Attitude and Readiness towards Artificial Intelligence and its Utilisation: A Cross-sectional Study among Undergraduate Medical Students in a Medical College, Kolkata

AI is a hot topic in recent days... We students of IPGME&R, Kolkata, India have done a study on Attitude, Readiness and Utilization of AI by medical students.
Artificial Intelligence (AI): The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.
Our study showed that:
1. Nearly half of the study participants showed a favorable attitude towards role of AI in healthcare
2. Around three-fifth of the participants could define basic concepts of data sciences and AI and were ready to choose AI based applications for healthcare; they were willing to accept AI usage despite feeling a lack of cognitive skills
3. Most of them used AI-based applications for studying (ChatGPT), however, some of them faced difficulties in using them

Thank you!

1. Attitude and Readiness towards Artificial Intelligence and its Utilisation : A Cross-sectional Study among Undergraduate Medical Students in a Medical College, Kolkata Group B1 of Batch 2022 (Roll nos. 51-75)

2. Introduction 1

3. Introduction Artificial Intelligence (AI) : The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages. [1] The advantages of using AI in medicine are that many AI-driven software, chatbots, and robotics surgery can help in increasing knowledge and efficiency, diagnosing, counseling patients, and assisting surgeons during procedures to reduce human errors and increase precision. [2 ] 1) Catherine Soanes, Sara Hawker, Julia Elliott (eds). Pocket Oxford English Dictionary. 10th Edn.New Delhi.Oxford University Press.2005;44 2) Driver CN, Bowles BS, Bartholmai BJ, Greenberg- Worisek AJ. Artificial Intelligence in Radiology: A Call for Thoughtful Application. Clin Transl Sci. 2020 Mar;13(2):216-218. 2

4. Introduction (contd.) A study conducted by Al Zaabi A et al. in 2023 in Gujarat, India, revealed that 87.2% of physicians and medical students are interested in learning the application of AI in healthcare. [3] Another study by Garrel et al. in 2023 in Germany, showed that every fourth student (25.2%) uses AI-based tools and applications very frequently. [4] 3) AlZaabi A, AlMaskari S, et al. Are physicians and medical students ready for artificial intelligence applications in healthcare? Digit Health. 2023 Jan 26;9:20552076231152167. 4) von Garrel , J., Mayer, J. Artificial Intelligence in studies—use of ChatGPT and AI-based tools among students in Germany.  Humanit Soc Sci Commun   10 , 799 (2023). https://doi.org/10.1057/s41599-023-02304-7 3

5. Introduction (contd.) There is dearth of studies regarding attitude, readiness, and utilization of AI among undergraduate medical students in West Bengal. Hence, assessment of the attitude, readiness, and utilization of AI among the undergraduate medical students is crucial for understanding the way AI is gradually transforming healthcare, from the grassroot level, so that the future doctors would be better equipped with the knowledge and skills for applying AI in healthcare delivery judiciously. With this background in mind, the current study was conducted among undergraduate students of a medical college in Kolkata. 4

6. Research question 5

7. Research Question What is the attitude, readiness, and utilization of Artificial Intelligence (AI) among the undergraduate medical students of a medical college in Kolkata? 6

8.  Objectives 7

9. Objectives To evaluate the attitude towards AI among u ndergraduate m edical s tudents of a medical c ollege in Kolkata To identify the readiness towards AI among the study participants To assess the utilization of AI among the study participants 8

10.  Methodology 9

11. Methodology Study Type : Descriptive, observational study Study Design : Cross-sectional Study Setting : Institute of Post Graduate Medical Education and Research (IPGME&R), Kolkata Study Duration : 10 th June 2024 to 6 th July 2024 10 The M.B.B.S. course started in IPGME&R, Kolkata in 2004 with a batch strength of 100 and since 2019 with the new C.B.M.E. curriculum, the annual batch strength has been increased to 200. So, our study population including the current four phases of M.B.B.S. was 800.

