Authors: Kyle Lamb, Michael D. Abramoff, Jose M. Balibrea, Stephen M. Bishop, Richard R. Brady, Rachel A. Kallkat, Manish Chand, Justin W. Collins, Marcus K. Diener, Matthias Eisenmann, Kelly Fermont, Manuel Galvao Neto, Gregory D. Hager, Robert J. Hinchliffe, Alan Horgan, Pierre Jeannin, Alexander Langerman, Kartik Logishetti, Amit Mahadik, Lena Mayer-Hein, Esteban Martin Anton, Pietro Mascagni, Ryan K. Matthew, Beat P. Muller-Stich, Thomas Neumuth, Felix Nickel, Adrian Park, Gianluca Pellino, Frank Rudzich, Sam Shah, Mark Slack, Miles J. Smith, Naim Soomro, Stephanie Speidel, Danil Stoyanov, Henry S. Tilney, Martin Wagner, Ara Darzi, James M. Kinross, and Sanjay Purkaista.
![](https://mvsystem.com/wp-content/themes/mvs/assets/img/blog/blog-detail-banner.png)
21.05.2024
Consensus agreement on the Delphi Method for Digital Surgery
The use of digital technologies is growing rapidly in all surgical specialties, but a consensus on the term "digital surgery" has not yet been formed. This is extremely important because digital healthcare technologies create technical, managerial and legal challenges that are unique to the surgeon and the patient. We strive to define the term "digital surgery" and pay attention to ethical issues related to its clinical application, as well as identify barriers and research goals for future practice. 38 international experts in the field of surgery, artificial intelligence, industry, law, ethics and politics participated in the survey using the Delphi method, consisting of 4 rounds. The questions were formulated by expert and public groups using a questionnaire on key topics identified in the literature during the two subsequent rounds of the survey. Consensus was determined if more than 70% of experts considered the statement important, and less than 30% considered it unimportant. A final online meeting was held to discuss consensus statements. There is 100% agreement on the definition of digital surgery: the use of technology to improve preoperative planning, surgical performance, therapeutic support or training in order to improve outcomes and reduce harm. We cover key ethical issues related to data, privacy, confidentiality and public trust, consent, law, litigation and liability, as well as commercial partnerships in the field of digital surgery. We also identify barriers and research goals for future practice. Developers and users of digital surgery should be aware not only of the ethical issues associated with digital healthcare applications, but also of the ethical considerations inherent in digital surgery. Future research on these issues should involve all stakeholders in digital surgery, including patients.
Introduction
Digital technologies ranging from robotics (1), virtual and augmented reality (2), and artificial intelligence (AI) (3) offer high-precision data-based surgery in order to improve patient outcomes, surgical efficiency, on the one hand, as well as the productivity and efficiency of surgeons and their teams, on the other hand (4,5). The introduction of digital technologies is not limited only to the operating room. Currently, digital technologies play an important role in such areas as preoperative planning (6), prediction of surgical risk (7), as well as evaluation of the effectiveness of surgical intervention (8,9).
In many ways, the rapid adoption of these technologies is driven by commercial opportunities and the promise of better results for surgeons and patients. The robotic surgery market, which accounts for only a fraction of the global digital surgical technology market, is estimated at US$5 billion with an estimated value of US$16.77 billion by 2031 (10). Thus, surgeons and healthcare providers are gaining more and more freedom in choosing technologies to use in their operating rooms.
Despite the growing role of digital technologies in surgery, the definition of the term “digital surgery” remains unclear, while it is of great importance for a number of reasons. Firstly, surgery is a high-risk clinical intervention, and failures in these technologies can cause serious harm. Surgery differs from other clinical disciplines because it depends on analyzing heterogeneous data in real time, and patients and surgeons deserve to standardize new technologies to reduce risks. Secondly, digital surgery cannot be effectively tested or understood unless there is a clear definition. Without this, it is impossible to ensure the quality of clinical interventions or trials. As digital surgery is rapidly being introduced into clinical practice, it is also important that we can clearly and consistently explain its role to patients, especially in the context of data collection and processing for digital surgery applications. Finally, the lack of clarity in digital surgery is hindering progress. Rapidly developing areas, which include digital surgery, require clarification of research priorities and areas for collaboration.
