Chronic or persistent pain presents as a complex, common, and costly health problem across the life span. Most approaches to the management of chronic pain incorporate education about pain neuroscience and the evidence for different clinical interventions such as surgery, pain relief medications, and nonpharmacological approaches. Education about pain is a clinical tool that empowers people to self-manage their condition through a better understanding of pain and the treatments strategies that are likely going to both safe and cost-effective.
Chatbots, or more formally conversation agents, are increasingly being used in the health field. Chatbots are computer systems that enable interactive 2-way communication through a variety of modalities including videos, images, voice or text. Due to the versatility of these systems, chatbot have the potential for improving clinical-client communication and delivering education in more engaging ways that are aligned with the client’s age, learning styles and learning preferences.
How we developed Dolores
Dolores was designed and developed through a collaboration between the University of Queensland, CSIRO, the Tess Cramond Pain and Research Centre, and the Queensland Interdisciplinary Paediatric Pain Service. Dolores was designed to be embedded in the Pain ROADMAP monitoring app. The idea for Dolores resulted from a working group with researchers and clinicians who discussed how to adapt Pain ROADMAP to a paediatric setting.
The chatbot framework used to build Dolores is an in-house framework developed at CSIRO. Two interactive experiences were built: a structured pain history interview and an education session. Key assessment constructs for the pain history interview were drawn from existing assessment proformas used by local pain centres and the electronic Persistent Pain Outcomes Collaboration (ePPOC) referral questionnaires. For the educational experience, an interdisciplinary team of researchers mapped chatbot pain education topics, commonly asked questions on these topics and conversation flows. Where possible, existing resources for information were referred to in Dolores’ responses to questions, including web pages, YouTube videos, mobile phone apps, and web-based education modules.
Our design allowed for the provision of age-appropriate responses in terms of both content and readability level using 3 app settings (i.e., for younger adolescents, older adolescents, and adults). Dolores generated responses using a combination of speech and written outputs with a “chat bubble” produced as typically seen in a conventional web-based chatting program. The user could interact with Dolores through text, speech, sketching (using a drawing widget) and optional interactive buttons. For the educational session, users were able to browse topics, select a random topic, take a quiz, or ask Dolores questions to cater for different learning styles. If a user desired to continue learning about a topic after the chatbot interaction, the information provided by the Dolores could be saved and then exported to a PDF format for future reference. Key features of the Dolores app are displayed visually in Figure 1 below.
Pilot Testing Dolores with people with chronic pain
For pilot testing, we recruited 60 people with chronic pain from waiting rooms of tertiary multidisciplinary pain centres in Brisbane, Australia. The 60 individuals consisted of 20 adolescents (aged 10-18 years), 20 young adults, and 20 adults (aged >35 years). Participants spent 20 to 30 minutes completing the interactive chatbot activities that enabled the Dolores app to gather a pain history and provide education about pain and pain treatments. After the chatbot activities, participants completed a custom-made feedback questionnaire measuring acceptability constructs pertaining to health education chatbots.
Overall, acceptability was high, across all age groups, for the following constructs: engagement, perceived value, usability, accuracy, responsiveness, adoption intention, aesthetics, and overall quality. The positive feedback provided to the open-ended questions also outweighed negative feedback. The participants provided positive feedback on the multiple options available for interaction and commented that the app was easy to use. The majority of negative comments referred the unpleasantness of Dolores’ speech, which was set to Siri Female (Australia).
Future Research
Our research team are currently investigating ways chatbots can provide an effective adjunct to traditional clinical care for people with chronic pain. Current projects explore how chatbots can enhance clinician-client communication and support people with pain once discharged from a tertiary multidisciplinary pain centre.
Links to Relevant Research Publications
Andrews, N.E., Ireland, D., Vijayakumar, P., Burvill, L., Hay, E., Westerman, D., Rose, T., Schlumpf, M., Strong, J., Claus, A. (2023).Acceptability of a Pain History Assessment and Education Chatbot (Dolores) Across Age Groups in Populations With Chronic Pain: Development and Pilot Testing. JMIR Formative Research, 7 ARTN e47267, 1-17. doi: 10.2196/47267
Ireland, D., Vijayakumar, P., Andrews, N. (2024). Dolores: a mobile chatbot for people living with chronic pain. 19th World Congress on Medical and Health Informatics, Sydney, NSW, Australia, 8 - 12 July 2023. Amsterdam, Netherlands: IOS Press. doi: 10.3233/shti231028
Links to other Media Content
News articles
Ireland D, Andrews, NE. Pictures of pain: how a visual chatbot can help people with chronic pain. Published: August 8, 2023 in The Conversation. Available from: https://theconversation.com/pictures-of-pain-how-a-visual-chatbot-can-help-people-with-chronic-pain-210810
Podcasts
Everyday AI: A CSIRO Podcast. Episode #3, Chatbots & Health Apps – AI in Healthcare. Available from: https://www.csiro.au/en/news/podcasts/everyday-ai-podcast
The Philosophy and Therapy Podcast. Episode #7. David Ireland. Available from: https://www.philosophyandtherapy.com/1570933/7819792