Nazanin Soghrati
When the word “revolution” is mentioned, the image most often conjured to mind is of historical human revolutions across the world. Yet, for the past decade or so, a quiet but powerful AI revolution has taken place instead. Machines have slowly risen to replace humans in labour-driven jobs, and even surpassed them in certain areas. Tech giants such as Google and IBM are only looking to further advance their reach in integrating AI in everyday industries. One of the largest and most promising of such industries is the healthcare industry. Of course, this prospect hasn’t exactly been met with enthusiasm in the physician community.
Many members of the healthcare industry, as well as the general public, are hesitant about such extensive integration of machines in hospitals and clinics, because even a small fault in the system could so easily, and quite literally, end a life. Another concern remains in the ability of machines to accurately analyze information when given tasks such as diagnosing a patient. Despite impressive efforts to construct computer systems completely off of the Bayesian-network, machines still have trouble handling uncertainty and ambiguity with normally black-and-white facts. The Bayesian-network however –developed based on British mathematician Thomas Bayes’ work– still remains as one of the primary methods to approaching complex computational problems. This network is, at its core, a probabilistic model which tries to determine the level of influence one random variable has on another variable. For example, this could be determining the influence a specific symptom has on a disease. With more symptoms or data points for the machine to operate on, the more accurately these relationships or “networks” can be calculated, potentially to a point where the machine can make a correct diagnosis all on its own.
Image Source: Dice Tower News
Nevertheless, the AI revolution is only rising, especially in healthcare. AI research company DeepMind, which was acquired by Google in 2014, has recently launched a program along with the UK’s National Health Service. Google DeepMind’s most infamous project occurred last year, when its AI-run machine, AlphaGo, was able to beat a professional, human Go (a strategy board game dependent on players’ concentration and retention of memory) player for the first time. Currently, DeepMind’s work with UK hospitals has increased as well, and the company is attempting to improve the storage and communication of data in hospitals through digital solutions.
IBM Watson, Google DeepMind’s foremost rival, has also partnered up with children’s hospitals to be further tested for use in oncological diagnoses, and its success so far has been undeniably impressive. At the most institutional level, however, the uses of AI in health remain first in storing data and second in helping to eradicate repetitive everyday tasks, such as administrative work. In Canada, hospitals have started to implement Epic HIS (Hospital Information System) in an attempt to better store patient information.
Image Source: Mobile Marketing
Undoubtedly, there will continue to be huge obstacles for us to face in the new age of AI. Causality in data remains to be one of the biggest challenges for machines, but in order to accurately analyze information, uncertainty and imperfections in data must be accounted for as well. With the resolution of such problems, health care too will begin to see an extensive revolution in its industry.
Author
