Physician perceptions of artificial intelligence in Northern Italy healthcare: a survey of fears and expectations

Abstract

Introduction: Artificial Intelligence (AI) is more and more spreading but despite the clear evidence of benefits related to its implementation, many physicians worry about ethical, legal, employment and professional changes that AI is going to induce. The purpose of this paper is to assess whether and why physicians worry about AI. Methods: This study is a cross-sectional survey addressed to a group of 362 Northern Italy hospitals physicians, both specialists and residents from selected specialties were asked to fill in a 27 multiple-choice online survey submitted by e-mail. The survey aimed to evaluate their opinions and expectations about the impact of AI on clinical, employment and ethical topics. The results were evaluated by the software Stata that enabled to carry out a multivariate analysis with the evaluation of the statistical significance of the results obtained. Results: 176 physicians (48%) answered the survey. The knowledge of the topic "AI" was reported as mild in 47%, poor in 30% and good in 15%; 98% of the responders believe that AI will improve medical activities, in particular by reducing medical errors. The legal problems, the worsening of the relationship with the patients and the deep changes of the medical role have been considered its most negative expected consequences. From an employment point of view, most responders believe that the AI cause the replacement of physicians by other professional figures. The most frequent sensations caused by AI are optimism (34%), worry (30%) and enthusiasm (13%), while anxiety is reported by 9% of the responders. The responders also believe that new dedicated digital technologies and new skills will be needed. Deep changes in the formation of physicians and residents are deemed to be necessary. Gender influences the response given on the effects of AI: women tend to be overall more pessimistic, predicting greater impacts on training, with a substantially negative feeling and with a lower probability of easing litigation. The responses are not correlated with the doctor's specialty of the respondent. The region, which influences the responses on training and feelings, does not influence the response on the effect of AI on litigation. The respondents' origins in some regions of northern Italy and the selection of some medical specialties must be considered limitations of the reported analysis.

Ähnliche Arbeiten