Due to the increasing use of digital technologies, surgery is not unique among medical specialties. However, it will include not only the use of digital technology, but also a breakthrough in the culture and practice of surgery as a specialty that has historically focused on postoperative outcomes with little emphasis on data collection in the operating room. As a result, the potential benefits and risks of digital technology adoption are unique to surgery, which requires digital surgery to be clearly defined.
The use of digital technologies in surgery can create risks that are not brought to the attention of patients as part of the current practice of obtaining consent. In addition to the identified risks associated with the introduction of new technologies into clinical practice, digital technologies often depend on large-scale processing of personal data, which creates certain ethical and data management problems.
A survey on the use of AI in healthcare conducted in 2018 showed very or extremely important factors contributing to the development of AI (11) — the creation of an ethical framework to strengthen/preserve trust and transparency (88% of respondents) and clarity regarding data ownership rights (82% of respondents). Therefore, to be successful, digital applications must not only be accurate, but also be based on ethical principles (12).
Lessons can be learned from the ethics of AI in other sectors. Global key ethical principles in the field of AI are reduced to 5 main ones: transparency, fairness and impartiality, non-harm, responsibility, autonomy (13,14).
These principles can serve as a general guide for those who develop and use digital instruments in surgery. But there is a lack of guidance to cover specific ethical and data management issues related to surgical practice. The Artificial Intelligence Laboratory of the National Health Service of the United Kingdom has published a management and data collection system for the safe implementation of AI systems in healthcare (15). In addition, the World Health Organization recently announced new guidelines on AI ethics in healthcare (16). This creates a common basis for the development of AI in healthcare, but does not address specific issues of AI ethics in surgery. Firstly, the process of making surgical decisions is unique and requires promptness, which depends on the specific situation occurring “here and now”, while the patient often cannot be consulted. Secondly, databases on surgery are extensive and heterogeneous, including surgical videos, sensor data and data on the work of the surgical team (17, 18). Data management issues in surgery affect not only the patient, but also the surgeon and the surgical team as a whole. The recent publication of ethical standards for the use of AI in teaching robotic surgery (19) signals the specifics of the nature of surgical practice and the need to study ethical issues within the framework of digital surgery. In addition, the ownership of surgical data is controversial and does not fit into the existing legal framework.
There have been few works in the existing literature dealing with ethical principles and data management issues related to digital technologies in surgery. In areas that cover several areas of knowledge and where there is insufficient information, consensus methods such as the Delphi method (20) are effectively used. Therefore, we conducted a study using this method. Firstly, to identify key ethical principles and formulate data management issues in digital surgery, and secondly, to compare these opinions among key stakeholders in digital surgery to reach consensus. The objectives of the study are, firstly, to agree on a definition of the term “digital surgery”, which can be used in both clinical and academic environments, and secondly, to identify key ethical principles and formulate data management issues related to digital surgery. Next, identify key barriers and research goals for the future of digital surgery.
Results
52 experts completed the first round, 44 participants (84.6%) completed the second round and 38 participants (86.4%) completed the third round. 20 members of the public also participated in the 1st round, and the issues raised in this round were combined with expert questions in the 2nd round. Cronbach’s alpha (a coefficient showing internal consistency of characteristics describing a single object) was 0.981 and 0.881 in the 2nd and 3rd rounds, respectively, indicating high reliability between evaluators. The full text of the questionnaire and the results of rounds 2 and 3 can be found in Additional Notes 2 and 3. Consensus was reached on 114 questions, which were grouped into 7 key topics:
- Definition of Digital Surgery
- Data
- Privacy
- Confidentiality and public trust
- Approval
- Law
- Litigation and liability.
- Commercial partnerships
Consensus was reached on 38 issues related to the development, implementation and monitoring of digital surgical systems, and on 22 technical, clinical and organizational goals for future research in the field of digital surgery. A list of all the issues on which consensus has been reached can be found in Additional Note 4.
Definition of Digital Surgery
Digital surgery is the use of technology to improve preoperative planning, surgical performance, therapeutic support, or training to optimize outcomes and reduce harm..
71% of participants agreed that there is currently no clear definition of digital surgery. 86% of the participants agreed that digital surgery should include the preoperative period, surgical intervention and postoperative period. 82% of the participants agreed that the term “digital surgery” should include not only the operational aspects of surgery, but also others, including training, diagnosis and research.
