Readiness, knowledge, and perception towards artificial intelligence of medical students at faculty of medicine, Pelita Harapan University, Indonesia: a cross sectional study

Abstract

Introduction

Artificial intelligence (AI) enables machines to perform many complicated human skills which require various levels of human intelligence. In the field of medicine, AI helps physicians in making diagnoses and treatments for patients with more efficiency, accuracy, and precision. In order to prepare medical students who are the future healthcare workforce, it is important to enhance their readiness, knowledge and perception toward AI. This study aims to assess Pelita Harapan University (PHU) medical students’ readiness, knowledge, and perception toward AI.

Methods

A quantitative cross-sectional study was conducted to assess respondents’ readiness, knowledge and perception toward AI. An online questionnaire was distributed via Google Forms to all batch of medical students. Medical Artificial Intelligence Readiness Scale for Medical Students (MAIRS-MS) questionnaire was used to evaluate AI readiness, while an adapted questionnaires was used to evaluate knowledge and perception toward AI. Data were then analyzed using IBM Statistical Package for Social Sciences (SPSS) version 23.0.

Results

A total of 650 respondents were included in this study. Most respondents were in pre-clinical phase (88%) while the remaining were in clinical phase (12%). Overall, the total mean score for AI readiness was 73.34 of 100. Respondents had a mean score 24.52 ± 5.26 of 40, 27.78 ± 4.65 of 40, 10.57 ± 2.07 of 15, and 10.47 ± 2.00 of 15 in the cognitive, ability, vision, and ethics domain respectively. Generally, respondents had sufficient knowledge and positive perception toward AI. There were also significant correlation between readiness and knowledge with gender, having studied coding previously in high school, and having family or close friends working in AI field. Social media also had a good influence on enchancing readiness in the domain of ability and ethics, and perception towards AI.

Conclusion

Medical students of PHU mostly showed neutral to favorable response on readiness, knowledge, and perception towards AI. Incorporating AI into high school and medical curriculum is an important step to prepare medical students’ encounter and partnership with AI as the future workforce in medicine.

Introduction

Background

Artificial intelligence (AI) is a technology that allows machines to mimic many complicated human skills. AI is capable of carrying out tasks that would usually require basic to moderate human intelligence or interaction, either on its own or in conjunction with other technologies [1]. et al. Another review by Alowais et.al. depicted the role of AI in clinical practice which comprises of (1) AI assistance in diagnostics; (2) AI assistance in treatment; (3) AI assistance in population health management; (4) AI-powered patient care [2].

The application of AI in the field of medicine has been used since the second half of the 20^thcentury. Originally known as computer-assisted medical technology, it performed routine medical tasks. The Early Detection and Prevention System (EDPS) was established in 1998 with the purpose of providing guidance to nurses and paramedics in rural Indian clinics that lack medical professionals. The system demonstrated a diagnostic accuracy rate of 94% for a total of 933 patients, and patients reported it to be more precise. In 1986, the University of Massachusetts developed DXplain, a system that produces differential diagnoses, which rekindled enthusiasm for AI in the medical domain. In the 2000s, DeepQA was created as a system that utilized natural language processing and search algorithms to analyze data and generate responses. Subsequently, this approach was implemented in the field of medicine by extracting data from a patient's medical record in order to generate responses based on empirical evidence [3].

AI helps physicians to better identify patients who need extra attention and provide personalized protocols for each individual on a computer application. AI can be used by primary care physicians to analyze their discussions with patients, create notes and have the necessary information entered directly into the electronic health record (EHR) system. Several advantages of using AI in medicine include efficiency, accuracy, precision, decreased workload, increased patient face time, increased time on critical cases, saves money, and better monitoring [4].

et al.On the other hand, a study conducted by Baigi et al., found that medical students have a promising positive attitude towards AI in the medical field; however, most students have low knowledge and limited skills in working with AI [5]. This was also found by Tung et al.., whose results found that medical students in Malaysia had shown high awareness of AI but needed improvement in working together with AI [6].

Justification of study

In the medical field, AI rapidly transforms various sectors, including healthcare. AI holds immense potential to enhance treatment planning, diagnostics, and patient care. However, one factor that determines the successful integration of AI into healthcare relies heavily on the readiness, knowledge, and perception of medical professionals, particularly medical students who are the future workforce. By evaluating medical students' readiness, knowledge, and perception towards AI in healthcare, we could address an essential step for preparing the future workforce to harness the full potential of AI technologies; and therefore allow an effective integration of AI which will ultimately supports patient care and healthcare delivery.

