Hey siri, what's wrong with me? Watson health and the future of artificial intelligence in healthcare
Laura Mannion UCD School of Medicine, University College Dublin, Belfield, Dublin 4
In the sci-fi action film ‘I, Robot’ there is a scene in which police detective Del Spooner is interrogating Sonny - a robotic murder suspect. During questioning, the cyborg reveals he has been programmed to feel emotions, and even experience dreams. Spooner reacts strongly to this news, snapping back with contempt and derision. Certain that Sonny is nothing more than an imitation, he maintains that a machine cannot possibly contribute to society in any creative, meaningful way.
Multinational technology and consulting corporation IBM has a wildly different opinion! Two years ago they unveiled ‘Watson’, a computer system worth 1 billion US dollars . Although a far cry from the hyper-intelligent cyborg featured in ‘I, Robot’, some similarities are notable. Watson is part of a new generation of computers which operate by simulating human thought processes; a computing method known as ‘cognitive computing’ .
While Watson can’t have dreams or get sad, it does have the ability to solve problems that the average computer system cannot. Ask a search engine like Google to give information on a topic, and it will present a list of results ranked by popularity. However, Google will be unable to summarize this information, or to suggest how it might be applied in the real world. These situations require more introspective analysis, which Watson hopes to deliver.
Working to determine the underlying links between data, Watson offers suggestions based upon patterns and relationships, not popularity. This ingenuity allowed it to defeat the current human world champions in the contextual game show ‘Jeopardy’ . Watson took home the 1 million US dollar prize and undoubtedly wounded a few egos in the process.
Since then, Watson has invented a recipe for barbeque sauce and developed methods to help win fantasy football leagues. The things your dad does on a daily basis! However, these are only side projects in comparison to IBM’s latest Watson initiative. According to a recent press release, the application Watson Health “will help improve the ability of doctors, researchers and insurers to innovate by surfacing insights from the massive amount of personal health data being created daily” .
In short, IBM intends to use Watson to diagnose and treat disease. Watson Health is already in use at several cancer treatment institutes in the US and Canada where it assists in choosing individualised patient therapies . It is hoped that such an approach can use the vast trove of online data, finding crucial connections between specific cancers and their treatment.
Suggesting appropriate diagnoses and treatment represents a great step forward for Watson Health. In evidence of this, Watson business chief Manoj Saxena has announced that amongst nurses using the application, 90% follow its guidance . Still, IBM has been slow to place Watson on the frontlines of medical care, limiting system access to institutions with only the highest levels of funding and prestige.
This lack of application to the general public can lead us to wonder whether artificial intelligence is making headway in medical treatment at all. Like many medical students, Watson Health is currently studying for its USMLE exam and is waiting in the wings to prove its use in the real world. However, IBM’s partners in making Watson Health aren’t doctors and nurses, but multinational corporations that demand evidence of commercial success before a project like this can proceed.
As a result, IBM’s marketing strategy for Watson Health is aimed squarely at global industry. Big business has been intensively courted, with giants such as Apple, Johnson and Johnson, and Medtronic all signing on to incorporate the new software into their products. The public may see elements of Watson Health included in current Apple health tracker app HealthKit™ for example, laying the foundations for the real life implementation of AI in healthcare . What IBM may ultimately want however, is an even more lucrative collaboration. In 2012, expenditures for private health care insurance totalled $884 billion dollars in the US alone . The successful integration of this sector and Watson Health could spell incredible profit for the computing firm.
What could such artificial intelligence bring to private healthcare clients and who will ultimately benefit? The evidence to date points to worrying trends. In a 2012 executive paper Jamie Bisker, the Research Relationship Manager for IBM, describes how insurance premiums are set and maintained . He reports that social media has emerged as a prime source for the data which determines these premiums. For now at least a large proportion of the public are unaware of this development. Insurers are not omniscient, however. With four zettabytes (10²¹ bytes) of information as of 2013, the sheer volume of material contained within the internet presents any search engine with a formidable investigative problem. Enter Watson Health.
Bisker proposes that IBM’s new system, with its superior pattern recognition capabilities, will slip astutely through the online minefield of “irrelevant tweets and blog entries” to capture the valuable information amongst them. Along with other AI advocates, he maintains that the more appropriate the information Watson collects, the more effectively risk can be calculated and the more cost effective insurance policies can become.
Speculation aside, computing systems like these will confer large companies with even greater access to personal information. The idea of a multinational corporation trawling insidiously through Facebook, evaluating our daily life in terms of the liability it poses, is unsettling. However, there is also concern that simple overuse of AI could lead to problems. For example, doctor-patient relationships may suffer if physicians defer unduly to the machine’s authority. Medical teaching and learning might also be impacted, with apathy and disinterest setting in as computers effortlessly assimilate the trickier aspects of medical knowledge. In this sense, Watson resembles that high flying classmate; prodigious but infuriating!
Turning illness into a commodity is something to be fastidiously avoided. As big business gains access to increasingly powerful technology, the opportunity to misuse such technology increases in tandem.[KH1] To borrow a phrase, “knowledge is power - and with great power comes great responsibility”. Will responsibility matter if it stands in the way of profit?
It lies with the public, legislators, and medical practitioners to ensure that patient welfare unequivocally comes first. Artificial intelligence has already begun to change patient outcomes for the better. Watson Health has bolstered the treatment provided by oncologists and nurses, and started to interpret the masses of health data created each day.
If Watson can successfully integrate into everyday life, it may yet overturn the assumptions of the gruff ‘I, Robot’ protagonist, who sees no merit to what machines can contribute to society. With the potential to improve diagnosis and support, there is every indication that AI will soon be a significant benefit to a great number of people.
- IBM Online Press Release Team. IBM. [Online].; 2011 [cited 2015. Available from:
- Wikipedia Contributors. Wikipedia. [Online].; 2015. Available from:
- Adam Gabatt and agencies. The Guardian. [Online].; 2011. Available from:
- IBM Online Press Release Team. IBM. [Online].; 2012. Available from:
- IBM Online Press Release Team. IBM. [Online].; 2015. Available from:
- BBC Tech News. BBC. [Online].; 2014. Available from:
- Campbell M. Apple Insider. [Online].; 2105. Available from:
- MD DM. PNHP. [Online].; 2013. Available from:
- Jamie Bisker, IBM Research Relationship Manager. IBM. [Online].; 2012. Available from: