Abstract
Background
Artificial intelligence (AI) is poised to reshape cardiothoracic (CT) surgery. Prior studies suggest different attitudes and adoption rates of AI based on gender. This study explores these gender-based differences further to identify domains in which differences might exist.
Methods
A survey addressing the use of AI in CT surgery was created and distributed to members of multiple international CT surgery societies. Results were collected electronically. Responses were scored on a 5-point Likert scale and then collapsed into 3 categories for χ2 analysis. Free text responses were organized thematically.
Results
Data from 412 CT surgeons were analyzed, including 344 men and 68 women specializing in cardiac (n = 213) and thoracic (n = 176) surgery. Geographic representation included North America (65%), Europe (24%), and Asia (11%). In comparing subspecialties, female CT surgeons were more likely than male surgeons to assign a limited role to AI. North American female surgeons saw the smallest role for AI. Female surgeons’ hopes for AI focused on reductions in administrative burden and efficiency gains, while men sought improved diagnostic accuracy and decision making support. Women were worried about systemic biases and clinical dehumanization, while men were concerned with the technical implementation of AI.
Conclusions
Gender is associated with CT surgeons’ perceptions of AI in clinical care, with women, particularly thoracic subspecialists, expressing more reservations than men. It is important to understand these different views, as adoption of AI in CT surgery may have an important impact on gender equity in the field.
Key Words: AI, artificial intelligence, large language models, cardiothoracic surgery, survey
Graphical Abstract
Graphical abstract depicting the methods, results, and implications of this study.
Gender-based consideration of the risk and benefits of AI integration into CT surgery.
Central Message.
Gender-based differences in AI perception in cardiothoracic surgery are significantly associated with subspecialty and geography. Understanding these differences is crucial for integration of AI into the field.
Perspective.
Artificial intelligence (AI) is rapidly integrating into clinical medicine and cardiothoracic surgery. To optimize the benefit of this technology, it is important to understand surgeons’ perceptions of and goals for the technology. These insights are significantly influenced by gender and other demographic characteristics. Recognizing these differences will aid the safe and equitable adoption of AI.
The use of artificial intelligence (AI) in medicine is advancing rapidly, with applications spanning diagnostics, predictive modeling, and perioperative management.1, 2, 3 In cardiothoracic (CT) surgery, AI has the potential to improve imaging interpretation, patient selection, intraoperative support, and postoperative care4; however, optimal implementation remains uncertain, and surgeon acceptance likely will help determine the final role and degree of AI integration. Surveys of CT surgeons have revealed both diverse and varied uses of AI, as well as attitudes regarding its future adoption and role; generally, however, there is optimism surrounding the value of AI in diagnostics and administrative tasks.5
Previous work has suggested differences in the adoption of new technology—specifically AI—based on gender, with women often expressing greater caution. Women appear to use AI technology less often and express more concerns regarding its implementation. With the rapidity and scope of AI integration into business and medicine, such biased adoption potentially could yield differential productivity, quality, and pay gaps.6,7 Because CT surgery already exhibits significant gender disparities in workforce composition, remuneration, and leadership roles, such differences may further impact AI acceptance and gender inequity.8,9 We sought to assess the results of an international, multisocietal survey regarding AI implementation in CT surgery to determine whether similar gender-related trends might be found among practicing CT surgeons.
Methods
In collaboration with the University of Chicago Survey Lab, a 25-question instrument was developed to evaluate surgeons’ perceptions of AI in domains including its role, value, threat, and regulation. The majority of questions were answered using a 5-point Likert scale ranging from strongly negative to strongly positive, while several questions requested free-text responses. The survey was translated into French, Spanish, Italian, German, Japanese, and Korean and distributed electronically in May and June 2024 to members of the Society of Thoracic Surgeons, European Association for Cardio-Thoracic Surgery, European Society of Thoracic Surgeons, and Asian Society for Cardiovascular and Thoracic Surgery. Participation was voluntary and anonymous, and Institutional Review Board review was exempted (University of Chicago IRB24-0545; April 4, 2024).
