Paper Type

Short

Paper Number

PACIS2026-1562

Description

Front-line healthcare systems face persistent demand–capacity gaps in initial consultation, triage, and history taking. AI-enabled symptom checkers and diagnostic chatbots have emerged as scalable tools to ease primary care pressure, yet evidence remains limited on how interaction design shapes patient behavior and triage outcomes. This study examines how dialogue design and explanation type jointly influence patients’ intention to reuse the system and triage accuracy through psychological mechanisms. We conduct a 2×2 experiment (structured vs. unstructured dialogue * outcome-anchored vs. evidence-anchored explanation) with a physician baseline condition, using an AI diagnostic system for mental healthcare and recruiting patients and physicians. We expect unstructured dialogue to increase reuse intention, and explanation type to moderate psychological safety and perceived control across dialogue structures. The findings aim to advance theory on human–AI interaction in diagnosis and inform the design of patient-facing medical AI interfaces that improve user experience without compromising clinical accuracy.

Comments

14-Healthcare

Share

COinS
 
Jul 5th, 12:00 AM

Navigating the Dialogue Structure and Explanation Style in AI Diagnostics

Front-line healthcare systems face persistent demand–capacity gaps in initial consultation, triage, and history taking. AI-enabled symptom checkers and diagnostic chatbots have emerged as scalable tools to ease primary care pressure, yet evidence remains limited on how interaction design shapes patient behavior and triage outcomes. This study examines how dialogue design and explanation type jointly influence patients’ intention to reuse the system and triage accuracy through psychological mechanisms. We conduct a 2×2 experiment (structured vs. unstructured dialogue * outcome-anchored vs. evidence-anchored explanation) with a physician baseline condition, using an AI diagnostic system for mental healthcare and recruiting patients and physicians. We expect unstructured dialogue to increase reuse intention, and explanation type to moderate psychological safety and perceived control across dialogue structures. The findings aim to advance theory on human–AI interaction in diagnosis and inform the design of patient-facing medical AI interfaces that improve user experience without compromising clinical accuracy.