Subscribe to the Mindyra blog to get your free copy.
Subscribe to Mindyra for the latest information about mental and behavioral health.
// //
This post will discuss a 2019 study titled, “Using Machine Learning to Identify Suicide Risk: A Classification Tree Approach to Prospectively Identify Adolescent Suicide Attempters” recently published by Hill, Oosterhoff, and Do in Archives of Suicide Research. This study aimed to evaluate the utility of classification tree analysis in developing a screen for suicide risk.
// //
We understand that as a care provider or patient, it is challenging to keep up with all of the most updated and relevant research related to the care you provide or receive. Accordingly, we will be doing our best to provide you with summaries of recently published peer-reviewed research articles. To be clear, this is not research that we have conducted or produced - rather, these are summaries of research carried out by talented researchers across the world. Our summaries consist of our interpretation of these research studies based on information provided within research articles. These summaries are intended to be educational and full credit regarding the content of each study is to be given to the authors of the study (provided for every article we discuss).
// // Depression, Integrated Health, AI in medicine
Eric Topol makes the case for how artificial intelligence can improve health care, despite privacy concerns ILLUSTRATION: MICHAEL GLENWOOD By Sumathi Reddy March 4, 2019 9:39 a.m. ET There’s no doubt that artificial intelligence is transforming health care. But its use doesn’t come without controversy, as critics ask if AI could further dehumanize medicine and erode the doctor-patient relationship. Eric Topol, a cardiologist and director and founder of Scripps Research Translational Institute in San Diego, argues the opposite in his new book, “Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again.” The book, coming out March 12, makes the case that not only will AI improve the accuracy of diagnosis and treatment of illnesses, but it will also restore compassion to medicine. Dr. Topol is a paid adviser to two AI health companies, Verily Life Sciences, a Silicon Valley-based company that used to be part of Google, and Voxel Cloud, a China-based company. Here are edited excerpts from a recent interview: A lot of consumers and even health-care professionals are wary of artificial intelligence. They see it as something that depersonalizes medicine. Obviously, you disagree. Why? I had the same kind of bias going into this field. I spent a few years of research to figure out where we could take AI. My major conclusion was this would be the ultimate objective: that we use technology to enhance humanity. There’s many different ways that this can be achieved. The theme is the gift of time, because every type of clinician can have augmented performance by having AI support. So not just reading scans or slides but getting rid of keyboards completely, using voice recognition. And getting rid of all the administrative tasks that are such a burden. What specialties and tasks is AI particularly useful for? It’s perfect for anything that’s a pattern, be it an image, like a medical scan, or speech. For both those areas of image and speech, we now have deep learning algorithms that are as good as humans, even expert humans, in interpreting and making the right classification. So that’s the strength and that’s where we can immediately in the years ahead start to harness that to improve health care. AI brings with it a lot of privacy concerns. Consumers are worried about their private medical data being shared. How do you allay these concerns? Privacy and security is a fundamental concern. If we can’t guard that and take it to the levels of priority that it deserves then all this is a bust. Axiomatic is people owning their data—all their data, from birth on. That’s sensor data, that’s genomic data, that’s traditional medical records, everything. And if they own it, then it would be available for inputs for AI algorithms. How do people own their data? It should be a civil right. In Estonia, everyone owns their data and it sits on a blockchain platform. Many other countries around the world are moving to the model of people owning their health data and sharing it or sharing parts of it as they need or sharing it for medical research if they would like to. But in the U.S., there hasn’t been any movement in that regard. In order to do this right, to really use the power of AI, the more of your data that goes into the algorithms, the better the output will be. So right now people can’t even get all their data because they go to a lot of different providers and move around in their life and have to pay for it a lot of times. It’s just ridiculous. Our country is way behind in people having their data. And as a privacy issue, if it sits only in health systems that get hacked left and right, that’s not exactly ensuring your privacy. All the cybersecurity gurus say that the best way