Participants were asked to comment on the definition of the term “digital surgery”. This definition was discussed at the final online meeting, where 100% of the panelists agreed with the final agreed wording. This definition provides a practical formulation that can be used for both clinical and research purposes. It can also be used by those who have limited knowledge in this field. The participants of the discussion were also asked to coordinate the technologies included in digital surgery (Fig. 1). Finally, the panelists agreed on the existing and potential benefits of digital surgery (Table 1). The consensus elements were grouped into three themes: data, analysis and applications.
Fig. 1: Components of digital surgery identified by the Delphi method.
Advantages of digital surgery
detected by Delphi panel
Patient
Improvement of clinical results
Improving patient care
Improving diagnostics
Individual treatment of the patient
More rapid detection of deterioration of the patient's condition
Surgeon
The possibility of planning the preoperative period
Decision support by the surgeon
Reducing the cognitive load on the surgeon
Automation of surgical processes
Error prediction
Error detection
Standardization of surgical processes
Improving the ergonomics and health of the surgeon
Evaluation of the surgeon's work
Accelerated surgical education
Organization
Improving the effectiveness of surgery
Improving economic efficiency
Quantification of outcomes beyond survival and other standard criteria for evaluating outcomes
Understanding the advantages and limitations of surgical strategies
Understanding and improving team dynamics
Table 1. Advantages of digital surgery identified by the Delphi panel.
Data collection issues
Data access
Data plays a central role in both the development and use of digital surgical technologies. The panelists agreed that there is currently no infrastructure for data collection, and that a significant factor in this is the lack of interoperability (previously defined as the ability of two or more systems or components to exchange information and use information exchanged (21) between different devices and systems. Thus, the panelists agreed that data is not always available in digital format. In addition, they concluded that there is a shortage of reliable data sets, and this is complicated by the lack of data quality standards, annotations, and formatting. Finally, the panelists agreed that determining proper access to data is an important issue and that data management processes are currently overly complex.
Data storage and security
For digital surgery to be successful, data must be stored properly and securely. The panelists agreed that hospitals currently lack the technical capabilities and structure to properly store data. They also stressed that the cost of data storage and proper data encryption are important issues. In addition, the panelists agreed that institutions are not equipped and do not have sufficient resources to ensure proper cybersecurity. Finally, the consequences of data leaks are currently poorly defined.
Data exchange
Data exchange includes data exchange between different technologies, between hospitals, as well as between hospitals and commercial partners. The panelists agreed that there are currently no guidelines on data ownership rights, and international legal requirements regarding data exchange are unclear. In addition, the panelists agreed that compliance with the existing rules for working with data can hinder competitiveness. Also, data exchange across international borders is problematic, and there is no consensus on data exchange formats. Finally, the panelists agreed that surgeons have no motive to share data.
Privacy, confidentiality and public trust
Maintaining privacy and confidentiality is essential not only to protect patient autonomy, but also to ensure patient trust, which is vital for the future development of digital surgical applications. The panelists agreed that proper anonymity of data is of great importance. In addition, there was a consensus that agreements with patients on data exchange should be defined and that data should be used explicitly for the purposes for which it was collected. The panelists agreed that surgical teams are not sufficiently aware of the significance of the data they collect. This suggests that there is a need for more in-depth training in data management and security.
The widespread misuse of personal data in both clinical and non-clinical settings has led to public concerns about the collection of personal data, especially in collaboration with a commercial organization (22). The panelists agreed that building public trust in data sharing is an important issue. To date, there has been insufficient interaction with the public, among which there is a general lack of education in the field of AI. The participants of the discussion agreed that the following important factors affect public trust:
- lack of awareness due to the opacity of surgical AI systems;
- the fear that AI will increase bias in data sets;
- the inability to create an effective surgical artificial intelligence system to date.
Regardless of success or failure, the panelists agreed that there should be mandatory reporting of results. Privacy and confidentiality issues apply not only to the patient, but also to the surgical team. The panelists agreed that an important issue is the surgeon’s right to privacy and the potential impact of digital surgical systems on his behavior.
