ABSTRACT
Family caregivers of autistic children are susceptible to unconfirmed fads and false claims regarding to the efficacy of unproven interventions. This study aims to predict family caregivers’ participation and adoption of unproven interventions in online communities. Based on the Diffusion of Innovations theory, we first divided the family caregivers’ adoption process into five stages: awareness, persuasion, decision, implementation, and confirmation. Social network analysis and natural language processing methods were subsequently utilised to characterise personal, environmental, and behavioural factors for predicting the formation of last three stage. The results indicated promising evidence for the application of machine learning algorithms in predicting family caregivers’ decision (AUC = 0.823), implementation (AUC = 0.887), and confirmation (AUC = 0.921). Furthermore, the results showed that factors such as social interaction, social persuasion, and modelling significantly contributed to family caregivers’ online community participation and facilitated the adoption process of unproven interventions. Family caregivers with stronger negative emotions expressed were more likely to adopt unproven interventions and recommend these interventions to other community members, thereby accelerating the diffusion of these interventions.
Acknowledgements
The authors extend their heartfelt appreciation to Elia Gabarron for her insightful advice and meaningful suggestions that significantly enhanced the quality of this research
Authors’ contributions
NZ is the principal investigator for the study, generated the idea and designed the study, was the primary writer of the manuscript, and approved all changes. QY and LH supported the writing of the manuscript. MF is the owner of the funding and participates in the revision. MJ provides suggestions for revisions to the manuscript. All authors were involved in developing, editing, reviewing, and providing feedback for this manuscript and have given approval of the final version to be published.
Availability of data and material
The material described in this manuscript is freely available. Please contact the lead author for more information.
Disclosure statement
No potential conflict of interest was reported by the author(s).