Understanding Preference Heterogeneity in Digital Neurorehabilitation: Results of a Discrete Choice Experiment

Increasing demand, rising costs, and limited access pose major challenges for stroke rehabilitation. Digital interventions offer solutions, although acceptance varies. Understanding heterogeneity in preferences and decision consistency is essential.

A discrete choice experiment (N = 1055 after data cleaning) assessed preferences for 7 attributes. Latent class analysis identified preference-based subgroups; a heteroscedastic conditional logit model examined scale heterogeneity.

Three classes were identified. Class 1 (21.3%) prioritized low copayment (β_80€ = -3.355; β_0€ = 3.058; P < .001) and showed limited digital readiness, placing less emphasis on technical features. Class 2 (32.2%) focused on therapy success, with large utility differences across levels (β₆₀% = -4.035; β_100% = 3.909; P < .001) and higher digital affinity. Class 3 (46.5%) valued professional contact (β = 0.941; P < .001), location flexibility (β = 0.632; P < .001), and progress feedback (β = 0.558; P < .001), reflecting differentiated engagement preferences. The heteroscedastic conditional logit model showed that satisfaction with health (γ = 0.114; P < .001) and quality of life (γ = 0.043; P < .001) were linked to greater consistency, whereas (very) good health status (γ = -0.177; P < .001) and therapy experience (γ = -0.014; P < .001) corresponded to increased variability.

Tailored interventions should reflect distinct preference types to foster acceptance. Cost-sensitive users may need simple formats; digitally engaged individuals benefit from flexible solutions; and those with inconsistent decisions may require clearer guidance or structured onboarding. Conducted in stroke rehabilitation, the study also captured broader rehabilitation experience.