Approval
Large amounts of surgical data are vital for the development and evaluation of digital surgical technologies. To ensure the confidentiality of the patient’s personal data, it is necessary to obtain appropriate consent from him. In accordance with the General Data Protection Regulation of the United Kingdom and the EU (GDPR), consent must be specific and informed and, therefore, include the purposes of processing and the right to revoke at any time. With regard to the digital surgical context, the panelists agreed that when collecting data for unknown future applications, there may be problems with consent procedures. In addition, an important issue is the management of a patient who has decided to withdraw his consent. The “right to delete”, although with reservations, is problematic in the context of digital surgery. For example, for AI systems that are pre-trained on a patient’s data set who has decided to withdraw their consent.
It is unclear to what extent patients should be informed of consent to the transfer of their data. The panelists agreed on the importance of patients understanding what they are being asked to do when they agree to data sharing. Teaching patients to share data for digital surgical applications should be a priority. The panelists agreed that issues related to the various requirements for agreement between countries were important. Finally, the panelists agreed that it was necessary to consider the right of the surgeon and the surgical team to opt out of data collection.
Thus, the participants in the discussion came to a consensus that there should be a standardized methodology that allows patients to share their data for use in digital surgery. Procedures for obtaining consent for digital surgery must:
- determine the amount of data collection;
- determine who will access the data;
- explain why the data will be collected;
- allow data collection for future/unknown applications;
- patients must give separate consent if commercial partners have access to their data.
Law
The key legislation in Europe regulating the use of data in digital surgery is the General Data Protection Regulation The UK and the EU. This fundamental piece of legislation regulates health data (as well as other data) regardless of the format or method of their collection. GDPR is technology neutral without mentioning artificial intelligence or related technologies. However, considerable attention is paid to the large-scale processing of personal data.
U.S. law is more complicated. Most of the relevant legislation is governed by the Privacy Rule within the framework of the Law on the Transfer and Protection of Data from Healthcare Institutions (Health Insurance Portability and Accountability Act, HIPAA). Unlike GDPR, HIPAA is more limited and concerns only Protected Health Information (PHI), which is identifiable. Thus, depersonalized data is not subject to HIPAA. In addition, data ownership under HIPAA is a problem that has yet to be resolved.
The participants of the discussion agreed that an important problem is the lack of standardization of terminology in legislation related to AI, and the lack of special regulations related to digital clinical data. An important issue that the panelists agreed on was that the current model of ownership of both data and intellectual property is unclear according to the law. In addition, there is a lack of clarity regarding the legal basis for data collection and exchange. It is currently unclear who is responsible for data integrity in accordance with the law. Other issues agreed upon by the panelists included differences in data legislation in different countries and unclear rules regarding international data transfer. Finally, the panelists agreed that all stakeholders in the field of digital surgery are not sufficiently knowledgeable in the field of data law, and medical institutions do not have enough knowledge in the field of data law.
Litigation and liability
Although digital surgical systems promise benefits for patients, surgeons, and institutions (Table 1), panelists agreed that there is a lack of regulation regarding litigation and liability, both in the event of failures in digital surgical systems and for surgeons who choose not to follow such systems, as AI-based decision support tools. In addition, if the surgeon follows AI decision support and it leads to a negative outcome, it is unclear how responsibility will be considered. However, it should be noted that a recent guideline published by the American Medical Association regarding AI in healthcare states that autonomous AI creators must take responsibility (23). Other important issues that the panelists agreed on included the potential impact of digital surgical systems on medical liability compensation and the use of enhanced surgical data collection for the purpose of determining medical negligence.
Commercial partnership
The future success of digital surgery is likely to depend on the development of commercial partnerships that can offer medical institutions resources and relevant expertise. The panelists agreed that the business and data exchange models between hospitals and commercial companies were not clearly defined. The panelists agreed that in most institutions there is no basis or experience for establishing fair partnerships between medical and commercial organizations. They drew attention to the problems related to the inequality of power and the different motives between hospitals and commercial companies. Finally, the panelists agreed that commercial partnerships could limit the ability of hospitals to report results.
Barriers to Digital surgery
The panelists were asked to identify the main obstacles to the development of digital surgery in three areas: development, deployment and monitoring. The ten consensus barriers that were identified as the most important in previous rounds were ranked during the final online consensus meeting (Table 2).