The total number of distributed surveys was not made available to us by the individual societies for the sake of member privacy; a conservative estimate is >5000. A total of 453 responses were collected, with 424 complete or nearly complete, and 412 responses included gender data. Gender was self-reported, and male/female and man/woman are used interchangeably in this report to reference that respondent preference. Demographic data also included specialty (cardiac, thoracic), practice type, region, and years of experience. For specialty categorization, we defined specialization by the specialty that accounted for ≥75% of a surgeon's practice. For practice type, private practice and military practice were excluded from analysis owing to the small numbers represented.
Likert scale responses were collapsed into 3 categories—negative, neutral, and positive—for statistical analyses. Free-text responses were categorized thematically and subjected to word frequency analysis. The χ2 test was used to compare categorical responses among demographic groups. Gender-stratified analyses evaluated differences in responses as a whole and in relation to other demographic data. Significance was defined as P < .05.
Results
The 412 respondents included 344 (83%) men and 68 (17%) women. The majority of respondents were cardiac surgeons; thoracic surgeons were both generally overrepresented and had a statistically higher representation among women (P = .027). Among the 87% of surgeons who stated their type of employment, 78% reported working at an academic institution or university. Male surgeons were skewed significantly toward being older, with 54% having >20 years in practice. On the other hand, 59% of women respondents had been practicing for ≤10 years or were in training (P < .001). The majority of respondents practiced in North America, and this was even more pronounced in the female surgeon population; conversely, a much lower number of women were practicing in Asia compared to their male counterparts (P = .020) (Table 1). For statistical analyses, regions were grouped as “North America” and “other,” given the relatively low number of European and Asian respondents.
Table 1.
Demographic data of survey respondents categorized by gender
| Domain | Variable | Males (N = 344), n (%) | Females (N = 68), n (%) | P value |
|---|---|---|---|---|
| Specialty | Adult cardiac | 185 (53.8) | 28 (41.2) | .027 |
| General thoracic | 138 (40.1) | 38 (55.9) | ||
| Other or missing | 21 (6.1) | 2 (2.9) | ||
| Practice setting | Academic or university | 217 (63.1) | 50 (73.6) | .149 |
| Community system | 68 (19.8) | 9 (13.2) | ||
| Other or missing | 59 (17.1) | 9 (13.2) | ||
| Practice location | North America | 208 (60.5) | 48 (70.6) | .020 |
| Europe | 77 (22.4) | 18 (26.4) | ||
| Asia | 44 (12.8) | 1 (1.5) | ||
| Other or missing | 15 (4.3) | 1 (1.5) | ||
| Practice duration | In training | 18 (5.2) | 7 (10.3) | <.001 |
| ≤10 y | 67 (19.5) | 33 (48.5) | ||
| 11-20 y | 73 (21.2) | 16 (23.5) | ||
| >20 y | 186 (54.1) | 12 (17.7) |
Likert scale responses were evaluated in relation to gender, and significant differences were noted in multiple domains (Table E1, Table E2, Table E3).
Preoperative Evaluation and Postoperative Care
Surgeons generally trended toward predicting a larger role for AI in the next 5 years in preoperative patient evaluation, although there was a statistically significant difference in responses between genders (P = .036). While only 20% of male surgeons envisioned a small role for AI in this domain, female surgeons’ responses were distributed more evenly. Survey responses reflected a similar trend when surgeons were asked about the future role of AI in postoperative care. Women envisioned a significantly smaller role for AI compared to men (P = .003) (Figure 1).
Figure 1.
Gender-based differences in surgeon perceptions of the future role of AI. Numbers are the percentage of responses per value category.
Surgical Decision Making
Surgeons overall saw a limited future role for AI in surgical decision making, both preoperatively and intraoperatively. Regarding the role of AI in helping choose the proper surgical treatment, there were no important gender-related differences, with balanced distribution among those predicting small, medium, and large roles. On the other hand, when asked about the role of AI in intraoperative decision making, surgeons nearly unanimously predicted a small role, with no important gender differences. Only 13% of respondents predicted a large role for AI in intraoperative decision making over the next 5 years (Figure 1).
