Artificial Intelligence and Its Role in the Health Care System

 

Our current society has mixed feelings about Artificial Intelligence (AI) and the ability of how machines think and recreate themselves just like any other species; however, this idea is far from the truth. Opinions of the industry leaders are deeply divided. For instance, Elon Musk thinks that AI could lead to the demise of human civilization, whereas, Mark Zuckerberg is very “optimistic” about AI (The Economic Times, 2017). AI will disrupt every part of our society. It is better to adapt to it early to stay ahead in the competition rather than staying behind. The most interesting part of AI is that it can help us achieve faster and more accurate results without human supervision. For example, in existing setups it takes months or years to complete a clinical trial as most of the analyses are done by a group of scientists working with minimal AI based systems. With the help of AI based algorithms this overall process could be done in a much shorter time. Implementation of AI in the Canadian health care system can help in the following scenarios: disease prediction, hospitalization services, and performance of surgeries and treatments.

The history of AI

The conceptualization of AI is nothing new. The beginning of AI can be traced back to ancient times when philosophers were trying to explain human thought processes as a symbolic system. Artificial intelligence as a term was coined at a conference at Dartmouth College in 1956 by John McCarthy (Crevier, 1993). The big step in AI development was the theory of computation developed by Alan Turing depicted a machine, that shuffled binary digits: 0 or 1 could simulate any mathematical deduction. This led to a revolution and, eventually, led to the creation of neurobiology, cybernetics and modern computing devices. Over past decades, exponential growth in computation of mathematical algorithms powered by major software projects such as Google DeepMind, and IBM Watson all changed the landscape of AI. Many software applications are being developed for medical research programs and medical equipment. This has resulted in dramatic improvements in diagnosis of diseases, drug research, patient care, and the overall health care system.

Artificial Intelligence in Disease Prediction

The modern lifestyle has given birth to new types of diseases which were unknown to mankind before industrialization. Diseases such as diabetes, cancer, and obesity have reached a pandemic level, and the current health care system, which is reactive, is falling behind in the race to cure diseases. The health care system needs to be active rather than reactive, which means we have to predict the disease before it occurs (Dickon, 2017). It is impossible to fight new diseases with rudimentary techniques; instead, we need to change the approach. In a fast-paced environment, it is necessary to have an even faster way to predict issues, which cannot be done using traditional approaches. Advanced AI based algorithms could be used to dramatically improve the performance of scientists, which will result in quicker data analyses, and faster disease predictions. Doctor Krittanawong stated that AI could help doctors in improving the quality of health care but not replace physicians (2018). The image processing technology such as x-rays, electrocardiogram, electroencephalogram, and MRI powered by AI and big data applications, can compile through millions of interpretations which could aid doctors and scientists to gather the patterns of certain diseases and medical conditions. Based on those patterns, various prediction data models could be created. One of these models is – a system designed to predict children’s fever type based on symptoms – shown on Figure 1. To understand the origin of fever, which could be triggered by a variety of conditions, by analyzing symptoms, the machine predicts the cause of fever in children.

Another group, certified dermatologists from the Stanford University, USA used AI based algorithms that could learn from past data, analyze clinical images of melanomas and nevi and predict skin cancer (Kubota, 2013). The results were scored more accurately than those made by dermatologists. Despite the tremendous progress made by AI in certain fields of healthcare, it must be noted that still not enough evidence exists to suggest that AI can replace physicians. Due to the fact that conversation is a missing piece in machine interaction, it is unfeasible that robots or machines will substitute medical professionals. The interpersonal communication is a crucial part of everyday doctor-patient or nurse-patient interaction which requires critical thinking, and robots cannot perform this role. AI cannot be involved in high-level conversation or therapeutic communication to gain patient trust, show them empathy, and reassure patients – all important parts of the medical professional-patient relationships (Krittanawong, 2018).

Artificial Intelligence in Hospitalization Service

Old hospital operation systems are not agile enough to accommodate the growing needs of patients as new diseases are being discovered. The new architecture has to be developed which will improve the patient care by removing paper-based records to some data-centric operations. According to Groves, Telemedicine (feature where patient and doctor can interact via videoconference across geographical boundaries) powered by computerized AI algorithms with electronic medical records which can be used to service a remote intensive care unit without healthcare professionals being on site. Also, Telemedicine can monitor several patients on a constant basis. For example, intelligent intensive care could be completely automated using advance AI systems without the presence of any healthcare professionals. The accessibility of sophisticated intensive care units in remote areas could be increased (Groves, 2008). Furthermore, AI now can help to provide optimal care for the patients through triage. Using advanced deep learning algorithms, Dr. A.I. platform helps route patients to the right level of doctor-preferred care. Dr. A.I. asks follow-up questions and collects data to suggest the next step: whether they should to go to a primary care doctor or emergency department. Although Dr. AI’s predictive response is pretty good, it cannot be 100% accurate, and it might lead to a misdiagnosis. Apart from a misdiagnosis, security breaches of systems like Dr. A.I. could have serious repercussions. According to Ouellette, (2013) there was a case of a massive security breach where hackers got access to patient’s unencrypted personal information, which is a serious concern for a hospital regarding patient privacy as the top priority.