Research aims

This study aims to assess the readiness, knowledge and perception of Pelita Harapan University Faculty of Medicine students' regarding AI, assess students’ readiness to adopt AI technologies in medical settings, assess students’ level of knowledge about the basic concepts of AI in medicine and assess students’ perception towards AI in medical settings.

Methods

Study design

A quantitative study with a cross-sectional design was used to identify the readiness, knowledge, and perception of Pelita Harapan University medical students’ towards AI. Researchers conducted a survey on pre-clinical and clinical phase medical students of the Faculty of Medicine, Pelita Harapan University using a questionnaire via Google Forms. The respondents gave their consent by ticking an agreement check box, and then they filled out the questionnaire anonymously. Respondents were given an information sheet about the research being conducted attached to the form.

study population

Study population of this study is active pre-clinical or clinical students who are registered as medical students of undergraduate study at the Faculty of Medicine, Pelita Harapan University. In Pelita Harapan University, medical students undertake a six-year program, consisting of three and a half years in the pre-clinical phase followed by two and a half year in the clinical phase. The curriculum is based on outcome-based education, with a strong emphasis on problem-based learning. There is currently no specific course about AI in the curriculum of Faculty of Medicine, Pelita Harapan University. Topics about AI are integrated in elective course of Health Information System in 7^th semester.

All students were sent an invitation to participate in this research on a voluntary basis. Medical students who were dropped out from their undergraduate program, switching majors, filled in questionnaire incompeletely, and did not provide consent to participate in this study were excluded.

Data collection

Data collection was conducted over a period of three months, from January to March 2024. This study was ethically approved by the Medical Research and Ethics Committee of the Faculty of Medicine, Pelita Harapan University (No: 205/K-LKJ/ETIK/XII/2023) on December 19^th, 2023.

Questionnaire information

The first section of this questionnaire collected demographic data from respondents such as age, gender, batch, college financing, educational background, parents' last education, and where the information about AI was obtained. The second section assesses readiness for AI, respondents filled out the Medical Artificial Intelligence Readiness Scale for Medical Students (MAIRS-MS) questionnaire, an instrument developed by Karaca et al. [7]. It was developed to assess readiness by dividing it into 4 domains, including cognition (8 questions), ability (8 questions), vision (3 questions) and ethics (3 questions) using 5-point Likert scale. Each question on the MAIRS-MS questionnaire has a minimum score of 1 and a maximum of 5. Score 1 will be intepreted as "strongly disagree", score 2 as "disagree", score 3 as "neutral", score 4 as "agree" and score 5 as "strongly agree". MAIRS-MS has previously been used to evaluate readiness of medical student in Malaysia towards AI [7, 8]. The researchers of those studies provided validity evidence from a health professions education setting. In current study, MAIRS-MS was used to assess readiness towards AI in medical students in Indonesia, because of the similarities in culture and demographic composition between the two countries as fellow countries in Southeast Asia. Therefore, this questionnaire was chosen to ensure the relevance and accuracy of the research results. Sections three and four evaluate the level of knowledge and perceptions of medical students using an adapted questionnaire previously developed and used by Al Hadithy et al. [8]. Knowledge variable consists of 5 questions and perception consists of 13 questions. Knowledge questions use a 5-point Likert scale with similar interpretations as the MAIRS-MS questions, while perception questions rating from 1 which were interpret as "Extremely unlikely", score 2 as "unlikely", score 3 as "neutral", score 4 as "likely" and score 5 as "extremely likely" [9]. The knowledge questionnaire added more detailed terms on AI that is not covered in MAIRS-MS, while the perceptions questionnaire measures medical students’ perception towards AI as a technology that could impact healthcare professionals' employment prospects [10]. Currently, the cut-off value for the MAIRS-MS, attitude and perception scale have not yet been established. Therefore, it is not possible to determine whether a candidate is "adequately ready" or not, has “positive attitude” or negative attitude”, or ‘positive perception” or “negative perception” towards AI.

Data analysis

The data obtained was then analyzed using the International Business Machines Statistical Package for Social Sciences (IBM SPSS) version 23.0. The data normality test was carried out for numeric data using the One-Sample Kolmogorov-Smirnov test. Mean scores and standard deviations were calculated for numeric data from each domain on readiness, knowledge, and perception. Descriptive data is also presented for each variable using percentages and horizontal bar charts. Independent Sample t-Test is used to see whether there is a difference in the average score of two unpaired samples among variables associated with readiness, knowledge, and perception. p < 0.05 were considered statistically significant and reported in this study. The data visualization was created using Microsoft Word’s tool for drawing chart.