Differences by Specialty
When survey responses were categorized by gender as well as surgeon specialty, several interesting trends emerged. In the domain of preoperative care, surgeons generally predicted a large role for AI in patient screening; however, there were statistically significant differences in responses regarding the role and value of AI in preoperative patient evaluation, preoperative decision making, and identifying surgical treatment options (P = .001, .047, and .025, respectively). In all these cases, male cardiac surgeons saw a large role of AI, while female cardiac surgeons envisioned a much smaller role, often less than their thoracic counterparts. These differences also were evident in the realm of postoperative care, with 43% of male cardiac surgeons predicting a large role for AI and female surgeons predicting a smaller role (P = .001). Interestingly, on this question, female thoracic surgeons predicted a much smaller role than their cardiac counterparts, with 50% of respondents expecting a small role for AI and only 11% anticipating a large role. Mid-career female CT surgeons envisioned a particularly small role for AI (63%) (P = .009).
Finally, there were significant differences between genders and specialties (P = .031). When asked whether AI would make accurate, independent, and reliable clinical decisions within the next 10 years, a plurality (42%) of male cardiac surgeons thought AI was very likely to make good decisions, while all other groups responded in the other extreme, saying that AI was only slightly likely to make good decisions. Of these, female cardiac surgeons were least likely to predict good AI decisions (18%), followed closely by male thoracic surgeons (26%) (Figure 2).
Figure 2.
Gender-based perception differences by surgeon subspecialty.
Differences by Region
Similar to surgical specialty, there were significant differences in survey responses by gender and location of practice. Nearly one-half of North American female surgeons predicted a small role for AI in preoperative evaluation, while more than one-half of their non–North American counterparts predicted a large role. This difference between North Americans and others also was seen in male respondents, although to a lesser degree (P < .001). The trend of North American surgeons envisioning a lesser role for AI continued in the realms of preoperative decision making, determination of surgical options, intraoperative decision making and technique, and even postoperative care. In the latter 2 categories, differences were minimal between men and women but large in relation to geography, with non–North American women surgeons predicting the largest role for AI (P < .001 for each category). There also were significant differences in responses regarding the effect of AI on advanced practice providers, overhead costs, and physician job security. Non–North American women surgeons predicted the greatest threat of AI in all of these categories. There were also significant differences relating to the impact of AI on physician–patient relationships, with non–North Americans predicting the largest positive impact on relationships but, concurrently, the largest possible threat (Figure 3).
Figure 3.
Gender-based perception differences by surgeon region of practice.
Consensus
Despite the aforementioned differences, there was general agreement between the genders in multiple other categories. In the clinical realm, all surgeons saw a small role for AI in intraoperative decision making (61%) and technical tasks (67%) but predicted a large role in image interpretation (78%), and 90% of respondents predicted improved diagnostic accuracy. The predicted value of AI in aiding patient interactions was mixed but did not differ by gender or specialty; surgeon responses did differ significantly by the experience of surgeon, although without a clear pattern (P = .021). Surgeons anticipated little effect on job security, operative volume, patient relationships, or physician autonomy, and in fact predicted improved operative outcomes (59%) and decreased error rates (63%) with AI.
Surgeons were particularly enthusiastic about the potential of AI in the administrative realm, regardless of gender or specialty. Most respondents thought AI would be very valuable in note writing (69%), outpatient testing (59%), billing and coding (84%), and quality improvement (64%).
Finally, regarding AI regulation, responses were largely similar between groups. While most surgeons felt that AI could be used for administrative tasks without trials or extensive oversight (66%), they agreed that clinical use can and should be regulated externally (67%). They also agreed that as part of AI development, regulation, and integration, the CT specialty and societies should partner with industry (80%), allow access to educational content (81%), and allow access to clinical databases (69%).
Qualitative Hopes and Fears
Word frequency analysis of subjective survey responses, followed by summarization and categorization by AI, revealed that male and female surgeons typically shared hopes and fears regarding AI. The top 3 hopes of AI for both genders were reduction of administrative burden, increased diagnostic accuracy, and improved overall work efficiency. Both men and women also agreed on fears, the top two being clinical decision errors and loss of autonomy (Tables E4 and E5). Further examination of these categories revealed several small differences in how genders prioritized each hope and fear, however. Female surgeons prioritized operational improvements more than their male counterparts, hoping for less administrative burden and hoping for efficiency gains. In contrast, male surgeons focused on clinical enhancements such as improved diagnostic accuracy and clinical decision support to a greater degree than their female counterparts. There were also several areas that did not garner a large percentage of responses but were very skewed toward a particular gender. Among men, 3.1% hoped for technology integration into practice, while no women declared this hope. Conversely, female surgeons hoped for improved error reduction and greater AI equity and access twice as often as male surgeons. Similar themes continued in relation to surgeons’ fears regarding AI. Male surgeons expressed more concerns about technical implementation of AI, while female surgeons placed more emphasis on bias implications and clinical decision risks (Figure 4).