12. Methodology (contd.) Figure 1. Timeline of the research project represented with a Gantt chart 11

13. Methodology (contd.) Study Participants: Undergraduate medical students belonging to Phase I, II, and Part-1 and 2 of Phase III of IPGME&R, Kolkata Selection criteria : Inclusion criteria: Undergraduate medical students who filled up the Microsoft form Exclusion criteria: The students who did not provide consent to participate in the study Those who could not be approached 12

14. Methodology (contd.) Sample size : The sample size (n) was obtained by using Cochran’s formula, which is- n=z 2 pq/d 2 z → Standard normal deviate= 1.96 , p → Prevalence (assuming readiness of medical students towards AI= 50%), q → (1-p ) d → Relative error ( 10% of p) After applying 15% non-response rate, the final sample size was calculated as 443 [ At Confidence Interval (CI) of 95%, power (1- β ) = 80%] Sampling Technique : Simple random sampling 13

15. Methodology (contd.) Study tools : A predesigned, pretested, and semi-structured questionnaire, collecting data across the following domains- Sociodemographic characteristics of the study participants Attitude towards AI: 11 questions were adopted fr om a questionnaire previously developed by Sit et al. (to evaluate UK medical students’ attitudes toward AI ) to assess attitude in this study. The attitude questions were framed into a 5 -point Likert scale, ranging from “strongly agree” to “strongly disagree”. [4] 4) Sit C, Srinivasan R, Amlani A, Muthuswamy K, Azam A, Monzon L, Poon DS. Attitudes and perceptions of UK medical students towards artificial intelligence and radiology: a multicentre survey. Insights Imaging. 2020 Feb 5;11(1):14. 14

16. Methodology (contd.) Study tools (contd.) : 3. MAIRS-MS (Medical Artificial Intelligence Readiness Scale for Medical Students) scale [4] : It is a reliable tool for evaluating the perceived readiness levels of medical students on AI technology, and is validated by Karaca et al. It is a 5-point Likert scale, consisting of 22 items for assessment of readiness, divided under 4 domains: cognition, ability, vision, and ethics. 4) Sit C, Srinivasan R, Amlani A, Muthuswamy K, Azam A, Monzon L, Poon DS. Attitudes and perceptions of UK medical students towards artificial intelligence and radiology: a multicentre survey. Insights Imaging. 2020 Feb 5;11(1):14. 15

17. Methodology (contd.) Study tools (contd.) : Cognition domain (8 questions): knowledge of AI terminology and logic Ability domain (8 questions): readiness in choosing AI applications Vision domain (3 questions): the ability to explain AI's limitations, strengths, and weaknesses Ethics domain (3 questions): adherence to legal and ethical regulations when using AI 16

18. Methodology (contd.) Study tools (contd.) : Each item on the MAIRS-MS is scored between 1 (minimum) to 5 (maximum) points as follows- “Strongly disagree” is given a score of 1 “Disagree” is given a score of 2 “Neutral” is given a score of 3 “Agree” is given a score of 4 “Strongly agree” is given a score of 5 4. Utilisation of AI among the study participants 17

19. Methodology (contd.) Study technique: The questionnaire was prepared using Microsoft forms and shared with the study participants through online method The study participants were well-versed in English language, so the questionnaire was developed in English. Data collection was carried out for 1 week 18

20. Methodology (contd.) Statistical analysis: Data were tabulated and analysed using Microsoft Office Excel (version 2021) Descriptive statistics were represented using Mean (± S.D.), Median, frequency, and percentage, along with suitable diagrams wherever applicable Attitude towards AI was assessed by the Attitude scale developed by Sit et al. , which is a 5-point Likert scale, having a total of 11 items for the attitude domain, with options for each item ranging from ‘strongly agree’ (score 5) to ‘strongly disagree’ (score 1) 19

21. Methodology (contd.) Statistical analysis (contd.): The Item number 3 in the attitude domain of the scale was reversely scored (‘strongly agree’- score 1; ‘strongly disagree- score 5’ ) Attitude was categorized as ‘favourable’ and ‘unfavourable’ based on the median of the overall attitude score. Score ≥ median was classified as ‘favourable’ while score < median was considered as ‘unfavourable’ Mean (± S.D.) scores for each of the four sub-domains under the readiness domain of MAIRS-MS were calculated, and an overall Mean (± S.D.) score was obtained by summing up the mean scores of the four sub-domains 20