Results

Sociodemographics characteristics of respondents

Table 1. shows the distribution of respondents’ characteristic which included in this study. The total number of medical students who filled out the questionnaire was 650 respondents, of which 572 (88.0%) were pre-clinical students and the other 78 (12.0%) were clinical students.

Table 1 Sociodemographics characteristics of respondents

Respondents’ readiness towards AI

The results of this study show that respondents' scores were relatively lower in the cognitive and ability domains compared to the vision and ethics domains (Table 2).

Table 2 Mean MAIRS-MS scores of respondents

Respondents’ readiness in the cognitive domain

Table 3 and Figure 1 show the respondents’ readiness towards AI on the cognitive domain.

Table 3 Respondents’ readiness in the cognitive domain

Fig. 1

Chart of respondents’ readiness in the cognitive domain

Respondents’ readiness in the ability domain

Table 4 and Figure 2 show the respondents readiness towards AI on the ability domain.

Table 4 Respondents’ readiness in the ability domain

Fig. 2

Chart of respondents’ readiness in the ability domain

Respondents’ readiness in the vision domain

Table 5 and Figure 3 show the respondents’ readiness towards AI on the vision domain.

Table 5 Respondents’ readiness in the vision domain

Fig. 3

Chart of respondents’ readiness in the vision domain

Respondents’ readiness in the ethic domain

Table 6 and Figure 4 show the respondents’ readiness towards AI on the ethic domain.

Table 6 Respondents’ readiness in the ethic domain

Fig. 4

Chart of respondents’ readiness in the ethic domain

Respondents’ knowledge towards AI

Table 7 and Figure 5 show the respondents’ knowledge towards AI.

Table 7 Respondents’ knowledge towards AI

Fig. 5

Chart of respondents’ knowledge towards AI

Respondents’ perception towards AI

Table 8 and Figure 6 show the respondents’ perception towards AI.

Table 8 Respondents’ perception towards AI

Fig. 6

Chart of respondents’ perception towards AI

Factors associated with respondents’ readiness, knowledge, and perception towards AI

Table 9 shows factors associated with respondents’ readiness towards AI, while Table 10 shows factors associated with respondents’ knowledge and perception towards AI.

Table 9 Factors associated with respondents’ readiness towards AI (t-test analysis)

Table 10 Factors associated with respondents’ knowledge and perception towards AI (t-test analysis)

In the domain of cognitive, ability, vision, ethics, and knowledge, male respondents, respondents who had studied coding in high school and respondents who had family or close friends working in the AI field scored significantly higher than female respondents, respondents who had not studied coding in high school and respondents who did not have family or close friends working in the AI field. It was also found that respondents who know AI from friends scored higher on the cognitive, cognitive, ability and vision domain than respondents who do not know AI from friends. Respondents who knew about AI from social media also scored higher than respondents who did not know about AI from social media in the domain of ability and ethics, and perception. Then, respondents who knew about AI from blogs and television shows better score in domain of cognitive and ability, and knowledge than respondents who did not know AI from internet blog and television. Also, respondents on clinical phase scored higher than respondents on pre-clinical phase in the domain of ability and ethics (Tables 9 and 10).

Discussion

Respondents’ readiness towards AI

Respondents rated their readiness high in the domain of vision and ethics, but rated their readiness low in the domain of cognitive and ability. For ability, vision, and ethics domain, the results are similar, there are more respondents who chose agree than strongly agree, but there are also many respondents chosing neutral, while disagree and strongly disagree were low in result.

The majority of respondents had a neutral stance towards all statements in the cognitive domain of MAIRS-MS. All of the questions had a frequency over 40% for neutral, over 20% for agree, and less than 10% for strongly agree. These results differ from an earlier study conducted by Tung et al.. [6] whereas the majority of Malaysian Medical students had disagreed towards cognitive factors of AI, excluding two questions where, “I can define the basic concepts of data science” was neutral and “I can define the basic concepts of statistics” was agreed.