Artificial Intelligence in Surgery Performance and Treatment

The world’s most well-known hospitals already make routine use of surgical robots. AI assisted automated or semi-automated surgical procedures dramatically improve efficiency and accuracy. The typical process of automated or semi-automated procedures involve applying the advance sensors to collect data and this data is transferred to data models to assist surgeons during the surgery. This is very helpful and currently employed in many hospitals, where without opening the abdomen, insertion can be made, and sensors used so that the whole procedure can be completed with a high level of accuracy. These state-of-the-art techniques of operation performance make the operating time and recovery time shorter (Andersen, 2017). The most famous surgical robotic equipment that is used nowadays is da Vinci. Worldwide myriad of operations are performed with da Vinci surgical robot. In Canada only, more than 20 active da Vinci surgical robots are in operation (Kapoor, 2004). Lately, ophthalmology surgeons have offloaded most of their manual pre-operation work with automated or semi automated processes, and in some cases, both the minor and complex surgeries are being done with minimum supervision by doctors. Robotic-performed surgeries in ophthalmology have eliminated mistakes which had been occurred because of surgeons’ tremor. As a result, robotic-assisted eye surgeries are now performed with the precision and stability that is required during an operation. As an example, cataract removal operations have been done with automated techniques. Gebhart stated that surgeons in ophthalmology work with high concentration, so utilizing AI automated devices made doctor’s work more precise (2013). In addition, Gebhart highlighted that a surgical device “helps you work quicker and slicker with less trauma” (2013, para. 20). Although these AI assisted automated or semi-automated procedures are very helpful and cost efficient, the surgeons still need to know and be proficient in manual surgeries because any system failure during an operation could be fatal to the patient. Moreover, algorithms and software programs cannot create consciousness to work independently; instead, they can just mimic human-like capability.

All in all, to facilitate prediction of diseases, hospitalization related services, and performance of surgeries, Canada should give priority to the inculcation of Artificial Intelligence and deep machine learning algorithms in healthcare industry.

References

Andersen, M., (December 4, 2017). Surgical automation gets more precise vision, thanks to multiple data source. Robotic Business Review. Retrieved from https://www.roboticsbusinessreview.com/health-medical/surgical-automation-gets-precise-vision-data/

Crevier, D., (1993). AI: the tumultuous history of the search for artificial intelligence. New York, NY: Basic Books.

Dickson, B. (April 10, 2017). How healthcare can benefit from Artificial Intelligence. TechTalks. Retrieved from from https://bdtechtalks.com/2017/04/10/healthcare-artificial-intelligence-machine-learning/

Gebhart, F., (October 1, 2013). Automated IOL insertion in routine and complex cataract surgery. Modern medicine network. Ophthalmology Times. Retreated from http://ophthalmologytimes.modernmedicine.com/ophthalmologytimes/content/tags/alcon-laboratories/automated-iol-insertion-routine-and-complex-catar?page=full

Groves, R. Jr., Holcomb, B. Jr., Smith M., (2008). Intensive care telemedicine: Evaluating a model for proactive remote monitoring and intervention in the critical care set. Current principles and practices of telemedicine and E-health. Amsterdam: IOS Press (para. 2)

Hirzalla, N., (2013). Building a medical diagnosis intelligent system using simple tools. International journal of advanced research in computer science, 4(1), 24-30

Kapoor, A., (2004). The robotic invasion of Canada. Canadian Urological Association Journal = Journal De L’association Des Urologues Du Canada, 8(5-6), E466-E467. doi:10.5489/cuaj.2181

Krittanawong, C., (2018). The rise of artificial intelligence and the uncertain future for physicians. European Journal Of Internal Medicine, 48 e13-e14. doi: 10.1016/j.ejim.2017.06.017

Kubota, T., (January 25, 2017). Deep learning algorithm does as well as dermatologist in identifying skin cancer. Stanford news. Retrieved from https://news.stanford.edu/2017/01/25/artificial-intelligence-used-identify-skin-cancer/

Ouellette, P., (2013, August 27). Advocate medical group endures massive data breach. HealthITSecurity. Retrieved from https://healthitsecurity.com/news/advocate-medical-group-endures-massive-data-breach

The Economic Times. (2017, July 26). Zuck vs Musk! When Elon Musk dismissed Mark Zuckerberg’s understanding of artificial intelligence. Retrieved from https://economictimes.indiatimes.com/ magazines/panache/zuck-vs-musk-when-elon-musk-dismissed-mark-zuckerbergs-understanding-of-artificial-intelligence/articleshow/59766425.cms