22. Methodology (contd.) Ethical considerations : The proposal for the study was written to obtain clearance from the Institutional Ethics Committee (IEC) of IPGME&R, Kolkata The study participants were explained the purpose of the study and online consent (via Microsoft forms ) was obtained from them Anonymity and confidentiality of data were maintained throughout the study 21

23. Methodology (contd.) Operational definitions : 1. Attitude: the way that one thinks and feels about somebody/something or how you behave towards somebody/something that shows how you think and feel [1] 2. Readiness: the state of being ready or prepared for something [1] 3. Utilization: the act of using something, especially for a practical purpose [1] 1) Catherine Soanes, Sara Hawker , Julia Elliott (eds). Pocket Oxford English Dictionary. 10th Edn . New Delhi.Oxford University Press.2005;50, 751-2, 1017 22

24. Methodology (contd.) Workplan of the project: Literature review was conducted via dedicated online databases ( PubMed, Google Scholar ) for the selection of topic for the project After topic selection, the title was finalized, proposal was written and submitted for obtaining ethical clearance from the IEC of IPGME&R, Kolkata The questionnaire was designed (in Microsoft Forms) and pretesting was done Questionnaire was distributed via online method to the study participants 23

25. Methodology (contd.) Workplan of the project (contd.): The purpose of the study was explained to the participants and electronic consent was obtained from them through the Microsoft forms prior to data collection Data collection, analyses and interpretation were done A PowerPoint presentation of the project was prepared 24

26. RESULTS 25

27. Table 1- D istribution of study participants according to their Age-group and Gender (n=443) Socio-demographic characteristics Frequency   Percentage (%) Age  group (in completed years) 18 -21       256   57.78 22-24     153  34.53 ≥ 25    34  7.62 Total 443 100 Gender   Male       329   74.26 Female     114    25.74 Total 443 100 Inference: 74.26% of the study participants were males and 57.78% belonged to the age group of 18-21 years 26

28. Figure 2 : Pie diagram showing the distribution of study participants according to their P hase of MBBS (n= 443 ) Inference : 32.27% participants belonged to Phase I while 29.79% belonged to Phase II of MBBS 27

29. 28 Table 2- D istribution of study participants according to their current and permanent residences (n=443) Socio-demographic characteristics Frequency   Percentage (%) Current residence  Hostel     273   61.62 Home     130   29.34 Paying Guest 40 9.02 Total 443 100 P ermanent residence Urban and/or Suburban area     272   57.78 R ural area   171  34.53 Total 443 100 Inference: 61.62% of the study participants currently reside in hostel while 61.39% of the participants belong to urban and/or sub-urban areas

30. 29 Figure s 3 and 4 : Doughnut diagrams showing the distribution of study participants according to the highest level of education attained by their fathers and mothers (n= 443 ) Inference : Fathers of 67.49% of participants and mothers of 53.95% of participants had completed their graduation

31. Figure 5 : Bar diagram showing the distribution of study participants based on sources of information regarding AI (n=443)* Inference : Source of information for most of the study participants was socia l media (378) followed by friends (238) * multiple responses 30

32. Table 4 : Distribution of study participants according to their attitude towards AI (n=443) Sl. No. Items Strongly Disagree Disagree  Neutral  Agree  Strongly Agree 1. AI will play an important role in healthcare Frequency (%) 1. AI will play an important role in healthcare 25 (5.64) 15 (3.38) 59 (13.32) 196 (44.24) 148 (33.42) 2. I am less likely to consider a career in radiology, given the advancement of AI 54 (12.19) 99 (22.35) 150 (33.86) 96 (21.67) 44 (9.93) 3. Some specialties will be replaced by AI during my lifetime 44 (9.93) 64 (14.44) 97 (21.89) 167 (37.69) 71 (16.02) 31

33. Sl. No. Items Strongly Disagree Disagree  Neutral  Agree  Strongly Agree 4. I have an understanding of the basic computational principles of AI Frequency (%) 4. I have an understanding of the basic computational principles of AI 44 (9.93) 82 (18.51) 129 (29.11) 134 (30.25) 54 (12.20) 5. I am comfortable with the nomenclature related to artificial intelligence 34 (7.67) 66 (14.89) 149 (33.6 9 ) 145 (32.73) 49 (11.06) 6. I have an understanding of the limitations of artificial intelligence 26 (5.68) 45 (10.15) 109 (24.60) 199 (44.92) 64 (14.44 ) Table 4 contd. 32

