Tucked away in the Thomas J. Watson Center, IBM’s research headquarters in a small northern suburb of New York City, an unremarkable door leads to a large room. The room houses one of the world’s most powerful computers. Servers fill the space, covering several hundred square metres and cooled by a battery of loud fans. The dark mass of electronic circuits represents the latest and greatest hope of US physicians and promises to transform the country’s healthcare system.

The machine became a celebrity in January 2011 when it won the famed televised game show Jeopardy. Capably answering questions put by the presenter, Watson – for that is its name – systematically crushed the human competition.

“We wanted to show that our computer was capable of speaking and understanding English, not just deciphering codes and raw data, like other computers,” explains Adam Lally, one of the 12 members of the computer’s original development team. “For the contest, Watson read and assimilated nearly 200 million pages of text, including all of Wikipedia.”

The day after the televised victory, IBM went to work on finding a commercial outlet for the machine. “We decided the logical next step was to put it to use in medicine,” Mr Lally recounts.

Marty Kohn, in charge of Watson’s medical programme, states the starting point of the new project: “The number of data points and scientific articles is increasing every year at an ever-faster rate. No care provider can use all this information, or even access it all”. The supercomputer, on the other hand, is able to read 60 million pages of text per second and can instantaneously manage the new information. IBM’s objective is to assist physicians in making diagnoses and suggest the most appropriate treatments.

IBM hopes to market its service within a few years. Watson and similar computers should become accessible to every doctor on the planet thanks to cloud computing, a system of online storage and viewing. The machine’s medical skills are currently being tested at the Memorial Sloan-Kettering Cancer Center, a New York cancer treatment and research hospital, and the Cleveland Clinic, a hospital in Ohio. US insurer WellPoint plans to call on Watson’s talents soon to analyse the effectiveness of the treatments it pays for its customers to receive.

Shared responsibilities

Benoit Dubuis, Chairman of Swiss association BioAlps, warns of another effect: “With the ever- widening use of new technologies, there are more and more actors in the medical chain. Engineers and mathematicians are joining the physicians and life scientists. We must be careful not to end upin a Tower of Babel, where each person speaks a language specific to his own specialty.” To avoid this pitfall, the EPFL institute of technology in Lausanne has opened a school of life sciences, which aims to promote communication between engineers and biologists.

The increasing number of players is a challenge to physicians’ area of responsibility. “Currently, the doctor bears full medical responsibility. But what will happen in a few years?” asks Dr Panese. “The responsibility will also be borne by the person who inputs the data to the information system, or the creator of an algorithm. In future, it will not be out of the question for a mathematician to testify in a court of justice when a medical error has occurred.” May even Watson be incriminated? “The physician must absolutely stay in control of his diagnosis. Watson is there merely to helphim, not to take responsibility for anything,” comments Mr Kohn.

For Mr Gauthier, the danger is in the mathematical formulae of the new systems. In his view, “If machines like Watson are transformed into ‘black boxes’ and no one knows how they work, there will clearly be a problem of responsibility.” The expert throws down the gauntlet: “Physicians and society as a whole must think about the importance of machines in medicine, and clearly define their role.” ⁄

Herbert Chase, MD, professor of Clinical Medicine at Columbia University, is working with IBM on the project. He details Watson’s advantages: “The machine will allow patient treatment to be personalised as never before”. Beyond accessing the most relevant medical literature, Watson will be able to use a patient’s individual data: “The hospitalised person’s diagnosis will be customised based on their history. Watson will read their medical file and will know their allergies, the medications they have taken throughout their life, and even the specifics of their genetic makeup. It will be able to foresee drug interactions, even when thousands of drugs are involved or when they are nearly undetectable. The patient will also be able to indicate his treatment preferences and what side effects he is willing to put up with.”

