Artificial Intelligence in Surgery: Are We Ready?

Abstract

ARTICLE INFORMATION Mirnezami R, Ahmed A. Surgery 3.0, artificial intelligence and the next-generation surgeon. Br J Surg 2018;105:463-5. BACKGROUND Although artificial intelligence (AI) has been around for quite some time, it has been revolutionised since OpenAI released an early demonstration of the earliest version of Chat GPT in November 2022. The fact is that Chat GPT, being a simple chatbot, started adapting to its user at such a rapid rate that, even before its 1st birthday, we are at its 4th version.[1] “DeepMind” is a Google-funded company, which announced its AI programme known as AlphaZero. This programme, which learnt chess all by itself in just four hours, defeated the world champions of the game. In the medical field, AI-based platforms are performing similar to, if not better than, humans. Cornell university scientists reported AI to be more accurate in detecting lymph nodal metastasis compared to conventional pathologists. Either due to practical concerns or due to lack of necessity, AI in surgery is still in its infancy. However, the growth of AI can be exponential and it will be a prominent force here very soon. At present, AI is only an adjunct to surgical practice rather than a substitute. The role of AI in surgery, as the authors describe, is in making complex decisions. Choices related to the timing of surgery, its radicalness, need for multimodal therapy and personalised risks can be quite difficult for most surgeons. With the advent of complex tools such as algorithmic clinical decision support (CDS), analysing and comparing huge amounts of data has become simpler. Oncology is one such field where treatment decisions have to be individualised. Genetical analysis, pharmacological variables, complex side effects and gene-specific drug interactions require complex computing and AI is helpful in more ways than one. The newer CDS systems in surgery will have to integrate biochemical, radiological and sequencing data. With this information, they would predict variables such as duration of stay, personalised risk and cancer-specific outcomes. In future, AI may generate accurate scoring systems which guide clinical decision-making. For atypical cases, CDS systems will integrate the data from various international databases and provide a palatable outcome for experimental procedures. The authors remark that AI might well become a valued member of the surgical multi-disciplinary team. The full potential of future of surgical robot is beyond eliminating tremors and zooming in on the area of interest. It could pave the way for digital assistants similar to Siri and Alexa, from verbal responses to superimposing preoperative three-dimensional (3D) images on the table. Autonomous systems such as the Smart Tissue Autonomous Robot have been consistent in showing superior results in bowel anastomosis, in comparison to expert surgeons. The advent of semi-autonomous robots has posed specific problems in the ways of conflicting operative judgements. The above favourable developments are not without considerable risks. The future patient may be under the joint custodianship of both robot and surgeon which will lead to issues in ethics and accountability. Furthermore, the onus will be on the regulatory authorities to scrutinise the robots. With the lack of clear-cut definitions for robotic autonomy, this will be difficult to classify and define. Another concern is regarding competent regulatory bodies such as the United States Food and Drug Administration with regard to autonomous robots. COMMENTARY Many different technological companies have used AI for different applications. Ranging from chatbots that help in writing research papers to self-driving cars to mobile phones and now, helping doctors in treating patients. The term ‘Medical technologies’ has evolved from prosthetics and ultrasound machines to AI-powered tools. AI finds its place in a lot of critical decision-making in the field of medicine. AliveCor released their first mobile-based application known as ‘Kardia’ to detect atrial fibrillation based on the electrocardiogram in 2014. Apple also acquired the approval for its ‘Apple watch 4’ to detect atrial fibrillation in a person wearing its smart wristwatch. Similarly, seizure detection devices, glucose monitoring devices and endoscopic image detectors have been aiding physicians. Paige. AI is an AI-based algorithm that saves time for pathologists. Using computational histopathology, it helps them to focus on the important parts of the slides.[2] The field of surgery, which is pretty much a ‘hands-on’ branch, has also seen the influence of AI. Pre-operative 3D road maps can be constructed by integrating AI into computed tomography (CT) and magnetic resonance imaging. Fancy wearable devices have helped in detection and reducing post-operative complications and early recovery of patients. At present, AI is being used for risk stratification, genomics, imaging, diagnosis and drug discovery. Chilakmurthy et al. have demonstrated that deep learning can recognise intracranial haemorrhage, calvarial fracture, midline shift and mass effect from head CT scans. In minimally invasive surgeries, an intra-operative guide can be provided by applications such as tissue tracking and augmented reality. Endoscopic manoeuvring has never been this simple as AI learning-based aids in depth estimation, visual odometry and environmental mapping.[3-5] Although enhancing precision, robustness and increased ergonomics, AI in surgery comes with added ethical dilemmas and legal considerations. Patients’ privacy, data safety and vulnerability to cybercrime are the things to be watchful about. Zhou et al. raised some important issues in their article on medical robotics. They questioned the fate of medical and surgical trainees with respect to clinical diagnosis and dexterity when it comes to basic surgical skills. A day may come when a surgical robot with machine-based learning may outperform a surgeon.[3,6,7] AI as a medical colleague or an administrator may determine the strength and sustainability of a surgical team to perform an operation. By analysing the real-time data related to patient outcomes, AI is capable of generating a profile of an individual surgeon’s complications for individual procedures. In simple words, AI can suggest who is the best person to operate for a particular patient. Deep learning and interaction have helped systems such as ‘AlphaZero’ to be experts in a new operative technique overnight, which could take surgeons years to master. A potential problem arises with the use of semi-autonomous robots, in which the ‘Three Rules of Robotics’ by Issac Asimov, i.e. A robot may not injure a human being or, through inaction, allow a human being to come to harm A robot must obey orders given it by human beings except where such orders would conflict with the First Law A robot must protect its own existence as long as such protection does not conflict with the First or Second Law. May interfere with the judgement of the operating surgeon. Ethical and legal implications will arise in future, with respect to increasing autonomy of a machine. In spite of all insecurities, AI should be seen as an opportunity to complement rather than to replace the surgeon. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.

Ähnliche Arbeiten