Figure 4.
Categorized distribution of male and female surgeons with hopes (A) and fears (B) of/for AI integration.
Discussion
AI is evolving at a rapid pace and is integrating into many aspects of society, including the realms of medicine and surgery. While machine learning has been a part of developing medical technology for quite some time, natural language processing and large language models have opened new avenues for AI growth and also given the general public exposure to the technology.10 Gender-based differences in technology adoption have been documented in healthcare and in other industries,11,12 and this is also true for AI. In CT surgery, surgeons’ views regarding AI have been shown to be related to gender.5
Although the adoption of AI in medicine appears to be rapid and inextricable, the oversight of this process appears to be lagging far behind. Thus, understanding the nuances of CT surgeons’ perceptions of AI is critical to achieving successful and safe integration of the technology into the field. As such, we felt it important to further investigate these gender-based differences in perspective relating to AI in CT surgery.
This international, multisocietal survey demonstrates that gender is significantly associated with CT surgeons’ perceptions of AI in clinical practice, and that those influences are further related to other demographic factors. There were differences in response rates by gender, primarily relating to practice location and specialty type. Whereas our previous study demonstrated that women and thoracic surgeons envisioned a smaller clinical role of AI, these perceptions are in fact more complex when analyzed by multiple domains concurrently.5 In this subgroup analysis, in multiple domains, female surgeons, particularly female thoracic surgeons, saw little role for AI. However, subspecialty differences often overpowered gender difference; male thoracic surgeons frequently saw a smaller role for AI compared to cardiac surgeons of either gender.
The relationship between geography and gender was similarly complex. Overall, female North American CT surgeons perceived the smallest role for AI. On the other hand, non-North American women often envisioned the largest role for AI, and so while there were clear differences in opinion among subgroups, these differences appear to be impacted by both geography and gender in an intertwined manner.
These findings suggest that the integration of AI into CT surgery may be shaped not only by technical feasibility, but also by sociocultural and demographic factors that affect professional adoption. Our results align with prior studies reporting gender-based differences in technology adoption across medicine and other fields. The women in our study consistently prioritized AI's potential to reduce administrative burdens and improve workflow efficiency, reflecting broader concerns about equitable workload distribution and career sustainability in surgical practice. These themes parallel findings in psychology and workforce studies in which women have been shown to express heightened “AI anxiety” and to approach emerging technologies with greater caution, a reality that may be linked to concerns regarding systemic bias and unequal application.13,14 Men, in contrast, emphasized diagnostic augmentation and clinical decision support, generally demonstrating stronger optimism about AI's technical applications.
The implications of our findings merit attention. Female surgeons’ heightened concerns about systemic bias and clinical dehumanization amplify inequities already present in the field and described in AI training.15,16 Addressing these concerns will require rigorous external oversight, transparent algorithm development, and the active participation of diverse stakeholders, including women and underrepresented groups in surgery, throughout the design and implementation process. Importantly, our respondents agreed that professional societies should play a leading role in setting regulatory standards and fostering equitable access to AI resources, echoing calls from ethicists and surgical leaders.
Despite these differences, our study also identified several domains with strong consensus. Across genders and specialties, respondents predicted a limited role for AI in intraoperative decision making and technical execution, reflecting a persistent belief in the centrality of surgeon judgment and manual expertise. Conversely, there was nearly universal optimism about AI's role in imaging interpretation, diagnostic accuracy, and error reduction. These domains are consistent with areas of rapid AI development in medicine, such as radiology and dermatology, where deep learning systems have already demonstrated near-expert performance.2,17,18 Enthusiasm for administrative applications also resonates with broader discussions on physician burnout and the potential for AI to streamline documentation, billing, and quality monitoring.