This study found that the domain of ability had mixed stances towards the questions. Unlike the domain of cognitive where all of the eight questions had a neutral stance, the ability domain had two questions where the majority of respondents agreed with a frequency of almost 50%, with a close second being a neutral stance. However, the other six questions in the ability domain were similar in that they had a neutral stance followed by an agreed statement. Compared to Tung et al.. [6], majority of the answers were similar in terms of their stances. Both majority and second following were neutral and agreed for most questions. However, their study was more favored to the agreeing side compared to this study which was more lenient towards the neutral stance.

Specifically, the majority of statements in the vision and ethics domain of the MAIRS-MS scale garner more agreement from respondents than disagreement. These results are also consistent with the responses of Malaysian medical students [6]. In this study, respondents who chose strongly disagree and disagree are much lower than in Malaysian medical students. The results of this study showed that most respondents were neutral rather than showing agreement or disagreement. This response differs from that of Malaysian medical students, who typically give minimal neutral responses. In their case, most of the Malaysian respondents would either agree or disagree rather than give a neutral answer.

According to a study conducted by Chyung et al., respondents do not always interpret the question given and use a midpoint as a dumping ground when they are responding to a survey and questions they are unfamiliar with. [10] Past research proposes that when presented with a neutral response choice, individuals will be more likely to choose that choice than report their genuine conclusion. Three variables likely impact a participant's choice to falsely report through the neutral alternative: cognitive effort, ambivalence, and social desirability. People tend to gravitate towards neutrality in public supposition surveys because they need to strategically distance themselves from the negative sentiments associated with their conflicting views on an issue. Also, choosing a neutral choice allows individuals to avoid the cognitive effort required to select between their positive and negative sentiments on an issue [9, 11, 12].

Respondents’ knowledge towards AI

Knowledge wise, a majority of the students were familiar with the terms “artificial intelligence”, “machine learning”, and “algorithm” and while there was a balance knowledge wise on the term “neural network”. These results display that the students have at least some basic knowledge surrounding AI, but not fully understanding its intricacies. The results of this study were relevant to study conducted by Al Hadithy et al. [8] which also showed that most of the medical student in Oman had adequate knowledge as reflected in familiarity with the words "artificial intelligence", "algorithm", and "machine learning". These findings indicate that basical concept of knowledge in AI have been introduced and understood for medical students in Indonesia or Oman. Although, spesific term such as “neural networks” needs more-in-depth and specific learning to understand them because of its complexity. To close this gap, educational curriculum should be expanded to include in-depth discussions of more difficult AI ideas, such as neural networks.

Respondents’ perception toward AI

The findings in this study provide vital insights into the viewpoints of medical students regarding the potential impact of AI on healthcare. The survey data indicates that the respondents have a predominantly neutral viewpoint regarding the capacity of AI to replace or to assist the different jobs carried out by physicians or healthcare workers. This impartiality implies either a positive outlook or a lack of certainty regarding the capabilities of AI. The survey indicates that respondents possess a higher level of confidence in the capability of AI to aid or potentially substitute healthcare professionals in tasks such as diagnostic imaging, generating a prognosis, and documentation. The belief of the students are consistent with a review by Alowais et al. that describe how AI can help identify abnormalities and provide quantitative measurements for faster and more accurate medical diagnosis, and be used as a tool to analyze complex datasets, predict outcomes, and optimize treatment strategies to provide personalized treatments [2]. The survey data also reveals significant skepticism about AI’s ability to provide emotional support and providing psychiatric counseling. This skepticism is well founded, because AI currently lacks the ability to replicate empathy and creating a bond with humans which are all an important thing to consider in all the aspects written above. Studies have also noted that while AI can aid healthcare personnel, the component where human interaction is required remains vital. The survey data on AI indicated that respondents hold the belief that AI should not replace human healthcare workers. Instead, they believe that AI should be used to enhance and support these workers. AI has the potential to streamline processes and offer valuable insights [13]. However, it is important to note that the human touch is still considered essential, particularly for emotional support and personalized care.

Factors associated with respondents’ readiness, knowledge, and perception towards AI

From the analysis, it was found that males have higher score compared to females in domain of cognitive, ability, vision, and ethics. This finding is contradictory to a previous research made by Xuan et al., whereby in their research shows no significant average difference in the score based on gender [14]. Therefore, males scored higher than females in knowledge towards AI. These findings contradict with a study conducted by Jha et al. which stated that there is no significant difference score between males or females [15]. This might be caused that there are still educational gaps in terms of access and others factor such as peers, parents, teachers, school counselors, and gender stereotypes in the field of AI. This also influence that there is still lower participation of females in AI-related academic research compared to males [16].

