RESUMO
BACKGROUND: Information systems based on artificial intelligence (AI) have increasingly spurred controversies among medical professionals as they start to outperform medical experts in tasks that previously required complex human reasoning. Prior research in other contexts has shown that such a technological disruption can result in professional identity threats and provoke negative attitudes and resistance to using technology. However, little is known about how AI systems evoke professional identity threats in medical professionals and under which conditions they actually provoke negative attitudes and resistance. OBJECTIVE: The aim of this study is to investigate how medical professionals' resistance to AI can be understood because of professional identity threats and temporal perceptions of AI systems. It examines the following two dimensions of medical professional identity threat: threats to physicians' expert status (professional recognition) and threats to physicians' role as an autonomous care provider (professional capabilities). This paper assesses whether these professional identity threats predict resistance to AI systems and change in importance under the conditions of varying professional experience and varying perceived temporal relevance of AI systems. METHODS: We conducted 2 web-based surveys with 164 medical students and 42 experienced physicians across different specialties. The participants were provided with a vignette of a general medical AI system. We measured the experienced identity threats, resistance attitudes, and perceived temporal distance of AI. In a subsample, we collected additional data on the perceived identity enhancement to gain a better understanding of how the participants perceived the upcoming technological change as beyond a mere threat. Qualitative data were coded in a content analysis. Quantitative data were analyzed in regression analyses. RESULTS: Both threats to professional recognition and threats to professional capabilities contributed to perceived self-threat and resistance to AI. Self-threat was negatively associated with resistance. Threats to professional capabilities directly affected resistance to AI, whereas the effect of threats to professional recognition was fully mediated through self-threat. Medical students experienced stronger identity threats and resistance to AI than medical professionals. The temporal distance of AI changed the importance of professional identity threats. If AI systems were perceived as relevant only in the distant future, the effect of threats to professional capabilities was weaker, whereas the effect of threats to professional recognition was stronger. The effect of threats remained robust after including perceived identity enhancement. The results show that the distinct dimensions of medical professional identity are affected by the upcoming technological change through AI. CONCLUSIONS: Our findings demonstrate that AI systems can be perceived as a threat to medical professional identity. Both threats to professional recognition and threats to professional capabilities contribute to resistance attitudes toward AI and need to be considered in the implementation of AI systems in clinical practice.
RESUMO
INTRODUCTION: Quantitative computed tomography (qCT) is an emergent technique for diagnostics and research in patients with chronic obstructive pulmonary disease (COPD). qCT parameters demonstrate a correlation with pulmonary function tests and symptoms. However, qCT only provides anatomical, not functional, information. We evaluated five distinct, partial-machine learning-based mathematical models to predict lung function parameters from qCT values in comparison with pulmonary function tests. METHODS: 75 patients with diagnosed COPD underwent body plethysmography and a dose-optimized qCT examination on a third-generation, dual-source CT with inspiration and expiration. Delta values (inspiration-expiration) were calculated afterwards. Four parameters were quantified: mean lung density, lung volume low-attenuated volume, and full width at half maximum. Five models were evaluated for best prediction: average prediction, median prediction, k-nearest neighbours (kNN), gradient boosting, and multilayer perceptron. RESULTS: The lowest mean relative error (MRE) was calculated for the kNN model with 16%. Similar low MREs were found for polynomial regression as well as gradient boosting-based prediction. Other models led to higher MREs and thereby worse predictive performance. Beyond the sole MRE, distinct differences in prediction performance, dependent on the initial dataset (expiration, inspiration, delta), were found. CONCLUSION: Different, partially machine learning-based models allow the prediction of lung function values from static qCT parameters within a reasonable margin of error. Therefore, qCT parameters may contain more information than we currently utilize and can potentially augment standard functional lung testing.