Artificial Intelligence in the Medical Field

By Samuel Pimping

Artificial Intelligence (AI) has changed lives, and it will likely begin saving lives as well. Alas, many professionals still fear the threat of no longer use due to the currency of AI. Given the rapid development of the technology, it is high time to dispel those fears and finally embrace the impossible to avoid progress.

How does AI work?

First off, what is Artificial Intelligence? AI is any machine that has the size to pretend human intelligence. ChatGPT or any other chatbot are iconic examples of AI today as natural language processing tools. You can input any question or statement in the application, and it will respond in the most like a human manner it can. As a more amaze yet eerie example, Sophia is the world’s first robot citizen powered by AI that can pretend facial expressions and apparently have natural conversations with people.

While we think of R2-D2 in Star Wars or Jarvis in Ironman when we hear the term “AI,” technology has not yet reached that level of wit. It is to make sure that the industry will further develop and discover more like a human models in the future, but Artificial Intelligence today is focused more on increase and support humans in the success of hard tasks. For now, it is best not to exceed result because there are no Robocops or Wall-e’s in our midst thus far.

The ability to learn is the element that sets AI apart from all the other machines. Artificial Intelligence provides the learning that allows it to “think” for itself like any person can through learning, think, and to correct self. The program is fed huge amounts of data, and this information is then search for patterns and conflict so that the AI can develop rules (algorithms) for creating outputs that will be useful to users.

In simpler terms, the machine learns the data, reasons out a appropriate algorithm, and self corrects by continually adjust the methods employed to attain the most right, related, and useful outputs.

Given the skill of Artificial Intelligence what practical uses does it have in the medical field and how will doctors and physicians use it?

The Revolutionary Potential of Artificial Intelligence

In healthcare, Artificial Intelligence emphasizes detection. It is utilized to make faster diagnoses without compromising accuracy. The inputs comprise patient data that has already been validated by professionals to preserve the integrity and quality of the outputs. This way, whenever similar cases appear, the algorithm has a solid foundation to mine the best data and deliver reasonably certain results based on the identified patterns. Misdiagnoses can be prevented, and treatments can be made more holistic by knowing the exact cause. As for patient care, chatbots or virtual assistants help clients accomplish administrative tasks more conveniently including setting appointments, searching for medical information, and billing.

Artificial Intelligence has been used in highly specific areas of detection with regard to medical imaging, epilepsy seizures, hypoglycemia, and atrial fibrillation. It has been established that AI has greatly succeeded in image interpretations such as radiographs, histology, and optic fundi.

In fact, in 2018, an AI algorithm called Deep Learning based Automatic Detection (DLAD) was created by Seoul researchers to analyze chest radiographs and detect abnormal cell growth including potential cancers. Physician respondents were asked to manually detect critical areas in the image, and upon comparison with the respondents’ results, it outperformed 17 out of 18 doctors. In the same year, Google Artificial Intelligence Healthcare created a Lymph Node Assistant (LYNA) that analyzed stained tissue samples to identify metastatic breast cancer tumors from lymph node biopsies. LYNA was 99% accurate in classifying whether a sample is cancerous, and doctors were able to halve their review time using it as a tool.

Deep learning is Artificial Intelligence technique derived from how humans learn through example. Using artificial neural networks as inspired by the human brain, various layers of processing enable the machine to reach accurate conclusions with little to no human intervention. Deep learning algorithms give technology the capacity to extract vital information from massive amounts of patient data in our tech centric society. Applications powered by AI can encode dictated medical notes, and these apps are being further developed to make their own notes from the synthesis of patient interviews and test result without aid.

The general uses of Artificial Intelligence are extremely significant because it can perform tasks better and faster than humans, especially when such tasks are innumerable, tedious, and repetitive. Well-made AI, by itself, is efficient, effective, and leaves no room for error. Undesirable outputs only occur when the inputs are either incorrect or incomplete or when an external factor interrupts its normal processes. In other words, since humans are still involved in specifying inputs, erroneous judgment is nonetheless a risk. As compared to traditional data analysis methods that call for significant amounts of human judgment, the improved efficiency by AI allows the possibility of greater and more detailed medical insights in
every patient’s case.

This raises an important question. If Artificial Intelligence is so wonderful, then why are its current uses not widespread yet?

The Laws and Limitations of AI

Legal implications are a primary consideration. There are no general rules, standards, regulations, or laws for the control of Artificial Intelligence in medicine because its uses are highly specific and sporadic. One could argue that there are current standards surrounding medical technology. However, will these suffice considering the extremely diverse nature of AI, medicine, and illnesses?

The complexity of the technology involved compounds this regulatory problem. Black box Artificial Intelligence refers to AI lacking an explanation suitable for human comprehension due to its convoluted algorithm. It is imagined as a black box as one cannot perceive or “look into” the intricate processes that occur within the machine. How can the technology be approved and controlled if it cannot be understood by regulatory bodies in the first place?

Understandability is the first hurdle that Artificial Intelligence must overcome to become more accepted in the legal arena. Even if full transparency were practiced, it would be useless without the corresponding understandability. Likewise, this is the same challenge that innovators face in obtaining society’s trust—the court of public opinion. Would you, a patient, place your faith in the first models of AI in medicine without fully understanding it?

Assuming that you know AI-assisted physicians outperform those without the technology, who or what will you hold accountable should a misdiagnosis occur—the physician, the programmer, the technology itself, or all of them? Again, such an occurrence is possible if the inputs were incorrectly selected due to human error. AI only reinforces what it has learned, so if what has been taught is erroneous or misleading, a problem will most likely occur. It is still humans who select what the AI will work upon; therefore, learning bias is unavoidable.

Safety is an inherent obstacle. Compared to other professions, AI application should be extremely conservative within the medical field because it involves human lives. For instance, if the algorithm delivers an incorrect diagnosis which damages the patient’s welfare, then this constitutes medical malpractice. Widespread acceptance of AI is further mitigated due to physicians who are not as receptive to the consistently increasing and ever-evolving role of technology given the surrounding doubts.

Indeed, AI is still far from the universal remedy. It is used in highly specific cases, and it presently demands a great amount of financial resources as much research needs to be done. Because it is highly commercial in nature, it also runs the risk of being inaccessible to the marginalized in the future.

To sum it up, the current downsides or limitations of AI are that it is complex, narrow, and expensive for now. This can change once more developments take place in the rapidly paced industry.

The Consequences of AI

There exists a longstanding notion that AI will eventually take over all jobs. This claim is untrue. AI is meant to support, not replace. Human intervention is still necessary as AI is made to assist in the execution of laborious tasks and help deliver favorable results more efficiently. While it is likely that some jobs will be replaced or become obsolete in the years ahead, this has already been an inherent feature of all industries in mankind’s pursuit for progress. Advancements and automation have already rendered many jobs antiquated such as typesetters, movie projectionists, switchboard operators, and lamplighters, to name a few. AI is just another event in history that necessitates adaptation by rethinking education and career paths.

Artificial Intelligence is on its way to becoming the norm. Instead of taking it negatively, we should be open to transformation. Progress should never be mitigated and must always be welcomed. In this sense, we should envision the Integration of AI into the academic curriculum of all professions to equip future generations with the knowledge and capacity to embrace technological evolution more easily.

Let us adapt to progress—and not the other way around. In doing so, we will be able to save more lives not only today but also in the years to come.