This study has several limitations. Response rates could not be calculated owing to societies’ privacy restrictions, and the sample may not fully reflect the global distribution of CT surgeons. Women comprised only 17% of respondents, mirroring ongoing gender imbalances in the specialty but limiting subgroup analyses. Additionally, female surgeons were overrepresented in the early career and general thoracic categories. Although subspecialty and career length were statistically analyzed individually, this imbalance compared to male surgeons potentially introduces bias. Finally, non–North American surgeons were generally underrepresented. Given that overall, women predicted a smaller value of AI while non–North American women predicted the largest value, we can clearly see a geographic bias in the overall results due to 72% of female surgeon respondents practicing in North America. Nevertheless, the breadth of geographic and specialty representation strengthens the generalizability of these findings and underscores the importance of recognizing demographic heterogeneity in shaping AI adoption.
In conclusion, while gender is significantly related to perceptions of AI among CT surgeons, with female surgeons demonstrating particular caution regarding bias, equity, and clinical care, gender differences are influenced by multiple other domains, making simple generalizations inaccurate and unfair. These findings highlight the need for inclusive strategies in AI development, regulation, and integration to ensure that advances in surgical technology address not only technical performance but also equity, trust, and sustainability in the CT workforce.
Conflict of Interest Statement
The authors reported no conflicts of interest.
The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest.
Footnotes
Supported by the Donald J. Ferguson, MD Fund for Surgical Research at the University of Chicago.
Appendix E1
Table E1.
Comparison of outcomes related to role and value of AI by gender
| Statement | Category | Small effect, % | Medium effect, % | Large effect, % | P value |
|---|---|---|---|---|---|
| Role of AI in the next 5 y | |||||
| Screening | Male | 7 | 21.2 | 71.8 | .513 |
| Female | 4.4 | 26.5 | 69.1 | ||
| Preoperative evaluation and patient selection | Male | 19.8 | 34.9 | 45.4 | .036 |
| Female | 33.8 | 30.9 | 35.3 | ||
| Surgical management | Male | 31.1 | 34.6 | 34.3 | .126 |
| Female | 39.7 | 38.2 | 22.1 | ||
| Intraoperative decision making | Male | 59.9 | 27 | 13.1 | .054 |
| Female | 72.1 | 13.2 | 14.7 | ||
| Intraoperative technical tasks | Male | 66.6 | 24.7 | 8.7 | .289 |
| Female | 69.1 | 17.7 | 13.2 | ||
| Postoperative management | Male | 23.8 | 39.2 | 36.9 | .003 |
| Female | 39.7 | 42.7 | 17.7 | ||
| Value of AI in the next 5 y | |||||
| Image interpretation | Male | 5.5 | 16.3 | 78.1 | .445 |
| Female | 4.6 | 22.7 | 72.7 | ||
| Preoperative decision making | Male | 22.9 | 36.1 | 41 | .144 |
| Female | 32.3 | 38.7 | 29 | ||
| Intraoperative guidance/assistance | Male | 52 | 31.1 | 16.9 | .554 |
| Female | 50.9 | 26.4 | 22.6 | ||
| Postoperative management | Male | 26.2 | 40.7 | 33.1 | .079 |
| Female | 35.5 | 45.2 | 19.4 | ||
| Patient interactions | Male | 30.2 | 33.9 | 35.9 | .344 |
| Female | 26.9 | 44.2 | 28.9 | ||
Bold type indicates significant differences by χ2 analysis. AI, Artificial intelligence.
Table E2.