Barriers to digital surgery
identified and ranked from highest to lowest in order of importance by the Delphi method
Development
Lack of digitalization in hospitals
Outdated hospital IT systems do not meet their intended purpose
Insufficient data availability
Lack of a common ontology for annotations
Lack of a data registry and platform standards
Lack of standards in data formatting methods
Lack of data quality standards
Insufficient experience in the field of surgical AI
Poor interoperability between AI systems and embedded technologies in the operating system
Difficulties in data exchange between several centers
Implementation
Infrastructure costs
The difficulty of the process due to bureaucratic difficulties
Difficulties in establishing contractual relations
Compensation or business model is not clearly defined
Institutional reluctance to share patient data
Failure to demonstrate safety or clinical benefit to stakeholders
The difficulties of integrating AI systems with existing IT infrastructure
Differences in hospital IT systems
Regulatory requirements are currently unclear
Lack of mechanisms for obtaining consent and obtaining data
Monitoring
Clarity on responsibility for data monitoring
Lack of resources and staff dedicated to the task
Costs associated with monitoring
Lack of standardized criteria for evaluating outcomes for monitoring
Difficulties in quantifying improvements
Currently, monitoring is not given due attention
Separation between those who are engaged in monitoring and development of surgical artificial intelligence systems
Table 2. Barriers on the way to digital surgery
Future research goals
The panelists were asked to identify technical, clinical and organizational research goals for future practice. They were subsequently ranked in order of importance at the final consensus meeting (Table 3).
Future goals of digital surgery research
identified and ranked from highest to lowest in order of importance by the Delphi method
Technical
Standardization of surgical data processing and analysis platforms for data exchange and annotation
General ontology for data annotation
Improving the explainability of AI algorithms
Working with unlabeled or poorly marked data
Identifying inequalities in basic datasets
Effective data collection systems
Implementation of a single communication standard for surgical data
Open source dataset generation
Functional compatibility between different devices and systems
Clinical
Determining the most appropriate use cases/applications for surgical AI
Development of key results, reporting and measurement sets relevant to AI in surgery
Developing a framework for the implementation and evaluation of AI in surgery
Definition of the research methodology for evaluating surgical AI
Standardization of processes
Encouraging surgeons to share data
Organizational
Demonstration of the effectiveness of surgical artificial intelligence systems
Increasing public trust and education in the field of AI
Regulatory framework for the implementation and monitoring of surgical systems with artificial intelligence
Stimulating interdisciplinary education
Organization of a working group with the participation of all stakeholders to identify best practices in the field of surgical AI
Determining the impact of surgical AI systems on litigation and liability
Development of a typical business plan taking into account the interests of the industry
Table 3. Future goals of digital surgery research
Public reaction
A total of 20 members of the public responded to the questionnaire, which was adapted for a non-expert audience. The issues raised during this round were submitted for discussion in the 2nd round, together with questions prepared by experts, for voting by the expert group to ensure that the views of the public are properly taken into account. The public group consisted of representatives of various age groups, educational levels, and self-proclaimed people familiar with AI. Although the panelists were aware of digital surgery technologies and the application of surgery to AI, such as robotics, visualization and decision support, there were also common misconceptions about the use of AI replacing human interaction and the degree of autonomy in surgical robotics. The public has recognized the potential benefits of digital surgical technologies for patients as well as for surgical teams using surgical AI as an aid to surgeons.
A community group supported data sharing for surgical AI purposes. Nevertheless, common topics related to data sharing were identified among the participants of the public group, including effective cybersecurity, proper anonymization and understanding who will have access to their data. With regard to transparency and public trust, a common theme for the participants in the public discussion was the need to gain additional knowledge about AI in surgery. Panelists said that “the public has a poor understanding of AI” and that giving the public access to AI surgical applications would “build trust.” The Commission stated that there should be transparency in the presence of adverse outcomes, and that non-disclosure of information would affect public trust and perception.
Finally, the panelists expressed opposing views on the partnership between hospitals and commercial companies. While some members of the public understood the value and resources that such partnerships could bring, others were more skeptical, fearing that companies would sell or profit from their data, as well as because of poor historical records on user data protection.
