Paper Example. Artificial Intelligence in Prosthetic Limbs

In the current world, the difference between a real and a prosthetic limb cannot be noticed easily. People do not fear limb amputation as there are various alternative prosthetic limbs due to artificial intelligence. Engineers are also working on ways to decrease the problems amputees face with better and new prosthetics, which return some of their abilities. The prosthetics market was revolutionized by robotic limbs by combining various prosthesis movements with sensor technology forming bionic limbs that are AI-enabled. However, the generated bionic AI-enabled limb, in most cases uses a human-machine interface to interpret the intentions and feelings of a patient and then sends commands to the prosthetic limb (Alshamsi, Jaffar, & Li, 2016).

Is your time best spent reading someone else’s essay? Get a 100% original essay FROM A CERTIFIED WRITER!

Order now

The human-machine interface has eight electrodes that are used to carry weak electrical signals from the stump of the patient. After training the user and the algorithm, the model of regression is then uploaded to the embedded system and used to operate or control the artificial limb rather than a virtual cursor. Nonetheless, as in conventional control, the artificial limb is always controlled in a control-velocity mode such that the stronger the contracted user, the faster the artificial limb is moved. As time progresses, the bionic arm and the patient undergo training so that the machine-learning algorithm can learn the interpretation of the unique electrical signals (Vedaraj, Parijaat, & Rao, 2012). Also, there is a mini-computer located inside the prosthetic, which helps the limb to amplify and send signals to the system of the AI-enabled bionic limb. The mini-computer then operates the algorithm (machine learning) to translate the messages before commanding the motors of the limb to move in the direction the patient wants (Smart, 2016).

There is also a new application of artificial intelligence based on reinforcement learning (an automatic version of classic trial-and-error), which promises success in various small clinical experiments consisting of a non-disabled patient and an amputee. In most cases, human technicians or surgeons spend most of their time working with amputees to manually adjust the robotic limbs for them to function according to the patient's wish. Through comparison, the learning reinforcement technique automatically tunes the robotic limb granting the patients the opportunity to use their limbs freely within 10 minutes (Alshamsi et al., 2016).

The tuning process is then aimed at particular parameters that define the relationship between motion and force in using the prosthetic limb. For instance, some settings might state the stiffness of the artificial joint limb or the motion range that is allowed to swing a limb back and forth. In the depicted scenario, the prosthetic limb will have a dynamic combination of more than ten parameters that need trial-and-error tuning (Smart, 2016). In most cases, the starting point for the parameters are always imperfect for most freshly unboxed prosthetic limbs, but adequate for the patients or users to stand while making simple walking movements or while swaying the hands.

Training an artificial limb is always a complex and complicated co-adaptation process which requires the limb to know how to corporate with the mind, controlling all the body parts. The procedure might involve initial clumsiness, but as time progresses, the patient and the prosthetic limb shall adopt (Alshamsi et al., 2016). On the other hand, the reinforcement-learning algorithm must also prove its work with a fairly small set of training information from the artificial limb users. Therefore, most individual human amputees cannot walk for a long period for the sake of training the algorithm; they may walk for about 15 to 20 minutes then take a break to allow the limb to adapt (Vedaraj et al., 2012).

However, data training might have some limitations to the patient such that it is impossible to allow the artificial limb users to fall in their trial-and-error sessions for the algorithm to learn the cases due to safety reasons. Thus, despite the difficulties, initial results might be accurate and promising. The reinforcement learning algorithm is always trained to recognize specific motion patterns in the collected data from the embedded sensors in the artificial limb. Some initial constraints are then set on the algorithm to prevent more undesirable occasions, which might cause the patient or the wearer to fall. The algorithm will eventually learn to focus on specific data patterns that match a smooth and stable walking pattern (Vedaraj et al., 2012).