34. Sl. No. Items Strongly Disagree Disagree  Neutral  Agree  Strongly Agree 7. Teaching in artificial intelligence will be beneficial for my career Frequency(%) 7. Teaching in artificial intelligence will be beneficial for my career 27 (6.09) 33 (7.45) 118 (26.63) 165 (37.24) 100 (22.57) 8. All medical students should receive teaching in artificial intelligence 34 (7.67) 45 (10.15) 82 (18.51) 167 (37.69) 115 (26.00) 9. At the end of my medical degree, I will be confident in using basic healthcare AI tools if required 23 (5.19) 31 (6.99) 127 (28.67) 173 (39.06) 89 (20.00) Table 4 contd. 33

35. Sl. No. Items Strongly Disagree Disagree  Neutral  Agree  Strongly Agree 10. At the end of my medical degree, I will better understand the methods used to assess healthcare AI algorithm performance. Frequency (%) 10. At the end of my medical degree, I will better understand the methods used to assess healthcare AI algorithm performance. 22 (4.96) 33 (7.44) 118 (26.63) 193 (43.56) 77 (17.38) 11. Overall, At the end of my medical degree, I feel I will possess the knowledge needed to work with AI in routine clinical practice 22 (4.96) 38 (8.57) 119 (26.86) 178 (40.18) 86 (19.41) Table 4 contd. 34

36. Inference from Table 4: 44.24% of the study participants have agreed that AI will play an important role in healthcare 22 .35% of the study participants have disagreed about considering career in radiology, given the advancement of AI 37.69% of the study participants agreed that some specialties will be replaced by AI in their lifetime 30.25% of the study participants agreed to have a basic understanding of the computational principles of AI 35

37. Inference from Table 4 (contd.): 32.73% of the study participants agreed with the nomenclature related to AI 44.92% of the study participants agreed that they had an understanding of the limitations of AI 37.24% of the study participants agreed that teaching in AI will be beneficial for careers 37.69% of the study participants agreed that all medical students should receive teaching in AI 36

38. Inference from Table 4 (contd.): 39.06% of the m agreed that they will be confident in using basic healthcare AI tools if required at the end of their medical degree 43.56% agreed that they will have a better understanding of the methods used to assess healthcare AI algorithm performance at the end of their medical degree 40.18% agreed that they feel they will possess the knowledge needed to work with AI in routine clinical practice at the end of their medical degree 37

39. Figure 6: Pie chart showing distribution of the study participants based on their attitude towards AI (n=443) Inference : 56.20% of the study participants had favourable attitude towards AI The median of the overall attitude score was 15 38

40. Figure 7 : Component -Bar Diagram showing the d istribution of study participants according to their readiness on the cognitive factor towards AI implementation (n=443) 39

41. Inference from Figure 7: 24.61% of study participants disagreed with explaining the basic concepts of data science 36.12% of the study participants agreed to have a basic conception of statistics 28.89% of the study participants disagreed with explaining how AI systems are trained 26.18% of study participants disagreed to define the basic concepts and terminology of AI 40

42. Inference from Figure 7 (contd..): 22.57 % of study participants disagreed to properly analyse the data obtained by AI in healthcare Around one-fourth of the study participants (25%) disagreed with differentiating between the functions and features of AI-related tools and applications. 25.51% of the study participants disagreed with organising workflows by the logic of AI 29.35 % of the participants agreed to ex press the importance of data collection, analysis, evaluation, and safety; for the development of AI in healthcare 41

43. Figure 8 : Component -Bar Diagram showing the d istribution of study participants according to their readiness on ability factor towards AI implementation (n=443) 42

44. Inference from Figure 8 : 3 0.02 % of the study participants have agreed to choose AI applications for the problem encountered in healthcare 31.37 % of the study participants have agreed to e xplain the AI applications used in healthcare services to the patient 37.69 % of the study participants have agreed that they found it valuable to use AI for education, service, and research purposes 3 1.15 % of the participants agreed to explain how AI applications in healthcare offer a solution to which problem 43