Discussion
This study is the first in the published literature to define the term “digital surgery”. Despite the fact that digital technologies are widespread in healthcare, the exact meaning of the term “digital surgery” is unclear. We present a practical, consistent definition that can be used by clinicians and scientists, as well as other stakeholders in the field of digital surgery, including patients and policy makers. This is based on well-established definitions of terms such as surgical data science, which “aims to improve the quality of interventional medicine and its value through data collection, organization, analysis and modeling” (4,5). Although the Delphi method demonstrates the potential benefits that digital surgery can bring, we highlight the ethical and data management challenges that developers and users of digital surgical technologies will have to face. In order for digital solutions to be successful in the operating room, the identified ethical and data management issues should not be secondary. They should be a priority for those who develop and use digital surgery applications at all stages.
Despite the fact that many of the problems of digital surgery identified by the Delphi method have parallels with digital healthcare, it is necessary to highlight those that are unique to digital surgery. First, the term “digital surgery” is widely used within the specialty, although 71% of the participants in this analysis agreed that the definition of digital surgery is unclear. In order for these technologies to be safely transferred into clinical practice and applied in research, standardization of this terminology is necessary.
Secondly, surgery is a high-risk clinical environment where the consequences of disruptions in digital technologies can cause significant and immediate harm on a time scale that is not comparable to other areas of clinical practice. Thirdly, the high quality of surgical results depends on the effectiveness and behavior of a multidisciplinary team, and therefore the scaling of digital technologies will require broad cultural progress. Fourth, even routine recording and analysis of operating room video data creates unprecedented ethical obstacles unique to procedural specialties. These problems must be resolved urgently before scaling up these techniques in the operating room. This goes beyond patient privacy alone (usually the primary privacy consideration in digital healthcare applications) and touches on the privacy of surgeons and their teams, who may be subject to scrutiny for their every action. In this regard, the threat of litigation may be a more serious obstacle to the development and implementation of digital surgical instruments than in other areas of healthcare. For example, surgeons may be reluctant to allow their data to be used to develop algorithms, fearing that the same video could be used against them for judicial purposes. Finally, there is a unique set of barriers to access to potentially large and diverse sets of surgical data that lack standardization, ontology, or quality assurance. Many operating rooms remain analog, and many surgical departments lack the technical infrastructure to collect the digital information available to them, or they may simply not do so. Therefore, we identify 3 key areas of digital surgery in the future.
Firstly, digital surgery is already here, but hospitals and healthcare systems are not ready for it. Significant investments in infrastructure are needed for the success of digital surgery. The pioneers who introduced digital surgery had to face a double challenge: bureaucracy and cost. In order for digital surgery to become widespread, efforts must be made to optimize this process. Template data exchange agreements and commercial models designed specifically for digital surgery applications can serve as a starting point for hospitals involved in complex and lengthy negotiations. However, appropriate commercial and legal expertise must be provided in order to receive individual advice. The United Kingdom has established a National Centre of Expertise to oversee and provide guidance to hospitals involved in these partnerships. Success can also be achieved in establishing a national health research data center in surgery, similar to existing national health research data centers in areas such as pain, mental health, and cancer treatment (24).
The panelists noted that interoperability is a key issue in the implementation of digital surgery in healthcare. Surgical data standards should be defined. Steps in this direction have already been taken with the recent publication of data annotation standards (25). Interoperability issues go beyond digital surgery and pose challenges to the wider application of digital technologies in healthcare. Current problems with data exchange between devices or hospitals will be further complicated by future applications requiring global data aggregation. International data exchange processes will have to deal with interoperability issues amid changing privacy requirements, and future efforts should focus on standardizing and streamlining this process.
Modern operating rooms have the potential to generate vast and heterogeneous datasets, but most hospitals currently cannot benefit from this. They lack the technical means to store data, as well as the networks and cybersecurity tools, including funding, needed to keep up with the development of technology. Digital surgery also has to contend with broader digital health challenges, such as varying levels of digitalization in public and private hospitals or national healthcare providers, combined with the challenges of heterogeneous hospital IT systems and electronic medical records, which pose significant obstacles to the large-scale adoption of secure digital surgery technologies.
Finally, although they are not specific to digital surgery, cybersecurity issues should not be overlooked. Digital surgical systems operate in high-risk clinical environments, and cybersecurity breaches affecting them can cause significantly greater harm to patients compared to other medical specialties. Despite the fact that no harm to patients was reported as a result of the WannaCry malware attack in 2017, she highlighted the vulnerability of hospitals to cybersecurity threats (26,27). Cybersecurity measures should not only be robust enough to protect these systems, but also take into account the “worst-case but possible” scenarios.