to protect your data privacy security is in the smallest possible units, not sitting in massive servers. Dr. Eric Topol speaks in Philadelphia in this 2015 file photo. He advocates for using more AI in health care. PHOTO:NEILSON BARNARD/GETTY IMAGES One surprising chapter in your book talks about how arbitrary diagnoses are in the medical world. How can AI help doctors with diagnosis? Medical errors are pervasive. We have well over 10 million in the U.S. a year and this is also accompanied by a tremendous amount of waste and overtesting and overmedication. So what AI can do is bring in all the data of each patient and bring in the list of probable diagnoses, many of which aren’t even thought of by a doctor when they see a patient. So what’s great here is having all the data that no human being could assimilate. And that also includes all the medical literature, up to the moment. No doctors can integrate all that. They can’t keep up with the remarkable pace of medical literature. You talk about freeing up more time for doctors. But couldn’t AI create a tsunami of information that becomes too difficult for doctors to process? No, it’s just the opposite. AI takes the remarkable amount of data and crystallizes it, distills it quickly and accurately. So the currently overwhelmed doctor with data, that’s all diffused. You have a chapter on mental health that cites research showing that people are often more comfortable talking to virtual humans than a real one. Do you think there’s a role for virtual therapists in diagnosing and treating conditions like depression or anxiety and other mental-health disorders? Yes, it’s already happening. There are already AI tools, Dr. Woebot and many others, that are expanding to provide virtual support, taking advantage of the willingness for people to disclose their innermost secrets to an avatar. But also the fact that there is such a burden of depression and mental-health conditions and there’s already some small, randomized trials showing evidence that this can be of marked benefit. Another chapter talks about diets. How in the world can AI tell me what to eat? The reason we’ve never had an individualized diet is because we’ve never had AI to bring in all the data about each person. But when you start getting billions of pieces of data about your gut microbiome, everything you eat and drink, and your sleep, and your physical activity, and your genome, that can be integrated to predict what are the foods that cause glucose spikes. So it turns out if we all ate the exact same food at the exact same time and amount, we all have markedly variable responses to different constituents in our blood like glucose, like triglycerides and I’m sure many others. You talk about patients being monitored remotely in their homes rather than being treated in a hospital. Is this a real possibility? AI can decompress the need for hospital rooms and fix a lot of the economic woes we have today as far as the investment in human capital, because the large staff of doctors and nurses that would not be necessary in the future. We already have the tools of getting vital signs and monitoring people in their home, but it relies on developing the algorithms and proving that it’s at least as safe as having people in the hospital. But I’m confident that over time it will be proven that patients can be more safely monitored in their own bedroom at a fraction of the cost. What kind of role do you see virtual medical assistants playing in the future? Will Alexa soon start telling me when my blood pressure is too high or when I’ve gained a few too many pounds? I think that’s certainly one of the most exciting applications of AI, because here you have your constant coach helping you stay healthy. If you have all the data and it’s being fed back to you with all the medical literature and all your ongoing physiology from sensors, there’s so many things that are imminently preventable. It’s an opt-in thing—not everybody will like this—but for those who are willing to be coached, it might wind up being one of the most important advances in health-care history. Write to Sumathi Reddy at sumathi.reddy@wsj.com
// // Psychiatric data
Over 50 years ago, President John F. Kennedy presented a Special Message to Congress identifying mental illness as one of America’s “most critical health problems.” The result was the Community Mental Health Act of 1963, one of the final pieces of legislation he signed before his death in November of that year.
// // Psychiatric data
WORK & FAMILY More workplaces are allowing time off for employees facing stress, anxiety or depression, but not all bosses are understanding of their workers’ needs
// // Psychiatric data
US Representative Joseph P. Kennedy III wants to put a dollar figure on the amount the United States spends as a result of limited access to mental and behavioral health services, a tally of incarceration and addiction-fighting costs he says will help make the case that the money would be better spent addressing the root issue of mental health.
