Assessing the Impact of Automated Suggestions on Decision Making: Domain\n Experts Mediate Model Errors but Take Less Initiative

Abstract

Automated decision support can accelerate tedious tasks as users can focus\ntheir attention where it is needed most. However, a key concern is whether\nusers overly trust or cede agency to automation. In this paper, we investigate\nthe effects of introducing automation to annotating clinical texts--a\nmulti-step, error-prone task of identifying clinical concepts (e.g.,\nprocedures) in medical notes, and mapping them to labels in a large ontology.\nWe consider two forms of decision aid: recommending which labels to map\nconcepts to, and pre-populating annotation suggestions. Through laboratory\nstudies, we find that 18 clinicians generally build intuition of when to rely\non automation and when to exercise their own judgement. However, when presented\nwith fully pre-populated suggestions, these expert users exhibit less agency:\naccepting improper mentions, and taking less initiative in creating additional\nannotations. Our findings inform how systems and algorithms should be designed\nto mitigate the observed issues.\n

Ähnliche Arbeiten