45. Inference from Figure 8 (contd.) : 37.92% of the study participants agreed to access, evaluate, use, share and create new knowledge using information and communication technologies 41.08 % of the study participants agreed to use AI applications in accordance with its purpose 32.27 % of the study participants agreed to use AI technologies effectively and efficiently in healthcare delivery 31.82 % of the study participants agreed to use AI-based information in combination with my professional knowledge 44

46. Figure 9 : Component- Bar Diagram showing the d istribution of study participants according to their readiness on vision factor towards AI implementation (n=443) 45

47. Inference from Figure 9: 41.76% of the study participants agreed to foresee the opportunities and threats that AI technology can create 42.21% of the study participants agreed to explain the strengths and limitations of AI technology 34.76% of the study participants agreed to explain the limitations of AI technology 46

48. Figure 10. Component -Bar Diagram showing the d istribution of study participants according to their readiness on ethical factors towards AI implementation (n=443) 47

49. Inference from Figure 10: 40.63% of the study participants agreed to follow the legal regulations regarding the use of AI technology in healthcare 41.08% of study participants agreed to act in accordance with ethical principles while using AI technology 39.27% of the study participants agreed to use health data in accordance with legal and ethical norms while using AI technology 48

50. Mean ± SD Range Cognitive Factor 19.86 ± 6.84 8-40 Ability Factor 25.70 ± 8.09 8-40 Vision 10.04 ± 3.02 3-15 Ethics 10.13 ± 2.99 3-15 Total Mean 66.13 ± 17.45 22-110 Table 5 : Distribution of Mean and Standard deviation (S.D.) scores of each sub-domain under the ‘readiness’ towards AI domain (n=443) Inference: Overall mean and S.D. score of readiness towards AI was 66.13 ± 17.45 The ability factor domain was found to have the highest Mean ± SD score ( 25.70 ± 8.09 ) 49

51. Figure 11. Bar of Pie diagram showing distribution of study participants based on various types of AI application used by them (n*= 443) Inference : About 66% of the participants used an AI application, of which ChatGPT (248) was the most commonly used, followed by Google Assistant (215) ( Others- Meta AI, Bixby, copilot,gemini etc ) *Multiple responses 50 (Meta AI, August AI, Research Rabbit, Gemini AI)

52. Figure 1 2 . Pie of pie diagram showing the distribution of study participants according to the frequency of usage of AI-based applications (n 1 =292 ) Inference : Most of the participants used an AI-based application weekly (50%) and among them, 34% used AI-based application once a week. Occasional 51 Daily Once weekly

53. Figure 13 . Bar diagram showing the distribution of study participants based on the frequency of usage of AI-based applications (n 2 =87 ) Inference : Among daily users (30%), most of them (64%) used AI-based applications for less than an hour 52

54. Figure 14 . Bar diagram showing the distribution of study participants according to their purpose of AI application usage (n 1 = 292)* Inference : Most of the study participants used AI-based applications for the purpose of studying (283) followed by entertainment (187) (*Multiple responses) 53

55. Figure 1 5 . Bar of pie diagram showing the distribution of study participants based on various difficulties faced by them while using AI applications (n 1 =292) Inference : 12% of the study participants faced difficulty while using AI applications, as most of them were unable to interpret (11) Faulty data 54

56. Figure 1 6 : Bar diagram showing distribution of study participants according to the different AI based applications that were first used by them ( n 1 = 292)* Inference : Nearly two-fifth (127) of study participants used Google assistant as the first AI based application (*Multiple responses) 55

57. Summary 56

58. Summary A descriptive, cross-sectional study was carried out from 10th June- 6th July 2024, among 443 undergraduate medical students of Phase I, II and Part 1 and 2 of Phase III of IPGME&R, Kolkata, selected using a purposive sampling technique A predesigned, pretested, and semi- structured questionnaire was used to collect data through Microsoft forms , containing sociodemographic characteristics and attitudes towards artificial intelligence scale developed by Sit et al ., MAIRS-MS scale, and utilization of AI in healthcare Data were tabulated and analysed using MS Office Excel 2021 57