Comparison of outcomes related to role and value of AI by gender and surgical subspecialty
| Statement | Category | Small effect, % | Medium effect, % | Large effect, % | P value |
|---|---|---|---|---|---|
| Role of AI in the next 5 y | |||||
| Screening | Male cardiac | 7 | 21.1 | 71.9 | .691 |
| Female cardiac | 0 | 32.1 | 67.9 | ||
| Male thoracic | 5.1 | 21 | 73.9 | ||
| Female thoracic | 5.3 | 21 | 73.7 | ||
| Preoperative evaluation and patient selection | Male cardiac | 11.9 | 36.8 | 51.3 | .001 |
| Female cardiac | 25 | 42.9 | 32.1 | ||
| Male thoracic | 28.3 | 33.3 | 38.4 | ||
| Female thoracic | 36.8 | 23.7 | 39.5 | ||
| Surgical management | Male cardiac | 25.4 | 34.1 | 40.5 | .025 |
| Female cardiac | 50 | 28.6 | 21.4 | ||
| Male thoracic | 37 | 35.5 | 27.5 | ||
| Female thoracic | 31.6 | 44.7 | 23.7 | ||
| Intraoperative decision making | Male cardiac | 56.7 | 27.6 | 15.7 | .18 |
| Female cardiac | 75 | 10.7 | 14.3 | ||
| Male thoracic | 62.3 | 28.3 | 9.4 | ||
| Female thoracic | 68.4 | 15.8 | 15.8 | ||
| Intraoperative technical tasks | Male cardiac | 67 | 23.3 | 9.37 | .307 |
| Female cardiac | 75 | 17.9 | 7.1 | ||
| Male thoracic | 65.2 | 28.3 | 6.5 | ||
| Female thoracic | 63.2 | 18.4 | 18.4 | ||
| Postoperative management | Male cardiac | 18.4 | 38.4 | 43.2 | <.001 |
| Female cardiac | 21.4 | 50 | 28.6 | ||
| Male thoracic | 29.7 | 42.8 | 27.5 | ||
| Female thoracic | 50 | 39.5 | 10.5 | ||
| Value of AI in the next 5 y | |||||
| Image interpretation | Male cardiac | 5.4 | 15.7 | 78.9 | .408 |
| Female cardiac | 3.7 | 33.3 | 63 | ||
| Male thoracic | 4.4 | 18.2 | 77.4 | ||
| Female thoracic | 2.7 | 13.5 | 83.8 | ||
| Preoperative decision making | Male cardiac | 18.4 | 34.1 | 47.5 | .047 |
| Female cardiac | 38.4 | 30.8 | 30.8 | ||
| Male thoracic | 27.1 | 39.5 | 33.1 | ||
| Female thoracic | 25.7 | 45.7 | 28.6 | ||
| Intraoperative guidance/assistance | Male cardiac | 49.7 | 31.3 | 19 | .736 |
| Female cardiac | 55 | 25 | 20 | ||
| Male thoracic | 53.3 | 33.6 | 13.1 | ||
| Female thoracic | 46.9 | 28.1 | 25 | ||
| Postoperative management | Male cardiac | 21.1 | 38.3 | 40.6 | <.001 |
| Female cardiac | 19.2 | 57.7 | 23.1 | ||
| Male thoracic | 33.3 | 46.2 | 20.5 | ||
| Female thoracic | 45.7 | 37.1 | 17.2 | ||
| Patient interactions | Male cardiac | 26.8 | 36.3 | 36.9 | .0268 |
| Female cardiac | 15 | 60 | 25 | ||
| Male thoracic | 34.4 | 32.8 | 32.8 | ||
| Female thoracic | 34.4 | 34.4 | 31.2 | ||
Bold type indicates significant differences by χ2 analysis. AI, Artificial intelligence.
Table E3.
Comparison of outcomes related to role and value of AI by gender and region of practice
| Statement | Category | Small effect, % | Medium effect, % | Large effect, % | P value |
|---|---|---|---|---|---|
| Role of AI in the next 5 y | |||||
| Screening | North American male | 8.7 | 24 | 67.3 | .276 |
| North American female | 6.3 | 31.2 | 62.5 | ||
| Non–North American male | 5 | 19 | 76 | ||
| Non–North American female | 0 | 15.8 | 84.2 | ||
| Preoperative evaluation and patient selection | North American male | 23.6 | 39.9 | 36.5 | <.001 |
| North American female | 45.8 | 29.2 | 25 | ||
| Non–North American male | 15.7 | 28.9 | 55.4 | ||
| Non–North American female | 5.3 | 36.8 | 57.9 | ||
| Surgical management | North American male | 37.5 | 29.8 | 32.7 | .001 |
