Home / Technology / Google continues working to use machines for health analysis

Google continues working to use machines for health analysis



A woman with advanced breast cancer came to a hospital in the city, the fluids already flooding her lungs. She saw two doctors and obtained a radiology scanner. Hospital computers read his vital signs and calculated a 9.3 percent chance that he would die during his stay.

Then it was Google's turn. A new type of algorithm created by the company read about women-175,639 data points-and issued its assessment of its risk of death: 19.9 percent. She died in a matter of days.

The heartbreaking story of the death of the unidentified woman was published by Google in May in an investigation that highlighted the potential of neural networks, a form of artificial intelligence software that is particularly good at using data automatically learns and improves Google had created a tool that could predict a series of patient outcomes, including how long they can stay in hospitals, their chances of re-entering and the chances of them dying soon.

What most impressed the medical experts was Google's ability to analyze data that was previously out of reach: notes hidden in PDF files or scribbled on old graphics. The neural network engulfed all this ungovernable information and then spat out predictions. And he did it much faster and with greater precision than the existing techniques. The Google system even showed which records led to conclusions.

Hospitals, doctors and other health care providers have tried for years to better use electronic health records and other patient data. More information shared and highlighted at the right time could save lives, and at least help medical workers spend less time on documentation and more on patient care. But current methods of extracting health data are expensive, cumbersome and time-consuming.

Up to 80% of the time invested in today's predictive models goes to the "shear job" of making the data presentable, said Nigam Shah, associate professor at Stanford University, co-author of Google's research paper. , published in the journal Nature. Google's approach avoids this. "You can throw the kitchen sink and not have to worry about that," Shah said.

Google's next step is to move this predictive system to clinics, AI chief Jeff Dean told Bloomberg News in May. known as Medical Brain-is working on a series of AI tools that can predict symptoms and diseases with a level of accuracy that faces hope and alarm.

Within the company, there is a lot of excitement about the initiative. "They finally found a new application for AI that has a commercial promise," says a Googler. Since Google of Alphabet declared itself an "AI-first" company in 2016, much of its work in this area has been aimed at improving Internet services. The advances of the Medical Brain team give Google the opportunity to enter a new market, something co-founders Larry Page and Sergey Brin have tried several times.

Healthcare software is largely encoded today. In contrast, Google's approach, where machines learn to analyze data on their own, "can surpass everything else," said Vik Bajaj, a former Verily executive, a health care arm of Alphabet and CEO of investment firm Foresite Capital. "They understand what problems are worth solving," he said. "They have now done enough small experiments to know exactly which directions are fruitful."

Dean conceives the artificial intelligence system directing physicians towards certain medications and diagnostics. Another Google researcher said that existing models overlook obvious medical events, even if a patient had prior surgery. The person described the existing hand-coded models as "an obvious giant obstacle" in health care. The person asked not to be identified to discuss the work in progress.

Despite all the optimism about Google's potential, the use of artificial intelligence to improve healthcare outcomes remains a big challenge. Other companies, especially the IBM Watson unit, have tried to apply AI to medicine but have had limited success saving money and integrating technology into reimbursement systems.

Google has always sought access to digital medical records, also with mixed results. For his recent research, the Internet giant closed deals with the University of California-San Francisco and the University of Chicago for 46 billion anonymous patient data. The Google AI system created predictive models for each hospital, not one that analyzes data between the two, a more difficult problem. A solution for all hospitals would be even more challenging. Google is working to secure new partners to access more records.

A deeper immersion in health would only add to the large amount of information that Google already has. "Companies like Google and other technology giants will have a unique and almost monopolistic ability to capitalize on all the data we generate," says Andrew Burt, privacy director of data firm Immuta. He and pediatric oncologist Samuel Volchenboum wrote a recent column in which they argue that governments should prevent this data "from only a few companies," as in online advertising.

Google is treading carefully on patient information, especially public scrutiny about data collection increases. Last year, British regulators slapped DeepMind, another Alphabet AI lab, for testing an application that analyzed public medical records without telling patients that their information would be used that way. With the latest study, Google and its hospital partners insist that their data is anonymous, safe and used with the patient's permission. Volchenboum said the company may find it more difficult to maintain that data rigor if it expands to smaller hospitals and health care networks.

Still, Volchenboum believes that these algorithms could save lives and money. He expects health records to be mixed with a sea of ​​other statistics. Finally, artificial intelligence models could include information on local weather and traffic, other factors that influence patient outcomes. "It's almost as if the hospital were an organism," he said.

Few companies are better positioned to analyze this agency than Google. The company and its cousin of the alphabet, Verily, are developing devices to track many more biological signals. Even if consumers do not massively take portable health trackers, Google has many other data pits to take advantage of. Know the weather and traffic. Google Android phones track things like the way people walk, valuable information to measure mental deterioration and some other ailments. Everything that could be thrown into the soup of medical algorithms.

Medical records are only part of Google's AI health plans. Your medical brain has deployed AI systems for radiology, ophthalmology and cardiology. They are also flirting with dermatology. The staff created an application to detect malignant skin lesions; A product manager walks through the office with 15 fake tattoos on his arms to prove it.

