In a recent pre-print study posted to the medRxiv* server, a team of researchers assessed medical students' understanding, perceptions, and educational preferences related to artificial intelligence (AI) to equip them with knowledge and skills for ethical and effective AI utilization in medicine.
Study: Medical Students’ Attitudes toward AI in Medicine and their Expectations for Medical Education. Image Credit: PopTika/Shutterstock.com
*Important notice: medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.
Background
AI is now widely used in the field of medicine. AI uses algorithms and software to analyze digital information to diagnose and suggest therapies.
AI also plays a vital role in evaluating diagnostic images such as skin images and computed tomography (CT) scans and hence can act as a decision-support system for doctors while diagnosing diseases.
AI can also be helpful in various other medical fields like drug designing and personalization of treatment. However, there is limited knowledge about the level of understanding among prospective medical doctors regarding AI and its applications in medicine.
About the study
The authors of the present study conducted an online survey in which medical students were asked about their knowledge of associating AI with medicine. The questionnaire related to the survey was sent to medical students of a German university via e-mail on November 2022.
The students voluntarily participated and provided written informed consent, for which they were not given any compensation or rewards. The students took around 10 minutes to complete the survey in which they were first asked about their knowledge of employing AI in medical field and their understanding of AI in general.
The participants were then provided with a neutral definition of AI in medicine so that the participants could have basic knowledge about the topic.
Next, by using a combination of Likert Scales and semantic differential scales, the researchers collected valuable data on how participants perceived AI's reliability, technical competence, credibility, trustworthiness in the medical context, intelligence, and anthropomorphism.
These assessments helped in understanding the attitudes and beliefs of the participants towards AI, providing valuable insights into the human-AI interaction and acceptance within the studied context.
The selected scales were based on previous research methodologies, ensuring the study's validity and comparability with existing literature.
Finally, the participants were asked to share their views about including AI in university courses and about the particular aspects that should be incorporated into medical education.
Study results
The results showed that out of 84 individuals who initiated the survey, 26 discontinued, leaving 58 medical students (average age 24.51 years, standard deviation [SD]=3.56 years) who completed it.
A significant 94.83% of the participants were cognizant of AI's role in medicine, demonstrating a sound understanding of AI by identifying algorithms (58.62%), machine learning (48.28%), and neural networks (8.62%) as key components. They primarily associated AI's medical application with diagnostics (86.21%) and surgeries (27.59%).
Participants perceived AI in medicine as trustworthy (M=3.58; SD=0.71), fairly reliable (M=3.30; SD=0.69), and technically proficient (M=3.26; SD=0.71), but less credible (M=2.34; SD=0.71). They also considered AI to be intelligent (M=3.75; SD=0.66), but lacked anthropomorphic (M=1.99; SD=0.64) attributes.
Their experience with AI was moderate (M=2.85; SD=1.41), as was their exposure to AI in an educational (M=2.67; SD=1.47) or medical context (M=2.69; SD=1.43).
However, they expressed strong interest in AI's medical potential (M=4.52; SD=0.71), a desire to know more about AI in general (M=4.38; SD=0.83), and a wish for more extensive AI coverage in medical education (M=4.17; SD=0.92).
A substantial 86.21% of participants agreed that basic AI knowledge should be incorporated into medical studies. They particularly endorsed teaching about AI's operation modes (77.59%), ethics (75.86%), application areas (75.86%), reliability (94.83%), and potential risks (89.66%). However, they were less supportive of future developments (46.55%) and teaching legal aspects (46.55%).
Participants highlighted potential AI issues in medicine, including ethical concerns (53.45%), control loss (43.10%), and possible reliability issues (34.48%).
Conclusions
To summarize, the results of this study suggested that the participants had a strong interest in the application of AI in the field of medicine and had expressed their desire to learn more about it during their academic careers.
Therefore, it is now essential for medical schools to include AI education in their curricula as it will help students to gain the necessary knowledge and skills to effectively and ethically utilize AI in their future practice.
Further, to address the potential risks associated with AI in medicine, it is important to teach the importance of human oversight and ways to monitor and fix errors in AI algorithms, ensuring that the final decisions are made by human medical practitioners.
*Important notice: medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.
Study shows large language models susceptible to misinformation