This section describes the Artificial Intelligence devices or techniques proven helpful for medical settings. They are classified into two types: traditional machine learning and deep learning techniques [9].

In a double-blind validation study, Somashekhar et al. IBM Watson for Cancer has been proven to be an effective method for treating tumour cells [15]. To identify skin cancer subgroups, Esteva et al. colleagues analysed clinical pictures.

Fig. (4). Disease focus.

In Neurology, Bouton et al. developed an artificial intelligence method to help quadriplegic individuals regain control of their movements [16]. Farina et al. examined the efficiency of a physical man or machine link, which works according to the topmost implant release rates of lumbar neurons [17].

In Cardiology, Dilsizian and Siegel discussed how an AI system could diagnose heart illness using cardiac images. The US FDA lately gave Artery permission to market their Artery Cardio DL program; based on typical cardiac MRI data, AI is used to produce automatic, editable ventricular segmentation [18]. None of that is unexpected that these three ailments appear to be linked. Because these three illnesses are the biggest killers, getting treatment as early as feasible is critical to keeping the patient's condition from worsening. Additionally, by enhancing analytic methods on imaging, genomics, Equivalence Partitioning (EP), or Electronic Medical Records (EMR), which seems to be the strength of the Artificial Intelligence system, premature evaluation may be possible. AI has been primarily used to treat various conditions and the three primary disorders. Long et al. researchers diagnosed congenital cataract illness using ocular imaging data [19]. Also, Gulshan et al. used retinal fundus pictures to detect referable diabetic retinopathy, two recent examples [20].

Blood clotting in the vessel, known as cerebral infarction, causes a heart attack 85% of the time. Only a few individuals were able to receive prompt treatment due to a lack of recognition of early stroke symptoms. Villar et al. created a device that detects movement and predicts strokes early [24]. The two types of Machine Learning algorithms — in an attempt to provide such a methodology developed, the device was equipped with a biological blurry finite state machine and Patient-controlled Analgesia (PCA) [25]. The detection method included a phase of human detection and the phase of stroke onset detection. A stroke alert is initiated when the patient's behaviour differs from the regular pattern [26]. The sufferer is assessed as soon as possible for therapy. Mannini et al. proposed a smartwatch to estimate stroke that gathers information on physiologic and pathologic gait [26]. Hidden Markov models and Support Vector Machine (SVM) would be used to extract and model the data, and the method could accurately categorize 90.5% of the participants to the correct category [27].

Neuroimaging tools, such as MRI and CT scans are useful for stroke diagnosis and illness evaluation. Certain learnings have tried to use ML approaches for neuroimaging information to assist in stroke diagnosis. In nonactive functional MRI data, SVM was used to identify and classify endophenotypes of motor dysfunction following stroke by Rehme et al. [28]. With an accuracy of 87.6%, SVM can properly diagnose stroke patients. Griffis et al. used nave Bayes classification in T1-weighted MRI to identify stroke lesions [29]. The outcome is comparable to manual lesion delineation by a human specialist. In a multimodal brain MRI, Kamnitsas et al. used 3D CNN to separate tumours [30]. They have used a fully connected conditional random matrix framework for the last compositing of CNN's delicate segmented. When Rondina et al. employed Gaussian mixture analysis to evaluate stroke structural medical images, they found that cluster arrangements significantly improved predictive characteristics over damage amount on the region [31].

CT scans from stroke patients have also been analysed using machine learning algorithms. After a stroke, a free-floating intraluminal thrombus can become a tumour that is difficult to identify from the clogging of blood vessels that deliver blood to the brain in CT imaging. Jiang et al. colleagues employed two ML algorithms to categorize these two categories using quantitative shape analysis. The method's accuracy varies between 65.2% and 76.4% [32].

Diagnostic radiology implements a variety of imaging modalities to diagnose illnesses, the most popular of which are X-ray radiation, computerized tomography, MRI, and positron emission tomography. Radiologists employ a collection of images to search for diseases, diagnose them, pinpoint the origin of sickness and analyse the person's improvement over time in these operations [34].

Regarding detecting a variety of skin lesions, visual inspection is crucial. Typical skin melanoma, for example, includes visual characteristics that distinguish it from benign moles [35]. The most well-known ABCDE rule was created by a dermatologist as a rule of thumb for detecting skin carcinoma through observation. Criterion A refers to the tumour’s geometric irregularity, Criterion B to irregular borders, Criterion C to pigment variegation, Criterion D to a dimension of 6mm or higher and Criterion E to the lesion's surface extension or developing tumour [36].

Retina cinematography can be defined as a technique that does not involve the introduction of instruments in the body that uses retinal cameras to take pictures of the retina, optic disc, and macula. It can identify and evaluate diabetic retinopathy, glaucoma, retina neoplasms, and age-related vision problems and determine the reasons for avoidable blindness. The American Diabetes Association's clinical guidelines recommend screening yearly for diabetes patients with mild or no damage to the retina and more regular examinations for people with progressive retinopathy [37]. Ophthalmologists typically inspect and interpret fundus (part of the eyeball opposite to the pupil images), which is difficult to scale to the millions of diabetes individuals at risk of sight-threatening [38].

