Demographics
A total of 1019 responders from 94 countries participated in this survey. The highest number of respondents is from the UK (10%), followed by Malaysia (8%), the USA (8%), Australia (6%) and Japan (5%). Fig 1 shows the geographical distribution of survey respondents. Amongst them, 35% are female, 65 % are male. The highest number of responders comes from Generation-Y, sometimes also called the Millenials, contributing to 59% of the total responders. The second largest group were from Generation X (28%), followed by the baby boomers (11%).
In terms of academic qualifications, 91% holds one or multiple postgraduate degrees. Amongst them, 34% holds a PhD degree. However, only 5% were board certified. Seventy-nine per cent of the respondents are currently working as academic and/or clinical physicists, while 10% carry multiple roles. Eleven per cent of them are currently undertaking a postgraduate course. 67% are affiliated with hospitals, and 34% works in universities. The rest of the responders are affiliated with research institutes, consultancy, government agencies, regulatory bodies, etc. There are also 20% that have multiple affiliations.
Survey results
Fig 2 shows the survey summary for Q1 to Q10 and Q13. An exceedingly high number of survey respondents (91%) agree that AI will play an essential role in medical physicists’ practice. In comparison, 85% agrees that more and more applications such as quality control, treatment planning will be performed by AI.
When asked about their opinions if AI will threaten/disrupt the medical physicists’ practice and career, the polls are more spread out. Half of them do not agree with this notion, while 25% thinks otherwise. Albeit the apprehension, 92% acknowledges that all medical physicists should acquire at least some basic knowledge of AI. Towards that end, 87% express support that AI should be taught in the postgraduate medical physics programmes.
More than half (62%) of the responders claimed that they had a basic understanding of AI relevant to their field, while 21% did not agree. However, only 34% could confidently say they have a working knowledge of AI, while 24% neither agreed nor disagreed, and 41% did not have a working knowledge of AI. In terms of skills, only 22% think they have the relevant expertise in AI, while 25% sat on the fence, and 53% did not have the appropriate skills. Upon further probing, only 14% could confidently claim that they are proficient in AI, design, code, and implement an AI program; 15% were undecided, while 71% did not agree.
Fig 3 shows AI knowledge mainly was gained through self-taught (41%) and work-related activities (25%). Attending courses and postgraduate training only contributed to 12% and 5%, respectively. We also noticed similar approaches towards developing AI skills, with 39% being self-taught. Attending courses and postgraduate training only contributed to 10% and 7%, respectively. Interestingly, while only 16% claimed that they did not know AI, a more significant percentage (40%) admitted that they had no AI skill.
Contemporary AI systems are not without limitations. Just over half of the responders (59%) thought they understood the limitations of AI, while 16% did not agree. In terms of AI acceptance of AI, 67% thought they are ready to learn and apply AI in their practice, while 15% did not agree.
Table 1 shows the results of statistical tests on the different factors affecting the levels of agreement and method of knowledge and skill acquisitions, comparing across gender, age groups, country economy status, academic qualifications and practices.
Males, more often, strongly agrees that AI will play an essential role in the practice, and they possess the working knowledge and skills in AI. Gender association was found in the method of knowledge acquisition. More male responders claimed that they were self-taught, while more females learned from work-related activities.
Table 2
Results of statistical tests on the different factors affecting the levels of agreement and method of knowledge and skill acquisitions, comparing gender, age groups, country economy status, academic qualifications and medical physics practices.
|
Questions
|
Gender
|
Country economy status
|
Academic qualifications
|
Medical physics practices
|
|
Statistical Test
|
Mann Whitney U
|
Spearman Correlation
(r, p-value)
|
Spearman Correlation
(r, p-value)
|
Kruskal Wallis
(p-value)
|
1
|
AI will play an important role in the practice of medical physicists.
|
0.019
|
0.016
0.616
|
0.050
0.110
|
0.068
|
2
|
More and more applications such as quality control, treatment planning will be performed by AI
|
0.001
|
-0.073
0.021
|
0.066
0.035
|
0.192
|
3
|
AI will threaten/disrupt the medical physicists’ practice and career.
|
0.094
|
0.061
0.052
|
-0.058
0.068
|
0.924
|
4
|
All medical physicists should acquire at least some basic knowledge of AI.
|
0.385
|
0.096
0.002
|
0.002
0.949
|
0.520
|
5
|
AI should be taught in the postgraduate medical physics programme
|
0.750
|
0.115
<0.001
|
0.030
0.343
|
0.552
|
6
|
I have a basic understanding of AI (relevant to my field).
|
0.311
|
-0.170
<0.001
|
0.170
<0.001
|
<0.001
|
7
|
I have a working knowledge of AI (relevant to my field).
|
0.023
|
0.134
<0.001
|
0.134
<0.001
|
<0.001
|
8
|
I have relevant skill in AI.
|
0.045
|
0.091
0.004
|
0.091
0.004
|
<0.001
|
9
|
I am proficient in AI (able to design, code and implement).
|
0.558
|
0.005
0.884
|
0.005
0.884
|
0.006
|
10
|
I understand the limitations of AI
|
0.325
|
0.121
<0.001
|
0.121
<0.001
|
<0.001
|
11*
|
My knowledge in AI are developed through:
|
0.019
|
0.138
|
<0.001
|
<0.001
|
12*
|
My skill in AI are developed through:
|
0.288
|
0.047
|
0.004
|
<0.001
|
13
|
I am ready to learn and apply AI in my practice
|
0.541
|
0.191
<0.001
|
-0.032
0.305
|
0.159
|
Note: Bold font indicates significant differences, p <0.05. * Q11 and Q12 report the Pearson chi-square p-values.
There was a weak correlation between the age group with Q1, 7 and 13. Older medical physicists agree that AI will play an essential role in medical physics practice, and they are ready to learn and apply AI to their practice. Younger medical physicists tend to acknowledge that they had a working knowledge of AI relevant to their field.
The country economic status affected the perception of medical physicists. Higher-income countries correlated with higher academic qualifications (r = 0.210, p<0.001). medical physicists from higher-income countries were more likely to think that QC and treatment planning will be taken over by AI, and claim to have a basic understanding of AI relevant to their field. This corroborates the findings that more medical physicists from higher-income countries claimed they could design, code and implement AI. More medical physicists from the lower-income countries believed that all MP should acquire some basic knowledge of AI and education via postgraduate programmes. They were also keen to learn and apply AI in their practice. The country economic status association was found in the method of skill acquisition. More responders from upper-middle-income countries claimed that they learned from postgraduate programmes.
A significant correlation was found between academic qualifications and Q2, 6 - 8, and Q10. Medical physicists with a PhD degree were more likely to think that AI will have more and more application in the medical physics profession. They were also more likely to say that they had a basic understanding, knowledge, and skills of AI and understand the limitations of AI.
Q6-10 showed a significant difference in academic, clinical physicists’ opinions and those involved in both areas, henceforth called mixed-field physicists. Pairwise comparisons show that academic and mixed-field physicists were more likely to agree that they have a basic understanding, working knowledge and skills in AI, and understood the limitations of AI. However, the academic physicists were also more proficient in AI and able to design, code and implement AI.
Significant difference in preferences of methods of knowledge acquisition (ꭓ2 = 36.4, p < 0.001) and skill development (ꭓ2 = 28.1, p < 0.001). While the academic and clinical physicists were mostly self-taught, the mixed-field physicists do not have any particular preferences. However, a fairly large proportion of the academic and clinical physicists claimed they have no knowledge or skill in AI.