Crowdsourcing cancer survivors’ perceptions on the use of artificial intelligence in financial hardship interventions.

Abstract

610 Background: Financial hardship (FH) interventions are effective but are resource intensive and suffer from substantial implementation barriers. The use of artificial intelligence (AI) and automation as an implementation strategy for adaptation may improve the reach and scalability of FH interventions. However, cancer survivor perceptions of AI automation in FH interventions are understudied. Methods: Individuals were eligible for a short online crowdsourcing study if they were at least 18 years, a cancer survivor, and lived in the US. The survey first provided examples of applications of AI and automation (e.g., chatbots, risk algorithms, virtual humans) and then asked seven open-ended crowdsourcing questions about how AI or automation could have improved specific components of their cancer experience. We qualitatively analyzed two of the crowdsourcing questions focused on 1) financial assistance and 2) health insurance. First cycle coding was completed by ChatGPT to generate deductive categories of feedback. The analytic team then coded 10% of responses into the preliminary categories and made adjustments. The codebook was then applied to the remaining responses. Results: A total of N=200 cancer survivors participated and were on average 50.3 years (SD: 14.3) and 10.1 years (SD: 9.1) since diagnosis, most commonly non-Hispanic/Latino (90.9%), White (87.0%), cisgender women (69.5%), and heterosexual (85.0%). Seven categories emerged describing opportunities for AI and automation in financial assistance including: 1) Aid application support and process navigation 2) Conversational and interactive tools 3) Efficient search and discovery of resources 4) Insurance and billing support 5) Personalized matching to financial resources 6) Emotional and cognitive relief and 7) Connecting to human support. Six categories were identified that were focused on navigating health insurance including: 1) AI as a health insurance advocate 2) Enhancing understanding of insurance policies 3) Personalized guidance and decision support 4) Simplifying and automating administrative processes 5) Speed and convenience compared to traditional systems 6) Connecting to resources and human support. In response to both questions, survivors expressed some limitations and concerns about use of AI for FH interventions. Conclusions: Overall, cancer survivor feedback about using AI in FH interventions emphasized streamlining processes and eliminating redundancies. Findings demonstrate openness to AI or automation being used in FH interventions for cancer survivors—which could address some implementation challenges.

Ähnliche Arbeiten