After turning a biopsy or surgery sample into defining as the extraction and dyeing the slides with dye, professional pathologists analysed the slides underneath the microscope based on the visual evaluation. However, there have been observed differences among pathologists, and the approach is not easily scalable [39, 40].

Because human DNA information is always evolving, compared to an individual's genes, knowledge and control only through human filtering are difficult. In identifying deleterious gene mutations, deep learning models outperform standard methods like regression analysis and support vector machines [41], as well as identifying DNA activities that are not coded [42].

Wearable gadgets are categorized in this section:

Smartwatches collect information from body parts to monitor falls, warn of special cases, and monitor the elderly to ensure that they receive the required treatment after falling, improving their quality of life and promoting healthy aging.

The records could also be used to decide physical state, health management, chronic disease management, and prevention of disease for the elderly, monitor abnormal conditions in patients with heart disease, and detect ear diseases after physiological data of the human body is collected from integrated embedded intelligent electronic devices or which directly contact the body, such as heart rate, blood pressure, and ear infections. Smartwatches and smartphones, for example, are utilized to collect physiological data connected to Parkinson's disease for scientific treatment and monitoring [52].

Smartphones and wristbands record information such as emotion, sleep patterns, fatigue, overall wellness, and liquor intake or adrenaline. This even gathers information about the individual’s interactions, such as calls and messages, push notifications, and digital device usage. The emotional trauma levels of users are assessed using a range of data sources [53].

The use of robotics and automation in healthcare and related sectors is increasing now more than ever. Surgical robot demand is expected to increase in the upcoming era, according to the International Federation of Robots (IFR), with a business worth USD 9.1 Billion predicted by 2022. Robots not only support doctors and healthcare experts in executing specific and intricate tasks but also lower their workload, boosting the healthcare organization’s efficiency [54].

Fig. (8). Artificial intelligence in medical robots.

The application determines the kinematics and dynamics of a medical robot. Serial and parallel robots are used in various activities by surgical and rehabilitative robots and service robots click [55]. Flex Picker (ABB, Zurich, Switzerland), often called the “Delta” robot, is a Parallel Kinematic Manipulator (PKM) that was developed for clinical uses but is now majorly applied in the field of the food processing industry [56]. Most medical robotics technologies are ambulatory machines with a significant loading capacity but limited Degrees of Freedom (DOF). Medical robots with many variables on either side are adaptable, precise, and trustworthy equipment that functions like a well-trained human surgeon, with a minimal error range of centimetres or less [57].

The control of medical robotics is a dissimilar subject [58] since it needs excellent precision, reliability, and repeatability while limiting the effects of external disturbances. Furthermore, designers must give enough degrees of freedom (DOF) for the results to move in the logged axis by addressing the difficulty of control and dexterity. Cleaning, sterilizing, transport, nursing, rehabilitation, and surgery are all performed by medical robots, which utilize cutting-edge technology. Adaptive robust embedded controllers are often utilized to control and navigate such complex and nimble robots [59].

Robots used in wellness programs and medical must be free of viruses that convey easily transferable and contagious viruses to other victims; thus, they must be thoroughly cleaned [60]. The majority of surgeons and effectors are designed to be used once [61]. Service robots must be sterilized regularly to avoid becoming infective carriers. Because cooking robots may be washed after usage, they have their cleaning routine.

This is one of the most important needs in medical robotics because when working with a robot in a hospital, the operator's safety is critical [62]. The operator, medical staff, physician/surgeon, and patients should all be able to be near the robot within the hospital without being at risk. Surgical robots must adhere to the IEC 80601-2-77 standard's safety requirements. The IEC 80601-2-78 standard specifies the essential safety and performance requirements for rehabilitation robots [63].

Health personnel, surgeons, and other hospital employees with no mechanical skills are trained to operate robots. Therefore, for the brief use of such apparatus, designers must constantly ensure simple architecture, easy handling, and quick maintenance. Healthcare service automatons help individuals with simulators, prosthetics, ear monitors, and retinal prosthetics, all of which are low-maintenance devices [64].

Medical robots require continuous AC/DC electricity to operate. Several renewable energy sources are employed to provide reliable electricity to medical facilities ranging in size from majority numbers, middle situated metropolitan hospitals to small hospitals [65]. Wireless power transmission for mobile robots in hospitals is also being explored to decrease the need for frequent recharge.

This type of robot works best in a hospital's welcome area, where it may disseminate information about the hospital's various units/sections and direct patients and visitors. They can take care of a large group of people without growing weary and refer them to the doctor of their choice. They are especially enticing to hospitalized children since they astound them by providing exciting experiences. As a result, their symptoms of malaise will be reduced [65].