| North American female | 52.1 | 29.2 | 18.7 | ||
| Non–North American male | 24 | 43 | 33 | ||
| Non–North American female | 10.5 | 63.2 | 26.3 | ||
| Intraoperative decision making | North American male | 71.1 | 20.2 | 8.7 | <.001 |
| North American female | 87.5 | 4.2 | 8.3 | ||
| Non–North American male | 46.3 | 33 | 20.7 | ||
| Non–North American female | 36.9 | 36.8 | 26.3 | ||
| Intraoperative technical tasks | North American male | 75.5 | 19.2 | 5.3 | <.001 |
| North American female | 85.4 | 12.5 | 2.1 | ||
| Non–North American male | 57 | 28.9 | 14.1 | ||
| Non–North American female | 31.6 | 31.6 | 36.8 | ||
| Postoperative management | North American male | 26.4 | 40.9 | 32.78 | .016 |
| North American female | 45.8 | 41.7 | 12.5 | ||
| Non–North American male | 21.5 | 39.7 | 38.8 | ||
| Non–North American female | 26.3 | 47.4 | 26.3 | ||
| Value of AI in the next 5 y | |||||
| Image interpretation | North American male | 6.2 | 20.7 | 73.1 | .098 |
| North American female | 6.4 | 27.7 | 65.9 | ||
| Non–North American male | 5 | 10.8 | 84.2 | ||
| Non–North American female | 0 | 11.1 | 88.9 | ||
| Preoperative decision making | North American male | 32.5 | 32 | 35.5 | <.001 |
| North American female | 43.2 | 43.2 | 13.6 | ||
| Non–North American male | 8.8 | 45.6 | 45.6 | ||
| Non–North American female | 5.9 | 29.4 | 64.7 | ||
| Intraoperative guidance/assistance | North American male | 63.7 | 26.2 | 10.1 | <.001 |
| North American female | 70.6 | 20.6 | 8.8 | ||
| Non–North American male | 37.6 | 35.8 | 26.6 | ||
| Non–North American female | 16.7 | 38.9 | 44.4 | ||
| Postoperative management | North American male | 30 | 42.4 | 27.6 | .047 |
| North American female | 38.6 | 50 | 11.4 | ||
| Non–North American male | 22.8 | 38.6 | 38.6 | ||
| Non–North American female | 29.4 | 35.3 | 35.3 | ||
| Patient interactions | North American male | 34.6 | 34.6 | 30.8 | .079 |
| North American female | 31.4 | 48.6 | 20 | ||
| Non–North American male | 25 | 31.7 | 43.3 | ||
| Non–North American female | 18.8 | 31.2 | 50 | ||
Bold type indicates significant differences by χ2 analysis. AI, Artificial intelligence.
Table E4.
Hopes by gender regarding AI in CT surgery expressed as common themes
| Theme | Response frequency, % |
|
|---|---|---|
| Female surgeons | Male surgeons | |
| Administrative burden | 20.8 | 15.5 |
| Diagnostic accuracy | 20.8 | 24.7 |
| Efficiency | 22.9 | 18.0 |
| Patient care improvement | 10.4 | 13.4 |
| Decision support | 8.3 | 13.9 |
| Error reduction | 6.3 | 3.1 |
| Workflow improvement | 4.2 | 4.1 |
| Equity and access | 4.2 | 2.1 |
| Training and education | 2.1 | 2.1 |
| Technology integration | 0.0 | 3.1 |
AI, Artificial intelligence; CT, cardiothoracic.
Table E5.
Fears by gender regarding AI in CT surgery expressed as common themes
| Theme | Response frequency, % |
|
|---|---|---|
| Female surgeons | Male surgeons | |
| Clinical decision risks | 24.2 | 18.6 |
| Autonomy and control | 18.2 | 15.2 |
| Data governance | 15.2 | 16.2 |
| Systemic biases | 9.1 | 4.4 |
| Legal/liability exposure | 6.1 | 8.1 |
| Clinical dehumanization | 12.1 | 9.3 |
| Regulatory gaps | 6.1 | 5.3 |
| Workforce disruption | 3.0 | 5.0 |
| Technical implementation | 9.1 | 13.4 |
| Other | 12.1 | 15.5 |
AI, Artificial intelligence; CT, cardiothoracic.