Dean, the head of AI, emphasizes that this experimentation is based on serious medical advice, not just on curious software encoders. Google is starting a new test in India that uses its AI software to detect images of the eyes in search of early signs of a condition called diabetic retinopathy. Before publishing it, Google had three retina specialists arguing furiously with the first results of the research, Dean said.

Over time, Google could license these systems to clinics or sell them through the cloud computing division as a kind of diagnostics, like-a service. Microsoft, one of the main rivals of the cloud, is also working on predictive services of artificial intelligence. To market an offer, Google would first need to have more records in its hands, which tend to vary widely among healthcare providers. Google could buy them, but that may not be as good for regulators or consumers. The agreements with UCSF and the University of Chicago are not commercial.

For now, the company says it's too early to establish a commercial model. At the annual Google developer conference in May, Lily Peng, a member of Medical Brain, walked through the team's research to compare humans to detect the risk of heart disease. "Again," she said. "I want to emphasize that this is really early"

Bloomberg News

Health data management now only supports Internet Explorer 11 and later.
Update your browser.
Get more information: https://www.microsoft.com / en-us / WindowsForBusiness / End-of-IE-support

& # 39;). insertBefore (& # 39 ;. site-header-wrapper & # 39;); } }); } if (GetIEVersion () <= 8 && GetIEVersion ()! == 0) { var script = document.createElement (& # 39; script & # 39;); script.setAttribute ("src", "https://ajax.googleapis.com/ajax/libs/jquery/1.12.2/jquery.min.js"); script.setAttribute ("type", "text / javascript"); (document.getElementsByTagName ("head") [0] || document.documentElement) .appendChild (script); } window.setTimeout (setIEBanner, 3000);

(function () { var d = document, h = d.getElementsByTagName (& # 39; head & # 39;) [0] s = d.createElement (& # 39; script & # 39;), sc = & # 39; https: & # 39 ; == document.location.protocol? & # 39; https: // & # 39 ;: & # 39; http: // & # 39 ;; s.type = & # 39; text / javascript & # 39 ;; s.async = true; s.src = sc + & # 39; s.dpmsrv.com / dpm_c415505dca69be631ca5d391b3ccd2b44b52d017.min.js & # 39 ;; h.appendChild (s); }) ();

window .__ wtw_lucky_site_id = 80021;

(function () { var wa = document.createElement (& # 39; script & # 39;); wa.type = & # 39; text / javascript & # 39 ;; wa.async = true; wa.src = & # 39; https: //d10lpsik1i8c69.cloudfront.net/w.js&#39 ;; var s = document.getElementsByTagName (& # 39; script & # 39;) [0]; s.parentNode.insertBefore (wa, s); }) ();

// START Facebook tracking pixel ! function (f, b, e, v, n, t, s) {if (f.fbq) return; n = f.fbq = function () {n.callMethod? n.callMethod.apply (n, arguments): n.queue.push (arguments)} ; if (! f._fbq) f._fbq = n; n.push = n; n.loaded =! 0; n.version = & # 39; 2.0 & # 39 ;; n.queue = []; t = b.createElement (e); t.async =! 0; t.src = v; s = b.getElementsByTagName (e) [0]; s.parentNode.insertBefore (t, s)} (window, document, & # 39; script & # 39 ;, & # 39; https: //connect.facebook.net/en_US/fbevents.js&#39;); fbq (& # 39; init & # 39 ;, & # 39; 1255873067799046 & # 39;); fbq (& # 39; track & # 39 ;, & # 39; PageView & # 39;); fbq (& # 39; track & # 39 ;, & # 39; ViewContent & # 39;); // END Facebook tracking pixel

// BEGINNING Marketo Munchkin (function () { var didInit = false; initMunchkin () {function if (didInit === false) { didInit = true; Munchkin.init (& # 39; 555-ETU-514 & # 39 ;, {"wsInfo": "j1RQ"}); } } var s = document.createElement (& # 39; script & # 39;); s.type = & # 39; text / javascript & # 39 ;; s.async = true; s.src = & # 39; // munchkin.marketo.net/munchkin.js&#39 ;; s.onreadystatechange = function () { if (this.readyState == & # 39; complete & # 39; || this.readyState == & # 39; loaded & # 39;) { initMunchkin (); } }; s.onload = initMunchkin; document.getElementsByTagName (& # 39; head & # 39;) [0] .appendChild (s); }) (); // END Marketo Munchkin // BEGIN Kewyee (function (w, d) {w.kwa || (w.kwa = function () {(w.kwa.q = w.kwa.q || []). push (arguments);}); = d.createElement (& # 39; script & # 39;), fs = d.scripts [0]; se.src = & # 39; // cdn.keywee.co/dist/analytics.min.js&#39; ; fs.parentNode.insertBefore (se, fs)} (window, document)); kwa (& # 39; initialize & # 39 ;, 480); // END Kewyee .


Source link