July 20199 physician colleagues are not credentialed or trained to carry out. The role of the doctor has changed, and will continue to change.As a medical leader, I have a deep interest in understanding, foreseeing, and planning for such changes, so that the profession itself is not reacting, but proactively anticipating and leading the change itself. One of these potential disruptive drivers of change in medicine comes from the rise of artificial intelligence (AI) in health.Late last year, researchers surveyed over a thousand GPs in the UK, and found that the majority of them said that AI would never replace doctors. However, more and more commentators, computer scientists, futurists and medical researchers are publishing papers predicting a seismic change in the clinical landscape, where algorithms will start automating, and replacing, a lot of the duties carried out by health professionals, including doctors.Already, the FDA in the US has approved AI devices in radiology, ophthalmology and other areas of medicine. In February of this year, researchers reported in Nature Medicine that an AI algorithm outperformed junior paediatricians in diagnosing childhood illnesses (but not senior ones). In March of this year, a paper in the European Journal of Cancer found that a deep learning algorithm outperformed 136 of 157 dermatologists in classifying melanomas. I started off this article with using surgeons as an example of doctors who have moved to the apparent peak of their profession, who can validly argue that AI will never replace the need for their hands-on surgical expertise, but recent research indicates that even surgeons are not exempt. The makers of surgical robots are already developing autonomous versions of their machines, and early experiments show that robots can cut and sew up wounds just as well as a human surgeon.It is incumbent on the medical profession to be aware of the potential issues that the rise of AI in health can bring. We need to work with the technologists so that we can understand how and why clinical decisions are made by AI ­ currently, there is a "black box" that sits underneath deep learning algorithms that are not open to being deciphered easily. Data governance is also key ­ the power of machine learning is in the availability of big data that is accurate, clean and free of bias. Health information privacy is paramount, and cybersecurity essential. There are also legal considerations. A lot of AI code is open source, and as we develop AI apps and software, the issue of intellectual property ownership will come up.In addition, liability will become an issue, especially when things go wrong with AI. Who does a patient sue when a medical error occurs if a doctor relies on an AI to make a clinical decision? The European Union has even considered giving AI agents legal personhood status so the AI can be sued, a controversial idea that is still being debated in that jurisdiction.As we approach the "singularity", as predicted by futurists like Ray Kurzweil, it is important that we ensure that AI comes not to replace doctors, but enhances our role, so that it frees us from the more mundane and routine, to practice medicine in a more holistic, compassionate way. As experts like Eric Topol say, AI has the potential to bring us back to the heart of medicine, and free all of us to put our focus back on why we became doctors in the first place ­ to care for our patients.In other words, and without irony, artificial intelligence may very well help us to remove the artificial from health, and bring back the intelligence, and even more so, the humanity. AI MAY VERY WELL HELP US TO REMOVE THE ARTIFICIAL FROM HEALTH, AND BRING BACK THE INTELLIGENCE, AND EVEN MORE SO, THE HUMANITYErwin Loh
< Page 8 | Page 10 >