Respondents that have studied coding in high school were found to score higher on average in domain of cognitive, ability, vision and ethics, and knowledge compared to those that did not study coding in high school. A review stated that learning coding significantly enchanced cognitive and abilities, such as critical thinking and problem solving, which can be beneficial for AI-related skills [17]. On other hand, students who studied coding in high school before could increase vision-related AI to increase their understanding of complex concepts of thinking and learning opportunities. This promotes self-directed skill acquisition and continuing intellectual improvement regarding to AI [8]. Dabbagh et al. stated that introducing AI education in school curricula could improves ethical views on AI by provide student about the foundation to understand the risks, ethical considerations, and impact to society to augment human capabilities in the future [18]. Incorporating AI in educational curricula could help them to understand the responsibilities and impact of AI, so that it can encourage responsible innovation and right decision making [19]. Findings on knowledge towards AI in this study were contradict with Jha et al.. which showed that no differences according to training in programming/coding before [15].

Respondents with family or close friends working in AI field had higher average scores in the domain of cognitive, ability, vision, ethics, and knowledge than respondents who do not have family or close friends working in AI field. These findings are consistent with Gerlich et al., who said that social influencers such as family and close friends can be demonstrated to adopt the traditions and behaviours of their social groupings, particularly when it comes to intentions to use AI, hence enhancing readiness to AI. [20] Exposure to AI from an early age from close friends or family who worked in the AI field could provide informal learning opportunities and deeper understanding so that increase knowledge about AI [21].

Therefore, respondents who know AI from friends showed a higher score in domain of cognivite, ability and vision than who do not know AI from friends. These findings are supported by Kandoth et al., that stated social influences from friend has a significant positive impact on students intention to use AI and increase their personal innovativeness [22]. Then, other study have not researched about the influence of social media on readiness, knowledge, and perception towards AI on medical students. Respondents who know about AI from social media were found to score higher in domain of ability and ethics, and perception compared to who do not know about AI from social media. Past research showed that social media platforms could assist students to receive campaigns, educational, and promotional content regarding AI, especially in healthcare field [23]. Social media has influenced almost all medical students. So, social media could be an effective way for increasing readiness, knowledge, and perception of medical students towards AI in healthcare.

Respondents who know about AI from internet blog and television were found to score higher than who do not know about AI from internet blog and television in domain of cognitive and ability, and knowledge. This is in accordance with the result from a study in Lebanon that found that medical students receiving their knowledge from mass media such as internet blog or television had higher score in knowledge and attitudes toward AI [24].

Lastly, clinical students were also found to score higher in readiness, knowledge, and perception towards AI, and significantly higher on ability and ethics domain rather than pre-clinical students. This might be caused that students who are already in the clinical phase will receive more exposure to AI-based technology in the health sector, whether in hospitals or community health centers than pre-clinical medical students which was inline with a study by Doumat [24]. On the contrary, Xuan et al. found that pre-clinical students had significantly higher scores on ability, vision, and ethics domains of MAIRS-MS compared to clinical students, although they did not elaborate on the reasons of their findings [14].

Limitation and future directions

The respondents of this study are medical students in one university in Indonesia, Faculty of Medicine, Pelita Harapan University. The study's scope is limited and its conclusions cannot be extended to other groups, such as individuals from various academic fields, universities, or the general public. Furthermore, data for the study is acquired through a self-administered Google Forms questionnaire. This data collection method is susceptible to biases such as recollection bias, confirmation bias, and framing bias. Furthermore, a large number of the study's replies are neutral, which may be influenced by the respondents' prior experience with AI in the healthcare field. And lastly, most of the respondents are pre-clinical students, which limits the generalization the results of this study.

Conclusion

In conclusion, medical students of Faculty of Medicine, Pelita Harapan University mostly showed neutral to favorable response on readiness, knowledge, and perception towards AI.. Readiness in vision and ethics domain of readiness rated higher than domain of cognitive and ability. Besides that, exposure to AI from school, family, friends, social media, internet blog, and television mostly has a good influence in readiness, knowledge, and perception towards AI. Lastly, clinical students were also found to score higher on ability and ethics domain of readiness rather than pre-clinical students. In order to prepare the medical students as the future workforce in medicine, educational curriculum in Indonesia should start incorporating AI into the education system, starting from high school to the medical education curriculum.