Watson’s diagnoses should thus prove more precise than any contemporary doctor’s. According to the Institute of Medicine,a US NGO that deals with health matters, one in five diagnoses is incorrect or incomplete, and nearly 1.5 million medication errors are made every year in the USA. “Watson helps us to improve our aim,” says Daniel Kraft, a professor at Singularity University in the heart of Silicon Valley and Executive Director of FutureMed, a programme that explores the use of new technologies in medicine.

In the USA, the Watson project’s announcement has had a spec- -tacular impact. The question for the academic world and media is: will the robot replace the doctor? Vinod Khosla, one of the best-known investors in new Silicon Valley technologies, thinks so. He predicts that computers and robots will be able to handle most tasks performed by doctors – even surgical operations – and will replace four out of five physicians in the USA. Stanford University author and physician Abraham Verghese fiercely challenges this view: “People who say that don’t understand what treatment is. A physician’s work does not begin and end with diagnosis. It’s all about providing psychological support for our patients. And that’s something no robot can do.”

IBM’s Mr Kohn adds reassuring words: “Watson doesn’t pretend to be anything but a helpful tool. It will never replace the physician”.

But even as a mere assistant, the supercomputer seems to be redefining the role and working methods of medical personnel. “The physician will spend less time and energy in diagnosing his patient. He will be able to concentrate on psychological follow-up,” says John Eric Jelovsek, who heads the Cleveland Clinic’s Simulation Center.

In the eyes of Thomas Gauthier, a specialist in new medical techno-logies at Geneva’s Haute Ecole de Gestion, the transformation has already begun. “More often these days,” he says, “the physician is involved in mediation or negotiation. Patients frequently arrive at their doctor’s office with a handful of Wikipedia printouts. They have the illusion that they know what is wrong with them and sometimes refuse the treatments their doctor offers”. For Francesco Panese, Associate Professor of Social Sciences and Medicine at the University of Lausanne, the culture of medicine is in the midst of transformation. “The paradigm of the 1960s, when a doctor would diagnose his patient and decide his treatment, no longer exists. We are going to enter the era of the ‘negotiated clinic’. Patients have more and more access to information. They want to know what is happening to their bodies.”

Mr Kohn is counting on Watson to facilitate the transition to negotiated medicine: “The machine can explain, in English, precisely what the patient is suffering from and what the treatment options are. It has been proven that a treatment chosen with the patient’s agreement is more effective. We want to encourage that.”

But there is still the fear that Watson will threaten the relationship between patients and caregivers. “The supercomputer could disrupt the trust placed in the practitioner,” says Mr Gauthier. “When world chess champion Gary Kasparov lost against the Deep Blue computer, the human race realised for the first time that a machine could outperform a human being. It was a shock. The patient may be tempted to turn to the machine and reject the physician’s diagnosis in the belief that it
is of lower quality.”

Watson’s arrival on the market will also have an impact on physicians’ training. “The doctors of the future will need different skills. Memorisation will be less important, and the analytical skills to understand the information computers provide will be more so,” says Mr Kraft. “The computer will be the one who takes on memorising all the side effects of medications, which are too many for a single physician to know them all.”

Big data: a challenge for medicine
The data explosion now affects every corner of our society. But it is in the field of medicine that it has raised the greatest hopes. How will it be possible to store, share, analyse and view this mass of information (also called “Big Data”) to improve patient treatment? IT processing of gigantic mounds of data presents several advantages for the healthcare sector. According to consulting firm McKinsey, the information revolution will result in savings of $300 billion to $450 billion annually for the US healthcare system, representing a decrease of 12% to 17% in costs.
But hopes go beyond the merely financial advantages. “Combining the data from a patient’s file with global medical information will result in continuous improvement in diagnoses,” says Francesco Panese, Associate Professor of Social Sciences and Medicine at the University of Lausanne. “In future, it will be possible to predict what type of disease someone may contract many years in advance.” But there is another side to the coin: “We are going to move from an era of uncertainty to one of worry. We will know our ailments in advance, but in some cases, medicine will not know how to treat them”.