Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Theme-Specific Responses to Sustain Talk in AI-Delivered Motivational Interviewing: A Secondary Analysis of Emergency Department Counseling Sessions
ABSTRACT
Background:
Large language models can generate motivational interviewing-consistent responses, but their responses to client sustain talk require evaluation before clinical use.
Objective:
This formative secondary analysis examined whether sustain-talk themes varied by substance and whether an artificial intelligence counseling agent's response patterns varied by sustain-talk theme.
Methods:
We coded conversational turns from brief counseling sessions with non-treatment-seeking adult emergency department patients focused on alcohol (n=68), cannabis (n=48), or nicotine (n=32). Participant sustain-talk turns were classified into eight themes. Agent responses were coded as discrepancy-only, validation-only, balanced, or neutral-only. Generalized estimating equation models accounted for responses clustered within sessions. Secondary models examined language in the next participant turn.
Results:
Across 477 responses, 50.7% were discrepancy-only, 11.5% validation-only, 8.2% balanced, and 29.6% neutral-only. Relative to therapeutic-coping sustain talk, low-readiness sustain talk was associated with lower odds of discrepancy-only responding (OR, 0.20; 95% CI, 0.08-0.49) and higher odds of validation-only responding (OR, 11.05; 95% CI, 3.87-31.53). In a secondary multinomial model, validation-only responses were associated with a lower relative risk of next-turn sustain talk versus neutral language (RRR, 0.19; 95% CI, 0.09-0.41).
Conclusions:
This formative evaluation identified theme-specific response patterns and potential gaps for prospective testing AI counseling agents. These observational findings support prospective testing of theme-matched response strategies before large-scale deployment. Clinical Trial: NA
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.