**4.2 AI in medicine and its current applications**

Today AI is already being utilized in medicine, and thus far, its applications have shown promising results as demonstrated by improved patient outcomes, optimized clinical workflows, and accelerated research. To date, there are 521 AI-Enabled medical devices approved by the FDA [37], with the overwhelming majority of these products being in the field of radiology used to process images for all pathologies, excluding cancer [38]. Other AI applications currently available are used in the fields of anesthesiology, cardiology, gastroenterology, general and plastic surgery, hematology, microbiology, neurology, obstetrics and gynecology, ophthalmology, orthopedics, pathology and urology. Given the broad spectrum of applications within varying fields of medicine, one understands that AI utilization is not only based on type but

also on what its end goal is. Generally, AI technologies in medicine can be classified by the end goal they achieve; these include everything from screening and diagnosis, to triage and clinical trial management. Multiple applications are currently being utilized and under further development, including:

Computer-Aided Detection (CADe) technology is being developed to aid in marking/localizing regions that may reveal specific abnormalities. Its goal is to elevate the sensitivity of screening tests. Curemetrix, Inc. product cmAssist™, for example, has shown a substantial and statistically significant improvement in radiologists' accuracy and sensitivity for detection of breast cancers that were originally missed [39].

Computer-Aided Diagnosis (CADx) is being developed to help characterize or assess diseases, disease type, severity, stage, and progression. An example of the application of this technology is GI Genius™; an Intelligent Endoscopy Module by Medtronic, plc. That can analyze a colonoscopy in real-time and estimate the possible histology of colorectal polyps [40].

Computer-Aided Triage (CADt) aids in prioritizing time sensitive patient detection. VIZ™LVO is a software by Viz.ai, Inc. that detects large vessel occlusion strokes in brain CT scans and directly alerts the relevant specialists in a median time of 5 minutes and 45 seconds, as opposed to 1 hour which is the standard of care today, significantly shortening the time to diagnosis and treatment [41].

Computer-Aided Prognosis (CAP) can provide personalized predictions about a patient's disease progression. The EU-funded CLARIFY Project (Cancer Long Survivor Artificial Intelligence Follow-Up) is working in harnessing big data and AI to provide accurate and personalized estimates of a cancer patient's risk for complications, including rehospitalization, cancer recurrence, treatment response, treatment toxicity, and mortality [42].

Clinical Decision Support Systems (CDSS) are being employed to aid healthcare providers in the diagnoses and treatment of patients in the most effective way possible. Babylon AI, by Babylon, Inc. for example, is a system that uses data to decide on, and provide information about the likely cause of people's symptoms. It can then suggest possible next steps, including treatment options. The system has demonstrated its ability to diagnose as well as or even better than physicians [43].

Remote Patient Monitoring (RPM) systems are being used to monitor patients, and Virtual Rehabilitation is being developed to help patients recover from illnesses and injuries. Systems like CardiacSense Ltd. Medical Watch continuously monitor heart rate and blood pressure, process the data and update the physician in real time. This noninvasive monitoring system allows the physician to change treatment according to data that would not have been available otherwise [44].

Health Information Technology (HIT) is being employed to improve disease prevention and population health. Medial EarlySign, Ltd. mines data from electronic medical records for early detection of patients with high risk of colorectal cancer. Patients determined to have a high risk by the system are flagged and consequently scheduled for colonoscopy. This system has achieved early detection of an additional 7.5% of colorectal cancers that would otherwise have been caught in more advanced stages [45].

Clinical Trials Management Systems (CTMS) are being developed to help streamline all aspects of clinical trials including preclinical drug discovery, clinical study protocol optimization, trial participant management, as well as data collection and management. These types of systems enable researchers to improve study design by utilizing the guidance in choosing the best study design, determination of number of patients needed for each study arm, optimizing candidate selection, as well as tracking and analyzing large amounts of data. CTMS are helping researchers create stronger and more efficient trials [46].

As demonstrated by the above systems, the implementation of these types of AI has significantly and measurably improved the field of medicine. As the benefit of AI continues to be appreciated, via the understanding as to how it aids in providing better and more efficient care to patients, more professionals will begin to utilize it. With improved acceptance, the previously discussed adoption model that Barkun et al. [13] proposed will continue to shift towards long term implementation.

#### **4.3 Potential benefits of AI in surgery**

Improved patient care has historically been linked to technological advancements. Laparoscopic cameras have evolved from simple VHS quality to HD and 4 K cameras and even 3D vision with Near Infra-Red capabilities that allow the surgeon to see beyond the naked eye. Laparoscopic instruments evolved from simple straight and rigid instrumentation to articulating and flexible tools, providing a limitless range of motion. Standardization and precision-surgery have infiltrated the OR in the form of staplers for the creation of anastomosis, advanced energy tools for cutting and coagulation, and robotic assisted surgery that combines all of the above technologies together to enhance human precision. Most recently, AI has started to appear in the surgical field, albeit in the perioperative setting. These systems are helping surgeons with decision making processes both pre- and post-operatively by predicting complications and managing different aspects of patient variables [47]. Nevertheless, AI has yet to penetrate the walls of the OR.

The disparity between the advancement of AI in surgery and other fields in medicine is probably because most applicable AI technologies today are focused on vision and reporting, i.e. diagnosis and big data analysis. Surgery at its core is about both vision and action, which presents a much more complex challenge. This challenge, however, has not stopped research efforts in the field of Computer Assisted Surgery. A PubMed query revealed that in 2022, there were more than 5200 publications discussing AI in surgery, and according to The Growth Opportunities in Artificial Intelligence and Analytics in Surgery study, by 2024 the AI market for surgery will reach \$225.4 million [48].

Prototypes, proof of concept and pilot studies are being developed around the world, focusing mainly on improving patient safety and refining workflows in the OR [49]. There are already published reports of AI projects in Expert Systems, Computer Vision, image classification, as well as data acquisition and management that show promising results. Studies have reported success of Computer Vision systems for recognition of surgical tools, surgical phases and anatomic landmarks.

Research on videos of laparoscopic cholecystectomy, for example, has reported success of tool recognition such as graspers, hooks and dissectors; other studies have been successful in phase recognition during laparoscopic cholecystectomy. The tested systems have demonstrated the ability of understanding and reporting when the surgeon is dissecting the cystic duct, separating the gallbladder from the hepatic bed or removing it from the body. More advanced systems have demonstrated the ability to recognize and mark the critical view of safety [50, 51].

While these research efforts are certainly demonstrating promising results, the application of AI within the operating room itself remains in its infancy.

*Human-Machine Collaboration in AI-Assisted Surgery: Balancing Autonomy and Expertise DOI: http://dx.doi.org/10.5772/intechopen.111556*
