Isolating and Mitigating Sequentially Dependent Perceptual Errors in Clinical Visual Search

Remote-store-and-forward teledermatology has recently grown exponentially in popularity and use as an efficient, accurate, and cost-effective way to improve the health and well-being of countless patients. Despite advances in machine learning and computer vision, the screening and reading of dermatological images still depends on the visual system of human observers (e.g., clinicians), who receive extensive training to best recognize lesions and anomalies. In remote store-and-forward teledermatology settings, clinicians may examine hundreds of images on a daily basis, seeing several images one after the other. A main underlying assumption of their work is that clinician percepts and decisions about a current image are completely independent from prior viewings. However, we and other groups demonstrated that the visual system has visual serial dependencies (VSDs) at many levels, from perception to decision making, including in clinical tasks. These sequential dependencies, replicated hundreds of times in the literature, mean that what was seen in the past influences (and captures) what is seen and reported at this moment. Theoretically, VSDs are helpful in an autocorrelated natural world, but they are suboptimal in visual tasks conducted in artificial situations where images are not always related. Importantly, serial dependencies in perceptual processing could thus produce significant errors during diagnostic judgments of dermatological images. Our central hypothesis is that VSD can have a disruptive effect in asynchronous remote-store-and-forward teledermatology judgments that impairs accurate detection and recognition of lesions. This hypothesis is supported by our robust pilot data, which show that VSD strongly biases lesion classification in both untrained observers and expert clinicians. The rationale for the proposed research projects is that once it is known how serial dependence arises and how it impacts judgments, we can understand how to control for it. Hence, accuracy of lesion detection and diagnosis can significantly improve. The specific objectives of this proposal are to establish (Aim 1), identify (Aim 2) and mitigate (Aim 3) the impact of VSD on remote-store-and-forward dermatological judgments.

NCT Number	NCT04332783
Lead Sponsor	University of California, Berkeley
Conditions	Vision
Enrollment	10,120 participants
Start Date	2019-04-01
Primary Completion	2031-06-30 (estimated)
Study Completion	2032-10-30 (estimated)
Updated on ClinicalTrials.gov	2026-02-03