59. Summary (contd.) Sociodemographic profile: 57.78% of the study participants belonged to the age group of 18-21 years 74.26% of the study participants were males and 32.27% of the study participants belonged to Phase 1 of the academic year 67.49% of fathers and 53.95% of mothers of the study participants had completed their graduation The most common source of information for the study participants on AI was social media (378) 58

60. Summary (contd.) Attitude towards AI: Around 44% of the study participants have agreed that AI will play an important role in healthcare, while 37% of the study participants agreed that some specialties will be replaced by AI in their lifetime 30% of the study participants agreed to have a basic understanding of the computational principles of AI Almost 38% of the study participants agreed that all medical students should receive teaching in artificial intelligence 39% of the study participants agreed that they will be confident in using basic healthcare AI tools if required at the end of their medical degree 59

61. Summary (contd.) Readiness towards AI: Around 45 % of the study participants disagreed with explaining how AI systems are trained while 42.70 % of the study participants disagreed with defining the basic concepts and terminology of AI 22.57% of the study participants disagreed to properly analyse the data obtained by AI in healthcare About one-fourth (25%) of the study participants disagreed to differentiate between the functions and features of AI-related tools and applications. 60

62. Summary (contd.) Readiness towards AI (contd.): 31.37 % of the study participants agreed to explain the AI applications used in healthcare services to the patient while 37.69 % of the study participants found it valuable to use AI for education, service, and research purposes 32.27 % of the study participants agreed to using AI technologies effectively and efficiently in healthcare delivery and 31.82% of the study participants agreed to use AI-based information in combination with their professional knowledge 61

63. Summary (contd.) Readiness towards AI (contd.): 42.21% of the study participants agreed to explain the strengths and limitations of AI technology while 41.76% of study participants agreed to foresee the opportunities and threats that AI technology can create 40.63% of the study participants agreed to follow the legal regulations regarding the use of AI technology in healthcare 62

64. Summary (contd.) Utilization of AI: 66.0% of the study participants use an AI-based application, out of which, 85.0% of the study participants use ChatGPT Among 96.9% of the study participants, studying is the most common purpose for using an AI-based application Almost half of them (50%) use it every week more specifically once a week (34.2%) 63

65. CONCLUSION 64 64

66. Conclusion 65 Nearly half of the study participants showed a favorable attitude towards role of AI in healthcare Around three-fifth of the participants could define basic concepts of data sciences and AI and were ready to choose AI based applications for healthcare; they were willing to accept AI usage despite feeling a lack of cognitive skills Most of them used AI-based applications for studying (ChatGPT) , however, some of them faced difficulties in using them 65

67.  STRENGTHS, LIMITATIONS AND RECOMMENDATIONS 66

68. Strengths 68 There is a scarcity of studies on attitude, readiness, and utilization of AI among undergraduate medical students of West Bengal Large sample size Study was done on undergraduate medical students in all four phases

69. Limitation The study could have been conducted among undergraduate students in all medical colleges across West Bengal, but it could not be done due to time constraints 69

70. Recommendations 67 Incorporation of AI-related foundation courses, lectures, workshops, seminars, etc. into the medical curriculum Creation of opportunity by the Government for the use of AI in healthcare and the teaching-learning process Development of ethical guidelines for AI utilization in healthcare, and recommend continuous monitoring and evaluation of AI systems once implemented

71. Acknowledgement 70 We would like to thank our respected Director, Prof (Dr.) Manimoy Bandopadhyay . We would like to thank Prof. (Dr.) Avijit Hazra , Dean of Students Affairs, IPGME&R. We would like to thank Prof. (Dr.) Mausumi Basu , Head, Dept. of Community Medicine, IPGME&R. We extend our gratitude to Dr. Kuntala Ray, Associate Professor, IPGME&R, for guiding us throughout the project. We would like to thank Dr. Shalini Pattanayak , Dr. Kalpana Gupta , and Dr. Subhosri Saha , Post Graduate Trainees, Dept. of Community Medicine, IPGME&R, for constantly guiding us throughout the project. We would also like to thank the entire Community Medicine department of IPGME&R, for giving us an amazing opportunity to work on this project. Finally, we would also like to thank all the participants for providing the data required for the project.