system capable of understanding and responding to user input in a natural and engaging manner.
  3. Create a diverse range of social scenarios and activities within the chatbot platform, covering common social skills such as greetings, turn-taking, and understanding emotions.
  4. Incorporate personalized feedback mechanisms to provide tailored support and encouragement based on the user’s progress and needs.
  5. Evaluate the effectiveness of the chatbot intervention through user testing and feedback analysis, with a focus on improvements in social skills, confidence, and overall well-being among children with ASD.

Project Justification

It is well-documented that children on the autism spectrum experience life-impacting difficulties in social situations. Examples may include things like delayed speech, hyperactive behavior, and difficulties with learning . As a result of these issues, children on the autism spectrum can have trouble with behaving in ways that are considered acceptable in social situations.

Research has shown that these difficulties regarding socializing can be reduced in children with autism through exposure to both in-person and technological social skills training (SST). Yet, as discussed by Soares et. al, in-person therapy of this type can have prohibitive costs which technology can alleviate. Additionally, technology can provide benefits such as reducing anxiety caused by social situations, reducing instructor fatigue, and provide more settings in which participants can participate in SST.

It has been shown that adults on the autism spectrum already successfully utilize chatbots in assisting with social situations—such as in a 2023 article by Wired writers that explored ways in which adults with autism use Chat GPT to train themselves to navigate potential social situations.

When discussing the different approaches that autistic children most effectively learn from, visual learning plays a huge role. The Treatment and Education of Autistic and related Communication-handicapped Children (TEACHH) program conducted numerous studies showing that autistic children learn from visual and interactive mediums far more successfully compared to traditional methods.

Given these insights, there is a clear opportunity to leverage AI-driven chatbots to create a novel intervention tool specifically designed to assist children with ASD in developing their social skills. By harnessing the power of visual and interactive learning methods, combined with the accessibility and scalability of technology, we aim to address the unmet needs of this population and improve their quality of life.

Development Plan

Initial Development Phase:

  1. Conduct a thorough literature review to inform the design and development of the chatbot, focusing on existing research on ASD interventions, AI-driven chatbots, and best practices in human-computer interaction.
  2. Define the scope and requirements of the chatbot application, including target age range, user interface design, and key features.

Prototype Development Phase:

  1. Develop a prototype version of the chatbot application, integrating basic conversational capabilities and visual elements.
  2. Test the prototype with a small group of high functioning young adults diagnosed with ASD to gather initial feedback on usability, engagement, and perceived helpfulness.
  3. Iterate on the design and functionality of the chatbot based on user feedback and usability testing results.

Feature Enhancement Phase:

  1. Enhance the chatbot’s conversational AI capabilities to improve natural language understanding and generate more contextually relevant responses.
  2. Expand the range of social scenarios and activities available within the chatbot platform, incorporating feedback from users and domain experts.
  3. Implement personalized feedback mechanisms to adapt the chatbot’s responses based on individual user profiles and progress metrics.

Evaluation and Validation Phase:

  1. Conduct a comprehensive evaluation of the chatbot intervention through a randomized controlled trial (RCT) involving a larger sample of individuals with ASD.
  2. Measure outcomes such as improvements in social skills, self-confidence, and emotional regulation before and after using the chatbot.
  3. Administer validated assessment tools and surveys to gather quantitative and qualitative data on user experiences and satisfaction.
  4. Analyze the results of the evaluation to assess the effectiveness of the chatbot intervention and identify areas for further improvement.

Progress Report 1

This document entails the work undergone and further developments needed based on literature review. Additionally, it includes a UI mockup and a conversational flow map on three areas we decided to focus on: educational, emotional, and social skills training (SST).