AI can detect agonal breathing but can't replace doctors, says Kevin Campbell, MD This past week, a study was published by researchers at the University of Washington in Seattle demonstrating the ability of Alexa -- the amazon home assistant tool -- to detect agonal breathing patterns associated with a cardiac arrest. In the study, researchers utilized 162 audio clips of abnormal breathing patterns and allowed the computer systems to "learn" and identify agonal breathing. A variety of devices including Alexa and iPhone 5s and a Samsung Galaxy S4 all were able to detect agonal breathing patterns. These findings were published in Nature on June 19th and the authors found that the devices accurately detected agonal breathing 97% of the time -- they believe that the technology can be provided to consumers one day in the form of a downloadable application. So, what about AI in medicine? First of all we must understand exactly what AI is. Think of the concepts of artificial intelligence, machine learning and deep learning or deep neural networks as Russian nesting dolls or matryoshka dolls. The largest doll -- AI -- is on the outside. Globally, AI is simply using computers to solve problems or make automated decisions for tasks that when performed by humans require intelligence. In simplest terms, AI is the simulation of human intelligence by computer systems. The next doll we come to inside of AI is machine learning. Machine learning involves computers using algorithms to analyze data, learn from it, and then make determinations, decisions, and predictions. In essence, machine learning is the ability for computers to "learn" without being specifically programmed to do so through complex pattern recognition. The smallest and innermost doll is deep learning and deep neural networks. Deep learning relies on two concepts: Neural networks are a set of algorithms designed to recognize patterns in an effort to cluster and classify data. Reinforcement learning centers around more goal-oriented algorithms. These algorithms are actually able to learn how to attain a complex objective (goal). Think of Watson playing (and winning games) at Chess. These technologies are what has led to developments such as the ability of Alexa to recognize agonal breathing patterns in patients suffering cardiac arrests. The FDA has recently announced that it will be focused on AI and deep learning algorithms and how they may be applied in medicine. In fact, outgoing commissioner Scott Gottlieb has committed to creating simple and safe pathways for expedited approval. This will hopefully encourage developers to create, test, and ultimately deploy more AI tools that have the potential to change the way in which medicine is practiced. But let's remember, medicine is still about the doctor and the patient -- this is one of the most sacred of all human interactions and why many of us became physicians in the first place. AI does not replace doctors -- it makes doctors better. Just as the stethoscope greatly improved a physician's ability to detect a heart murmur, and the echocardiogram enhanced a clinician's ability to correctly classify a particular valvular heart abnormality, AI will be a tool that will make doctors better. It will be yet another tool that will allow clinicians to more quickly identify and predict disease. Ultimately, AI will allow all of us to actually focus on PREVENTION of disease rather than on the TREATMENT of disease. AI and the predictive analytics that this technology will provide will be a game changer. We need to embrace AI and work to ensure that it is developed in a safe and consistent way. As more AI technologies are applied in medicine, we must critically evaluate each one, just as we critically evaluate new surgical techniques, treatments, and tools. Source
