It is undeniable that the influence of Artificial Intelligence (AI) has seen significant growth over recent years. Roles in various professions are drastically alternated, including but not limited to education, finance and computer science. Not even healthcare, which is arguably one of the most important pillars for a successful society, is exempt from the influence of AI. Decades of research have rapidly contributed to the development of medical technology, and we should continue to expect rapid changes in the healthcare industry throughout our lifetimes. While AI has, to some extent, improved the efficiency and effectiveness of healthcare professionals and physicians, therefore enhancing their roles in curing patients, there have been many debates on whether the usage of AI in healthcare is ethical or not, in addition to AI potentially replacing doctors. 

The earliest demand and research for technology-assisted clinical diagnosis can be traced back to as early as the 1960s. An initial system known as MYCIN was developed, to help treat blood diseases and save time for clinicians when diagnosing diseases. It actually drew inspiration from other professionals, such as commerce and telephone systems, which have  very similar functioning principles, as the technology was based on common computer science and AI algorithms such as decision analysis, Boolean algebra and pattern matching. Each rule consists of an “if” statement, which states a problem, followed by a “then” statement, which shows a solution. However, success has only been demonstrated in fields outside of medicine. Moreover, failure to determine a fault in telephone networks would definitely cause less of a problem in diagnosing a disease in seriously ill patients. Therefore, with little to no signs of success, and unpredictable risks, the proposal was soon abandoned and replaced. 

An improved and parallel approach to improve computer-aided diagnosis was soon invented. The first value represents the frequency of the occurrence of a particular disease; the second one is how strongly a system “hints” that such a disease is happening. The numerical values are then adjusted according to clinical importance (a cough would have relatively less clinical importance than certain lung diseases). Unlike the initial MYCIN system, it has shown impressive performance, but still contains flaws. Firstly, they are unable to identify different variants of the  same disease; secondly, they do not take into account that a disease can evolve as bacteria mutates. Thirdly, the system lacks valid explanations regarding how a diagnosis is reached. Hence, up to that time, human evaluation is still an absolute necessity. Later, pathophysiological knowledge was applied, which works by using certain details provided by the patient to eliminate irrelevant possibilities of a disease for diagnosis, providing convincing and reliable mechanisms for diagnosis. Additionally, the reasoning process can be explained in casual, comprehensible language for physicians to understand.

Fast forward to nowadays, much of the early research on the usage of AI  technology has been applied, contributing to the continuous development of technology in healthcare. An investment of 100 million USD has been allocated to implement cutting-edge technology in medicine, with robot-assisted surgery, virtual nursing assistance and administrative work being the most prominent. It still plays a major role in diagnosing diseases, mainly by analysing medical images (e.g., X-rays, MRIs, ultrasounds, CT scans, and DXAs), analysing massive amounts of information which includes data (e.g., ECG, EEG, EMG, and EHR) and symptoms (e.g., body temperature, pulse rate, respiration rate, and blood pressure), plus guiding immunotherapy through DNA and RNA sequencing. However, the source shows that it has a similar performance to human physicians. AI can also be used to assess the risk of diseases and prevent them from happening, which is most likely done by providing tailored advice to people. A recent project conducted by the Chinese University of Hong Kong (CUHK), Imperial College London and Shanghai Jiao Tong University Renji Hospital has successfully performed intercontinental telesurgery, which was  likely done by surgeons in one location coordinating the machines, and surgeons in another location supervising machines conducting surgery on a patient. While the website does not explicitly mention the use  of AI in telesurgery, it is highly possible that some of the technology mentioned above contributed massively to the surgery. This also demonstrates that technology, including AI, is capable of breaking geographical barriers, as it may be harder for surgeons to access certain areas.

