In 2011, the International Business Machines (IBM) Watson became the first computer system to defeat a human being in the popular quiz show “Jeopardy!”. Primarily a question-answering computer system, developed by IBM to tackle the game show’s trivial nature, Watson has since gone on to expand its repertoire of skills – including contributions to healthcare, natural language processing, teaching assistance, weather forecasting and even tax preparation.

Of these developments, one particular contribution stands out the most, Watson Health, IBM’s multi-billion dollar endeavour to improve global healthcare with the aid of artificial intelligence (AI) and computer learning. Now, Watson Health is aiming to arm Watson’s artificial intelligence in the battle against cancer. By utilising Watson to keep a close watch on cancer’s rapidly mutating genetics and finding their roots, IBM hopes to develop new anti-cancer therapies in the future.

The application of IBM’s supercomputer in cancer care

Unlike conventional methods and systems, Watson confers a significant advantage in identifying tumour by being able to cast a wide net for detection of malignancies, and then hone in a specific variation of the cancer when a positive detection is noted. "We have to change our whole behaviour in looking at tumours. We are missing too much and too often treatment does not work for patients… With Watson, all oncogenes become targets," explained Dr Ravindra Kolhe, breast and molecular pathologist and director of the Georgia Esoteric & Molecular Labs LLC in the Department of Pathology at the Medical College of Georgia at Augusta University.

Theoretically, this would allow clinicians to rapidly and accurately determine the specific type of cancer a patient has. Subsequently, the best choice of cancer treatment can then be administered by the clinician and, if need be, Watson can even step in to provide assistant on the best treatment modalities by looking through all the available scientific literature with respect to the specific variant of cancer.

In contrast to traditional systems and humans, Watson is able to complete the assessment of an abnormal cell sample in 20 minutes and compile a comprehensive report on the condition – a feat that would require 10 people 10 days to produce otherwise. "The majority of the time, we just tell patients they have a cancer," says Kolhe. "Watson can help us provide more comprehensive, personalized care to patients."

Watson has the ability to process a large amount of data rapidly while maintaining high accuracy. Photo credit: Phil Jones/ScienceDaily
Watson has the ability to process a large amount of data rapidly while maintaining high accuracy. Photo credit: Phil Jones/ScienceDaily

Watson Health: The pros and cons

Nevertheless, Watson is not perfect and its road to cancer care has not been a smooth one. Even now, IBM’s team is hard at work to compile medical databases from the US National Institutes of Health (NIH) and PubMed which serve to double-check Watson’s findings. Much of that is before we realise that, Watson for Oncology has already been released for use in the medical sector over the past three years and has been in development for nearly six years. Despite it all, IBM executives have openly acknowledged that Watson for Oncology is still in its infancy stage. By the end of the year, Watson would be able to offer guidance of up to 80% of the world’s cancer cases – but, it still has a tough road ahead of it which only serves to highlight the inherent flaws in the system.

For starters, Watson’s decision making in cancer care is not entirely driven by AI where several dozen physicians at the Memorial Sloan Kettering Cancer Centre in New York actually provide input on treatment recommendation. Then, there has been the difficulty in keeping Watson updated with every change in treatment guideline requiring significant amount of changes in cognitive computing. For hospitals which do utilise Watson, the cost of the supercomputer may sometimes be too expensive with costs ranging between USD200 and USD1,000 per patient. Even then, doctors are not inclined towards following Watson’s recommendations. Why? That possibly has to do with the poorly understood workings of Watson being one of a kind. Unlike research groups and scientists, Watson is not subject to peer review or vetting with many third-party studies on Watson still unreleased.

On another side of the coin, there are clinicians, too, who worry about the nature of Watson as an AI which lacks empathy and nuance. The question often posed: Is Watson able to ensure it does no harm? Despite pooling data from all over the world, Watson has trouble understanding the nuance of diseases affecting a global population. For instance, cancer treatment for the Taiwanese population may vary significantly from what pooled global data would otherwise demonstrate, and inadvertently cost harm to the patient. Mistakes such as these may even lead to increased mortality which for Watson then, would be anything but elementary.

With all the hurdles and challenges along the way, there are those who have begun to doubt the capability of Watson or an AI in the field of healthcare in light of recent developments. However, to do that would be to miss out on the bigger picture. While it may have its fair share of troubles, AI as a whole is very much in its infancy – and the benefits that Watson offers are undeniably game-changing.

As discussed prior, Watson has the ability to sift and sort through a tremendous amount of information in a very short amount of time. Moreover, like every other computer system, Watson will only continue to grow and become faster with each iteration. Unlike humans, AI does not have a ceiling when it comes to learning and growing. Crucially, Watson is able to do what no other human being is able to – to be in two places at one time.

The bottom line

The road to finding a cure for cancer will always be a difficult one. And for the better part of half a decade, IBM’s Watson has been doing just that. While there is no end in sight, the potential for AI is growing at an exponential rate. With the ability to refine data within minutes and continuously learn, the next big medical breakthrough may just be made by an artificial intelligence. MIMS

Read more:
Watson: Transforming healthcare with Artificial Intelligence
News Bites: AI to monitor and track hospital hygiene practices, Revolutionary beating Band-Aid for the heart
Can AI revolutionise drug pricing and supply the next generation of drug pipelines?

Sources:
https://www.sciencedaily.com/releases/2017/08/170807110400.htm
https://www.statnews.com/2017/09/05/watson-ibm-cancer/
https://spectrum.ieee.org/the-human-os/biomedical/diagnostics/ibm-watson-makes-treatment-plan-for-brain-cancer-patient-in-10-minutes-doctors-take-160-hours
https://www.technologyreview.com/s/607965/a-reality-check-for-ibms-ai-ambitions/