Navigation auf uzh.ch
Navigation auf uzh.ch
Will algorithms replace doctors some time in the future? Listening to Marcel Salathé, who directs the Digital Epidemiology Lab at ETH Lausanne, you might come to that conclusion. Last Thursday at the Digital Health: Exploring Ethics and Policy conference at UZH he reported on software that boasts a 99 percent success rate in terms of recognizing plant diseases on the basis of photos of leaves. What works for plants is also being tried out on humans. Salathé cited a study where an algorithm identified skin cancer in 60 percent of cases – with dermatologists managing only 46 percent in the same study. The more images fed into the software, the better the results – a process experts call “deep learning.”
“In the coming years we’re going to see the publication of thousands of studies showing that algorithms can perform certain tasks better than humans,” claims Salathé – although he does admit that rather than making firm diagnoses, algorithms will be providing pointers as to when it makes sense, for example, to take a tissue sample in cases where cancer is suspected.
How radically will digitization change medicine? And what ethical, legal and social implications do we have to keep an eye on? These questions were the focus of the conference, organized by the newly established Health Ethics and Policy Lab, part of the Epidemiology, Biostatistics and Prevention Institute at the University of Zurich.
Effy Vayena, the SNSF professor who heads the lab, is convinced: “Medicine and research into health are at a turning point, with digitization giving unprecedented insights into the causes of disease, and enabling us to develop new ways of treating them. At the same time, this digital transformation is posing huge ethical and legal challenges in areas such as data protection.”
Antoine Geissbühler, professor at the University of Geneva’s Department of Radiology and Medical Informatics, warned that Switzerland was lagging behind in terms of the digitization of healthcare. At various points in the conference, the fact that datasets are not sufficiently connected was identified as the main problem in this regard.
For Peter Meier-Abt, Vice President of the Swiss Academy of Medical Sciences (SAMS), universities and hospitals in Switzerland aren’t sufficiently connected, and use a wide range of different incompatible systems to manage their data. There is help at hand, however, in the form of the national Personalized Medicine initiative and the Swiss Personalized Health Network it has given rise to. The next step will be to network health data on an international basis.
According to Tania Simoncelli from the Massachusetts Institute of Technology (MIT), the digitization of healthcare in the US has progressed more rapidly. Since Barack Obama took office, the US administration has invested more than USD 29 billion in digitizing the health system. On the other hand some technology firms, hospitals and labs have been trying to block the exchange of data because they fear sharing information with competitors might jeopardize their business. Simoncelli suggested that Switzerland would be well advised to address all the issues related to the digitization of healthcare more slowly, but more rationally and consistently.
Indeed there’s still a lot to sort out, not just in Switzerland. At the conference Petra Wilson from the Digital Health & Care Institute in Edinburgh described how a third of nations worldwide don’t have a legal framework in place regulating e-health, and some don’t have even basic legislation governing data privacy. Petra Wilson also said that action was required in relation to health data security. According to the Cyber Security Intelligence Index 2016, the healthcare industry is the most attractive target for cybercriminals worldwide.
Various speakers emphasized that secure, socially compatible solutions to the problem of handling digital health data were in sight. A positive example cited by Ernst Hafen of the Institute of Molecular Systems Biology at ETH Zurich was the Midata.coop initiative. Midata is a cooperative project where citizens have a kind of bank account where they can deposit their data securely and centrally, and determine who has access. Any commercial proceeds from the use of the data are reinvested in research benefiting society as a whole. Midata has launched various pilot projects. There are also plans to set up similar cooperative schemes in countries including Germany, the Netherlands, Ethiopia, and Vietnam.
