Pharmacists’ Trust in Automated Pill Recognition Technology: The Role of Presenting AI Uncertainty Information (Preprint)

Abstract

<sec> <title>BACKGROUND</title> Dispensing errors significantly contribute to adverse drug events, resulting in substantial healthcare costs and patient harm. Automated pill verification technologies have been developed to aid pharmacists with medication dispensing. However, pharmacists’ trust in such automated technologies remains unexplored. </sec> <sec> <title>OBJECTIVE</title> This study aims to investigate pharmacists’ trust in automated pill verification technology designed to support medication dispensing. </sec> <sec> <title>METHODS</title> Thirty participants performed a simulated pill verification task with the help of an imperfect AI aid recommending acceptance or rejection of a filled medication. The experiment employed a mixed-subjects design. The between-subjects factor was the AI aid type, with or without an AI uncertainty plot. The within-subjects factor was the four potential verification outcomes: AI rejects the incorrect drug, AI rejects the correct drug, AI approves the incorrect drug, and AI approves the correct drug. </sec> <sec> <title>RESULTS</title> Participants had an average trust propensity score of 72 out of 100 (SD = 18.08), indicating a positive attitude towards trusting automated technologies. The introduction of an uncertainty plot to the AI aid significantly enhanced pharmacists’ end trust (t(28) = -1.854, p = .037). Trust dynamics were influenced by AI aid type and verification outcome. Specifically, pharmacists using the AI aid with the uncertainty plot had a significantly larger trust increment when AI approved the correct drug (t(78.98) = 3.93, p &lt; .001) and a significantly larger trust decrement when AI approved the incorrect drug (t(2939.72) = -4.78, p &lt; .001). Intriguingly, the absence of the uncertainty plot led to an increase in trust when AI correctly rejected an incorrect drug, whereas the presence of the plot resulted in a decrease in trust under the same circumstances (t(509.77) = -3.96, p &lt; .001). A pronounced “negativity bias” was observed, where the degree of trust reduction when AI made an error exceeded the trust gain when AI made a correct decision (z = -11.30, p &lt; .001). </sec> <sec> <title>CONCLUSIONS</title> To the best of our knowledge, our study is the first attempt to examine pharmacists’ trust in automated pill verification technology. Our findings reveal that pharmacists have a favorable disposition toward trusting automation. Moreover, providing uncertainty information about the AI’s recommendation significantly boosts pharmacists’ trust in the AI aid, highlighting the importance of developing transparent AI systems within healthcare. </sec>

Ähnliche Arbeiten