Therefore, the healthcare industry would have numerous benefits from the implementation of AI. Firstly, the use  of AI is more cost-effective. As mentioned above, a hundred million US dollars was invested in implementing AI in healthcare. This funding eventually led to a massive amount of savings; it is estimated that approximately 150 billion USD would be saved. The top three applications leading to the highest amount of savings are robot-assisted surgery, virtual nursing assistance and administrative working assistance, saving USD 40 billion, 20 billion and 18 billion, respectively. Secondly, AI can identify various problems related to dementia, as it has many complicated underlying issues beneath, which include social, clinical, scientific and even economic impacts. Conversely, physicians may not be able to fully address these issues as their specialties are bound to the field of medicine, not to mention the challenges faced when diagnosing dementia with the interference of other neurodegenerative diseases such as Alzheimer’s. Thirdly, AI enables preventive treatment. As the old saying goes, “prevention is better than cure”. AI is also capable of detecting early signs of over 1,000 diseases, providing enough time for both doctors and patients to respond to the disease when the effects are still negligible. Additionally, certain tasks can be automated, and other tasks that cannot. Having the AI system complete the automated tasks can spare extra time and effort for physicians to work on tasks that cannot be automated. This also solves underlying issues caused by a shortage of doctors and long waiting times, which is very critical in Hong Kong, as well as various other places like the United Kingdom.

Yet, some limitations and downsides remain for implementing AI in the medical field. The arguably most prominent reason, is the unforeseeable risks that AI may pose. This includes ethical issues, as there is still a lot of research and discussion which should be done to evaluate whether it could improve healthcare or not; technology issues, such as a shutdown or a fault in the internet or AI system may even more profoundly endanger already seriously ill patients; social issues, as patients would most likely perceive a human doctor to be more reliable and empathetic; and finally privacy issues. Recently, the  award-winning healthcare platform, Medisafe, claimed to use AI to help doctors prescribe the most appropriate and effective medicine by using previous patient records.This raised various concerns and criticisms amongst the public, one of them being disrespecting the privacy of former patients. Althought not being directly linked to the implementation of AI, it showcases that privacy issues like these  may arise due to using AI to process large amounts of data.

This sparked considerable debate and discussion. Nowadays, everyone is debating whether a certain profession would be replaced by AI, including medics. Few are confident enough to proclaim that AI will definitely replace doctors, but famous billionaire Bill Gates said that “with the increased usage of AI, humans would no longer be needed for most things”. While he did not clearly mention whether healthcare professionals would be replaced as well, it is highly likely that he also thinks that doctors are threatened by the replacement of AI, as another source shows that the time of physicians, on average, spent on administrative work, is approximately 16.6% of working hours, with the average psychiatrist spending one fifth of their working time on administration, which can almost certainly be assisted with AI. As mentioned before, the work that AI can  do could easily solve the shortage of doctors and mental health professionals. While it seems highly possible that physicians would be replaced by AI, the ultimate decision still belongs to us humans.

Still, there are many reasons why doctors are not yet able to be replaced by AI. Although in some specialties, like surgery, where emotion is less important or may even hinder the process of surgery, it is essential for there to be an emotional bond between patients and doctors. AI merely consists of programs and instructions and therefore cannot, at least at the current stage, be able to feel emotions. While it can provide valuable instructions and recommendations for the treatment and curing of diseases, or even motivate patients suffering illnesses to have a positive attitude in life, it cannot adequately copy the empathy of human physicians. Moreover, doctors have a non-linear working method, which means, in simpler terms, we do not process information through lines of programming. This gives us humans, an advantage, as there are some cases where it is even too complex for AI systems. In an episode of the series House MD, the doctors could not correctly diagnose why a boy was poisoned and therefore prescribed the wrong medicine, until they surprisingly found out that his skin absorbed poison from his pants, which were kept in a truck. Similarly, it would be very unlikely for AI or any technology to have come up with such a diagnosis and reasoning. 

In conclusion, while AI and various other technologies have been thriving in the field of medicine, it remains uncertain whether it is ethical and reliable or not, therefore continuously sparking lots of debates. Currently, amidst the rapid development of AI, human doctors and other health professionals are still essential, but may face the threat of replacement. With extensive and rigorous research in both biomedicine and artificial intelligence, it is expected that there will be drastic changes in the healthcare industry that we know today.

Up Next