This type of robot is designed to assist doctors like human nurses in the hospital. Robots as nurses are widely utilized in hospitals in Japan, as the country has the greatest proportion of old people (over 75 years) among Organisation for Economic Co-operation and Development (OCED) countries. Medical facilities in the country are facing an increasing problem because of this. More Japanese residents are socially compelled to look after aged family members at home instead of working due to inadequate recruitment for senior care [67]. Furthermore, nurses and healthcare workers are stressed and exhausted due to the heavy patient load. As a result, the Japanese government is considering technology alternatives to care for the country's elderly patients [68].

Fig. (9). Utilization of robots used for healthcare.

Immediate medical attention is necessary following an accident to prevent the trauma from worsening. Because of the faster recovery, more lives can be spared. This is especially true in drowning, heart failure, shocks, and breathing issues. Emergency medications, cardiopulmonary resuscitation (CPR), and Automated External Defibrillator (AED) equipment are made to be small and can convey to an emergency site by a flying drone [69]. These robots can provide emergency treatment to a mobile or remote patient with a short response time.

These robotics are utilized in telemedicine services, in which a nearby doctor collects all physical data and treatment of virus via audio-visual help [70]. These devices are especially beneficial for diseases in rural parts with few healthcare practitioners.

Transfer of products is essential for numerous tasks in hospitals. Using serving robots, these heavy-duty chores may be completed quickly. Robots are also used to deliver food to a variety of hospital patients. They are used to distribute edibles, dispense drugs and discard material to be washed, bring beddings, and transfer waste, among other things, within the hospital [71].

Vacuuming and/or mopping robots are utilized in cleaning. They are also used for sanitizing hospital environments that appear capable of delivering the needed non-industrial robot system inventors predicted years ago. These robots sanitize hospitals and eliminate bacteria and pesticides [72].

They are commonly employed to spray antiseptic concoctions over vast outdoor areas, such as city residential neighbourhoods. Such robots are being controlled remotely to avoid dangerous interaction with the antibacterial mist. Hand sanitizer spraying robots with autonomous directions are being created to prevent diseases on individuals’ hands and faces. Alcohol-based sanitizers prevent infection, parasites, and other microorganisms and minimize the transmission of infectious agents among many individuals [73].

Compared to human surgeons, surgical robots provide precise and accurate minimally invasive surgery (MIS). For remote surgery, many teleoperators have been developed [74]. Fourth-generation Da Vinci surgical systems (Intuitive Surgical, California, USA) continue to advance MIS across many surgical methods. For surgeons utilizing Da Vinci systems, this provides an upgradeable design with variable setups and a trustworthy interface. Instrument and component standardization aids hospital inventory management and efficiency [75].

Radiography is indeed one of the advanced components wherein robotics is employed with a greater reliance due to high levels of radiation and safety concerns for system interaction. The Siemens Twin Robotic X-ray [76]. Siemens Healthineers, Henkestr, Germany, is a radiology device that can perform imaging techniques, angioplasty, and 3D photography [77]. It can do a variety of X-rays in the same room, and the surgeon can watch 3D visual information as the motor is running rather than just the patient. A computed tomography (CT) 3D scan is essential to confirm the diagnosis because typical 2D X-rays frequently overlook small cosmetic cracks inside the marrow. The MultitomRax Twin Robotic X-ray system can collect a 3D image on the same system, eliminating the need for a CT system [78].

The machines can be beneficial for the recovery of individuals who have had an accident or have had a stroke [79]. It can be useful in helping and caring for persons who are paralyzed, old, or in awkward situations. The machines encourage serviceable reorganization, recompense, and nervous system rejuvenation, which successfully reduces muscle wither. Hence, helpers are relieved of arduous work, allowing healthcare resources to be better utilized.

These machines are essential to a clinic's cafeteria and storage, ensuring that elevated food is delivered by hygiene requirements. Robots have devised a variety of mechanization and autonomous systems, ranging from dining to service. In Mandarin hospitals, a robotic cook is in use. In the hospital cafeteria, waiting robots will bring the food. Cookie (Sereneti Kitchen, Atlanta, Georgia, United States) and Moley (Moley Robotics, London, United Kingdom) are two different types of culinary robots having one and two robotic hands, respectively [80].

The delivery robots are used in transporting/delivering drugs and blood samples to/from the hospital. These fully autonomous robots can operate on the ground or in the air autonomously or with the man-in-the-loop operation, whereas an operator at a distance can remotely control them [81].

Starship robots (San Francisco, USA) are another example of surface delivery robots that can transport things weighing below 100 pounds within a 4-mile (6-kilometer) radius. Pharmaceuticals, packages, foodstuffs, and meals are among the places of interest, shipped from clinics and retailers by orders placed by customers using a smartphone app. The robot's whole travel and whereabouts are tracked on a cell phone when the order has been placed.

Additionally, an electronic barrier is employed to lock the cargo compartment across the journey to assure secure delivery. The mobile phone app is used to open it at the user's end [82].