Secondly, the participation of the public and patients is vital for the development and implementation of digital surgery. Our community group has shown that patients support digital surgery and are willing to donate data. Concerns arose mainly due to a lack of awareness about what digital surgery entails and how patient data will be used. This lack of awareness about digital surgery can lead to a poor understanding of the benefits available to digital surgery compared to current surgical practice.
Transparency of the digital approach to patient care and public trust have consistently been highlighted as key issues for both the public and experts, as well as in the guidelines for the implementation of AI in other areas. The public is a key stakeholder in digital surgery. Her participation at all stages of development and implementation is vital for building trust. We must not forget that patients are in the center of digital surgery. The public acceptance of digital surgery applications, as well as the collection and sharing of data that they may require, should not be overlooked.
However, our community group has shown that the levels of understanding of digital surgery and artificial intelligence vary significantly. Patients may not fully understand the scope of data collection, how it might affect them, or what digital surgery entails. Digital surgery patient education can build on existing initiatives such as the Wellcome Trust’s Understanding Patient Data Program (28). They allow patients to be informed about what data is being collected and how it is being used in digital surgery applications. It is only through education and interaction with the public that they can provide properly informed consent as to whether they want to share their data.
Thirdly, training is necessary not only for patients, but also for all stakeholders. Although an important research goal identified by our group was the need to identify the most appropriate applications for surgical AI, this will only be achieved if there is sufficient interdisciplinary education. Developers should have an understanding of the surgical problems that can be solved with the help of digital technologies. Similarly, clinicians need to understand the fundamentals of the technologies they use if they want to protect the interests of their patients.
In addition, the Delphi method has revealed broader ethical, managerial, and legal issues related to digital surgery. The panelists noted that there is a poor understanding of legal issues, as well as a lack of legal knowledge in hospitals. Efforts should be made to educate stakeholders and seek expertise on these issues, as well as to ensure that they are aware of the changing legal and regulatory environment, which may go beyond data privacy legislation. And also include issues of competition law and intellectual property rights, including commercial issues such as liability, damages and data ownership. The future digital surgeon will not only be a surgeon. They should have an understanding of artificial intelligence and technology, as well as awareness of legal, ethical and data management issues related to their use.
While the Delphi method is successfully used in the literature to provide a consensus opinion (29,30), it has limitations. Firstly, the conclusions drawn from the Delphi method are the subjective opinion of one group. To mitigate this effect, efforts have been made to reduce systematic error and ensure the representativeness of the findings by involving a large number of experts with a national or international profile from a number of key areas of digital surgery. Moreover, in areas such as digital surgery, where existing knowledge is limited and there is a need for knowledge from several different fields, the Delphi method has proven to be a highly effective methodology (31).
Secondly, the reliability of Delphi’s methodology has been criticized due to the lack of methodological standardization (20). We have sought to increase the reliability of this study based on the existing methodology used in the literature. We also addressed this limitation through extensive discussion at the final meeting to ensure that the conclusions reached are sound and appropriate.
Finally, it can be argued that our final definition of “digital surgery” may lack specifics, for example, due to the fact that we do not expand terms such as “technology”. This issue was widely discussed by the Delphic Commission, and the final definition was agreed upon for several reasons. First, it is unclear how to prioritize technologies that should or should not be included in the definition, and erroneous conclusions may be drawn from technologies that have been omitted. Secondly, listing all the technologies that should be included would significantly expand the definition, which would limit its practical use. Finally, there will be no perspective in the definition. Due to the strict definition of technologies included in digital surgery, technologies that are not currently being developed are excluded. Thus, we believe that this first agreed definition of “digital surgery” fulfills the purpose of creating a usable definition and can serve as a platform for future iterations.
In conclusion, this Delphi method defines digital surgery as the use of technology to improve preoperative planning, surgical performance, therapeutic support, or training in order to improve outcomes and reduce harm. Data generation technologies carry both opportunities and risks. This Delphi method has identified key ethical issues, barriers, and research goals that will serve as the basis for future research in this area. Issues related to data, privacy, confidentiality and public trust, consent, law, litigation and liability, and commercial partnerships should be addressed at all stages by those who develop and use digital surgery. Future research on the issues identified above should involve all stakeholders in digital surgery and, therefore, work in partnership with patients.
