Mapping Mpox vaccine hesitancy using an integrated machine learning and structural equation modeling approach.

This study employs an integrated computational approach to investigate Mpox vaccine intention in Bangladesh as Mpox immunisation strategies require a thorough understanding of behavioural determinants. Data from 405 participants were analysed using a structured framework that included frequentist statistics to examine demographic relationships, t-distributed Stochastic Neighbour Embedding (t-SNE) for unsupervised grouping, and structural equation modelling (SEM) to confirm underlying behavioural trends. About 77% of people were willing to get vaccinated, and this was mainly influenced by their age (p = 0.002), education (p = 0.014), and occupation (p < 0.001). The t-SNE analysis revealed two patterns of reluctance-one linked to practical barriers (mainly younger people with limited education), and another linked to more neutral or skeptical attitudes. Structural equation modelling (SEM) confirmed that perceived vaccine efficacy (vaccine knowledge) was the primary driver of intention (β = 0.346, p = 0.005). Social-structural factors significantly shaped this intent; notably, living without family was a positive predictor ( β = 0.185, p < 0.001), whereas 96% of the hesitant group resided in traditional family units. Furthermore, a strong correlation between personal intent and community advocacy (r = 0.58) suggests that prosocial motivation is a key driver in this cohort. Our results show that vaccine reluctance is not uniform. In South Asia, public health interventions need to move away from generic messages and towards customised approaches that target the particular "efficacy gaps" and structural obstacles found by this multi-method mapping.