Giving artificial users a way to inform the algorithm that a specific walking pattern feels worse of better might be challenging because such inputs in the AI-enabled system night fail to capture the complex coordination of an individual's perception and cognition in choreographing the body movements (Alshamsi et al., 2016). Moreover, in many scenarios where amputees have tried the AI prosthetic limbs, they depict that the limbs are easy to use as they can easily rotate their limbs. As both the wearer and the system adapt, movements become more natural and can control the movement of their limbs easily and independently (Smart, 2016). For instance, a patient can slowly rotate the hand and open it quickly at the same time. Therefore, due to artificial intelligence, the new bionic limb can be said to be more natural and superior in its daily tasks operations as opposed to other limb amputation options.

On the issue of user safety, AI-enabled systems in prosthetic limbs have provided precautionary safeguards for artificial users (Kristjansson et al., 2016). Notably, the presence of enhanced research methods has led to the invention of brain-controlled robotics; hence, making a great impact in current prosthetic limbs. However, user safety had not been emphasized in most of the prosthetic limbs especially for both upper and lower limb amputations. The response of nerve impulses in normal human assists the natural muscles in making movements. The stimulation of the nerves provides electrical activity that is measured through the Electromyography (EMG) (Keskinbora, 2019). However, the technique of using EMG to detect any reflex through muscle activity has not emphasized on user safety. AI has improved muscle and nerve feedback through the utilization of movement and sensory components. For instance, prosthetic limbs with human-like sensation and reflexes have been developed that are more effective compared to those using the EMG technology (Keskinbora, 2019). With the sensory components, there is an enhancement in the user safety since reflexes can be linked to brain activity. An example is the prosthetic leg that has an ankle fitted with microprocessors to determine whether the user is moving on uneven or flat terrains or up and down stair cases (Keskinbora, 2019). The functionality of prosthetic limbs on rough terrain has caused challenges to the users whereby amputees fall more times compared to healthy individuals (Nast, 2017). AI has reduced these falls through the use of sensors and micro-sensors that also assist in the movement of the upper prosthetic limbs with improved accuracy. Ongoing research on the integration of AI in prosthetic limbs will create life changing transformations despite the ethical implications (Keskinbora, 2019). Ongoing wars and accidents caused by different types of machinery have rendered individuals amputees; hence, the need for enhanced Artificial intelligence integration in prosthetic limbs (Nast, 2017). Ultimately, such limbs will not limit the life of amputees while engaging in daily activities.

References

Alshamsi, H., Jaffar, S., & Li, M. (2016). Development of a Local Prosthetic Limb Using Artificial Intelligence. International Journal of Innovative Research in Computer and Communication Engineering, 4(9). Retrieved from https://pdfs.semanticscholar.org/9a4d/2234806eedfe6350a89c9ca5603e3b7fe99e.pdf

Keskinbora, K. (2019). Medical ethics considerations on artificial intelligence. Journal Of Clinical Neuroscience, 64, 277-282. doi: 10.1016/j.jocn.2019.03.001

Kristjansson, K., Sigurdardottir, J., Sverrisson, A., Sigurthorsson, S., Sverrisson, O., & Einarsson, A. et al. (2016). Prosthetic Control by Lower Limb Amputees Using Implantable Myoelectric Sensors. Converging Clinical And Engineering Research On Neurorehabilitation II, 571-574. doi: 10.1007/978-3-319-46669-9_94

Nast, C. (2017). AI Is Fueling Smarter Prosthetics Than Ever Before. Retrieved 4 December 2019, from https://www.wired.com/story/ai-is-fueling-smarter-prosthetics-than-ever-before/

Smart, B. (2016). Military-industrial complexities, university research, and neoliberal economy. Journal of Sociology, 52(3), 455-481. Doi: 10.1177/1440783316654258

Vedaraj, I. R., Parijaat, S., & Rao, B. V. A. (2012). Material analysis for artificial muscle and touch sensing of cooperative biomimetic manipulators. The International Journal of Advanced Manufacturing Technology, 60(5-8), 683-692. Retrieved from https://link.springer.com/article/10.1007/s00170-011-3640-8