Developing a Service Quality Index System for AI Health Care Chatbots: Mixed Methods Study

TL;DR

This study creates a new system to evaluate the service quality of AI health chatbots using expert input and a structured framework.

Contribution

The paper introduces a validated service quality index system specifically for AI health care chatbots.

Findings

The final index system includes 5 primary and 17 secondary indicators grouped into five domains.

Assurance and reliability were the most weighted primary indicators in the system.

The Delphi process achieved 100% response rates and strong expert consensus.

Abstract

Artificial intelligence (AI) health care chatbots are gaining widespread adoption worldwide. It is imperative to understand the service quality of AI health care chatbots. However, there is limited guidance on how to comprehensively evaluate their service quality.

This study aimed to develop an index system based on the SERVQUAL framework for evaluating the service quality of AI health care chatbots.

An initial indicator pool was compiled through a comprehensive literature review and consultations with 4 experts. These indicators were mapped and categorized into 5 domains adapted from the SERVQUAL framework. The experts were recruited from hospital, university, and health commission settings by purposive sampling. The service quality index system was identified using a 2-round Delphi process, which included a virtual meeting between the 2 rounds. In the third round, indicator weights within each quality domain and subdomain were determined using the analytic hierarchy process.

There were 26 indicators identified in the literature, based on which the 2-round Delphi process was conducted. A total of 20 experts were invited. The response rates in both rounds of Delphi and the analytic hierarchy process were 100%, and the authoritative coefficients were both >0.7. The final service quality index system for AI health care chatbots comprises 5 primary indicators and 17 secondary indicators. There were 3 (18%) indicators on assurance, 4 (24%) on reliability, 3 (18%) on human-likeness, 4 (24%) on tangibility, and 3 (18%) on responsiveness. The primary indicators, ranked from highest to lowest weight, were assurance (0.239), reliability (0.237), human-likeness (0.187), tangibility (0.170), and responsiveness (0.167).

This study pioneers the development of a service quality index system for AI health care chatbots adapted from the SERVQUAL framework. The results provide a validated tool for evaluating the performance of chatbots and offer valuable insights for health service managers and developers to enhance AI-driven medical consultation services.

Figures