The Usability and Trustworthiness of Medical Eye Images

TL;DR

This paper introduces EyeGuide, a mobile app that improves user ability to capture medical eye images and enhances trust in AI diagnoses, addressing usability and trust issues in mHealth eye care solutions for developing countries.

Contribution

The work presents the design, development, and evaluation of EyeGuide, a novel app that guides users in capturing eye images and increases trust in AI diagnoses, filling gaps in existing mHealth tools.

Findings

Interactive tutorials enable faster image capture than tone guidance.

Providing disease-specific background information increases trust in AI diagnoses.

EyeGuide has potential applications in AI disease detection and telemedicine.

Abstract

The majority of blindness is preventable, and is located in developing countries. While mHealth applications for retinal imaging in combination with affordable smartphone lens adaptors are a step towards better eye care access, the expert knowledge and additional hardware needed are often unavailable in developing countries. Eye screening apps without lens adaptors exist, but we do not know much about the experience of guiding users to take medical eye images. Additionally, when an AI based diagnosis is provided, trust plays an important role in ensuring in the adoption. This work addresses factors that impact the usability and trustworthiness dimensions of mHealth applications. We present the design, development and evaluation of EyeGuide, a mobile app that assists users in taking medical eye images using only their smartphone camera. In a study (n=28) we observed that users of an interactive tutorial captured images faster compared to audible tone based guidance. In a second study (n=40) we found out that providing disease-specific background information was the most effective factor to increase trustworthiness in the AI based diagnosis. Application areas of EyeGuide are AI based disease detection and telemedicine examinations.