Methodology
The protocol of this Delphi Consensus Study was published earlier (32). The structure of the Delphi method consisted of four rounds (Fig. 2). Round 1 consisted of an initial round of scoping, during which the panelists were asked to formulate questions on topics identified in the literature. In the 2nd and 3rd rounds, the experts voted on the issues that arose in the 1st round, in terms of their importance or agreement. Round 4 consisted of a final online meeting of the panelists, where, firstly, non-consensual statements were voted on, and secondly, consensus statements from previous rounds were discussed.
Fig. 2. The structure of the Delphi method in rounds
Expert Council
Experts in the fields of surgery, artificial intelligence, industry, law, ethics and politics were invited to participate. All invited participants had a national or international profile in their respected fields and/or were authors of highly effective research in this field. Initially, 122 participants were contacted by email to express their interest in participating in the study. Of the 38 participants who completed all rounds, 24 were surgeons interested in digital technologies, 8 were scientists with expertise in artificial intelligence and its applications in surgery, 3 were from the healthcare industry, and the rest of the participants were involved in the fields of health policy, digital law and ethics. 18 participants were from the UK, 13 participants from the rest of Europe, 6 participants from North America and 1 participant from South America. The median (range) of the Hirsch index for participants was 26 (5-76), and the median (range) of participants was 15 (4-32) years of experience.
Round 1
A review of the literature on data management and ethical issues of digital surgery implementation identified key topics that formed the basis of the review round (13,33-37). In addition, participants were asked about their understanding of the term “digital surgery”, as well as how to identify key barriers and future research goals related to digital surgery (see Additional Methods for the complete questionnaire). The purpose of this initial round of evaluation was to stimulate the generation of problems on these topics.
Round 2
The issues raised by both the expert group and the non-professional participants of the 1st round were thematically analyzed using NVivo qualitative data analysis software (QSR International Pty Ltd. Version 12, 2018) in order to generate reports for the 2nd round. The statements made during the 1st round, in addition to the public responses, were presented to the expert group through the Qualtrics XM platform (Qualtrics, Provo, UT). The panelists were asked to rate the statements on a 9-point Likert scale according to importance or agreement. Consensus was determined a priori if a question was rated in the range from 7 to 9 (from important to absolutely important) by at least 70% of experts and from 1 to 3 (from completely unimportant to unimportant) by less than 30% of the group, a popular approach used in Delphi exercises (38). The panelists were also asked to recommend additional statements or changes to the statements.
Round 3
The statements on which consensus could not be reached in the 2nd round, in addition to the new statements made in the 2nd round, were presented to the participants of the 3rd round. The results of the previous round, along with summary statistics, were presented to all participants in the discussion in order to encourage convergence of views on inconsistent statements. The panelists voted on the statements in the same way as in the 2nd round.
Round 4
The panelists who completed all previous rounds of the Delphic Study were invited to the final consensus meeting held on the Microsoft Teams platform (Microsoft, Redmond, Washington) on 06/24/2021. The statements on which consensus could not be reached in the 3rd round were discussed and subsequently voted on during the meeting using the real-time polling software Mentimeter (Mentimeter, Stockholm, Sweden). During this final meeting, the panelists discussed the statements with a view to developing a consensus document. Finally, the obstacles to the development, implementation and monitoring of digital surgery, as well as future research goals on which consensus was reached in previous rounds, were ranked by the meeting participants from the most to the least important.
Ethical approval
Ethical approval for this study was granted by the local Research Ethics Committee at Imperial College London (20IC6136). All participants were provided with information about the participants in electronic form before the start of the 1st round. All participants provided electronic informed consent prior to the start of Round 1.
Summary of reporting
More detailed information about the design of the study is available in the Nature Research Reporting Summary, which is referenced in this article.
Data availability
The data sets generated in the course of this study can be obtained from the author responsible for the correspondence upon reasonable request.
Other articles
Thank you for your application!
We will contact you at the specified phone number
as soon as possible
Функция тестирования
находится в разработке
О запуске функции тестирования мы сообщим позже.
Оставьте вашу почту и вы узнаете о старте тестирования первым!