References
- 1.Topol E.J. High-performance medicine: the convergence of human and artificial intelligence. Nat Med. 2019;25(1):44–56. doi: 10.1038/s41591-018-0300-7. [DOI] [PubMed] [Google Scholar]
- 2.Esteva A., Kuprel B., Novoa R.A., et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017;542(7639):115–118. doi: 10.1038/nature21056. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Hashimoto D.A., Rosman G., Rus D., Meireles O.R. Artificial intelligence in surgery: promises and perils. Ann Surg. 2018;268(1):70–76. doi: 10.1097/SLA.0000000000002693. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Seastedt K.P., Moukheiber D., Mahindre S.A., et al. A scoping review of artificial intelligence applications in thoracic surgery. Eur J Cardiothorac Surg. 2022;61(2):239–248. doi: 10.1093/ejcts/ezab422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Platz J.J., Bryan D.S., Ferguson M.K., Naunheim K.S. Surgeon perception of artificial intelligence in thoracic surgery: insights from an international survey. Ann Thorac Surg. 2025 doi: 10.1016/j.athoracsur.2025.08.036. [DOI] [PubMed] [Google Scholar]
- 6.Russo C., Romano L., Clemente D., Iacovone L., Gladwin T.E., Panno A. Gender differences in artificial intelligence: the role of artificial intelligence anxiety. Front Psychol. 2025;16 doi: 10.3389/fpsyg.2025.1559457. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Koning R.M. Harvard Business School; 2025. Women are avoiding AI. Will their careers suffer? Working Knowledge. [Google Scholar]
- 8.Ortmeyer K.A., Raman V., Tiko-Okoye C., Espinosa J., Cooke D.T., Erkmen C.P. Women and minorities underrepresented in academic cardiothoracic surgery: it's time for next steps. Ann Thorac Surg. 2021;112(4):1349–1355. doi: 10.1016/j.athoracsur.2020.08.057. [DOI] [PubMed] [Google Scholar]
- 9.Shemanski K.A., Ding L., Kim A.W., et al. Gender representation among leadership at national and regional cardiothoracic surgery organizational annual meetings. J Thorac Cardiovasc Surg. 2021;161(3):733–744. doi: 10.1016/j.jtcvs.2020.11.157. [DOI] [PubMed] [Google Scholar]
- 10.Haug C.J., Drazen J.M. Artificial intelligence and machine learning in clinical medicine, 2023. N Engl J Med. 2023;388(13):1201–1208. doi: 10.1056/NEJMra2302038. [DOI] [PubMed] [Google Scholar]
- 11.Orser B., Riding A. The influence of gender on the adoption of technology among SMEs. Int J Entrep Small Bus. 2018;33(4):514. doi: 10.1504/IJESB.2018.090341. [DOI] [Google Scholar]
- 12.Ruetsch C., Davis T., Liberman J.N., et al. Prescriber attitudes, experiences, and proclivities toward digital medicine and how they influence adoption of digital medicine platforms. Neuropsychiatr Dis Treat. 2021;17:3715–3726. doi: 10.2147/NDT.S318344. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Rad A.A., Vardanyan R., Athanasiou T., Maessen J., Sardari Nia P. The ethical considerations of integrating artificial intelligence into surgery: a review. Interdiscip Cardiovasc Thorac Surg. 2025;40(3) doi: 10.1093/icvts/ivae192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Leivaditis V., Beltsios E., Papatriantafyllou A., et al. Artificial intelligence in cardiac surgery: transforming outcomes and shaping the future. Clin Pract. 2025;15(1):17. doi: 10.3390/clinpract15010017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Desai P., Wang H., Davis L., Ullmann T.M., DiBrito S.R. Bias perpetuates bias: ChatGPT learns gender inequities in academic surgery promotions. J Surg Educ. 2024;81(11):1553–1557. doi: 10.1016/j.jsurg.2024.07.023. [DOI] [PubMed] [Google Scholar]
- 16.Joshi A. Big data and AI for gender equality in health: bias is a big challenge. Front Big Data. 2024;7 doi: 10.3389/fdata.2024.1436019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Savadjiev P., Chong J., Dohan A., et al. Demystification of AI-driven medical image interpretation: past, present and future. Eur Radiol. 2019;29(3):1616–1624. doi: 10.1007/s00330-018-5674-x. [DOI] [PubMed] [Google Scholar]
- 18.Cellina M., Cacioppa L.M., Cè M., et al. Artificial intelligence in lung cancer screening: the future is now. Cancers (Basel) 2023;15(17):4344. doi: 10.3390/cancers15174344. [DOI] [PMC free article] [PubMed] [Google Scholar]