Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
Medicine Microsoft Communications Network Networking Software The Internet News Science Technology

Microsoft Analyzes Web Searches, Finds Clues For Early Cancer Detection (computerworld.com) 73

An anonymous reader quotes a report from Computerworld: Analyzing online activities can provide clues as to a person's chances of having cancer, Microsoft researchers showed in a paper published this week. Specifically, the researchers demonstrated that by analyzing web query logs they were able to identify internet users who had pancreatic cancer even before they'd been diagnosed. The study suggest that "low-cost, high-coverage surveillance systems" can be created to passively observe search behavior and to provide early warning for pancreatic cancer, and with extension of the methodology, for other challenging cancers," the researchers concluded. "Surveillance systems could also provide for automated capture and summarization of data and landmarks over time so as to provide patients with talking points in their discussion with medical professionals." The researchers used proprietary logs of 9.2 million web queries on Microsoft's own Bing search engine but focused exclusively on English-speaking people in the U.S. from October 2013 to May 2015. First, the team identified searchers in logs of online search activity who made "special queries" that are suggestive of a recent diagnosis of pancreatic cancer. Those queries included phrases such as "Why did I get cancer in pancreas," and "I was told I have pancreatic cancer, what to expect." The team then went back "many months" before the initial queries were made to examine patterns of symptoms as they were expressed by web searches about pancreatic cancer symptoms. "We showed specifically that we can identify 5% to 15% of cases, while preserving extremely low false-positive rates," the researchers said in their paper. The false positives ranged from one in 10,000 to one in 100,000.
This discussion has been archived. No new comments can be posted.

Microsoft Analyzes Web Searches, Finds Clues For Early Cancer Detection

Comments Filter:
  • Newsflash (Score:4, Insightful)

    by bengoerz ( 581218 ) on Friday June 10, 2016 @08:25PM (#52292691)
    Researchers have very low false-positive rate when analyzing past data.
    • "analyzing past data" - I've had great success in creating convincing horse race and stock picks using that method!

      Unfortunately, "Past performance is no guarantee of future results" is VERY true.
      • by AK Marc ( 707885 )
        That's simply not true. You take 100% of the data, run the analysis on 50%, then, when done, apply that analysis on the remaining 50% and measure your success. If successful, it should apply to any data source. Even future ones.
        • by ceoyoyo ( 59147 )

          Yeah, except that a lot of people don't bother doing that. Particularly physicians.

          • by AK Marc ( 707885 )
            This was big-date people doing medical research, not physicians, so hopefully they did it right.
  • by NotInHere ( 3654617 ) on Friday June 10, 2016 @08:27PM (#52292701)

    They just scan the user agent of the browser connecting, and if it contains "Linux" it means there is a cancer infection in the eyes of Microsoft.

  • by Anonymous Coward

    THAT'S what they'd use it for.
    To protect us from Cancer.

  • by Anonymous Coward

    They looked for people searching for "how to cope with stage 4 cancer" and sure enough, those people had it in most cases. How DO they do it?

    • Re:Wow!! (Score:5, Informative)

      by murdocj ( 543661 ) on Friday June 10, 2016 @08:35PM (#52292733)

      Actually, if you had patience to read the summary, they did something very clever. They did find the people who had been actually diagnosed with cancer. Then they went back months to their previous searches, and found that BEFORE they had any idea they had cancer, they were searching for information on their symptoms... symptoms of the cancer that would be discovered much later. How much would it be worth to you to find out you have cancer when it can be treated, rather than too late?

      • by Dunbal ( 464142 ) *
        And these non medically trained researchers are sure that these symptoms are specific to pancreatic cancer because after all abdominal pain, weight loss and vomiting could hardly be anything else right?
        • by Anonymous Coward

          Yes, if you read the fucking summary, you will see that they state, "The false positives ranged from one in 10,000 to one in 100,000."

          So yeah, the researchers are pretty certain the symptoms and terms they're analyzing for are pretty specific to pancreatic cancer.

          What would you rather have: No warning that you may have cancer until you're in the hospital shitting your guts out a straw? Or maybe, a warning 6 months earlier to get checked out by your doctor - when the cancer may be treatable, and you can av

          • "The false positives ranged from one in 10,000 to one in 100,000."

            If they had a false positive rate of 1% or even 10%, that would be amazingly good. Their claimed rate of 0.01% to 0.001% is completely implausible. It may just be a case of incompetent journalism, but if the researchers actually claimed those rates, I don't believe them, and I question their integrity.

            • Re: (Score:2, Informative)

              by Anonymous Coward

              > If they had a false positive rate of 1% or even 10%, that would be amazingly good.

              If it was a sensitive test, that'd be true.

              But it's not a sensitive test. It spots a relatively small percentage of pancreatic cancer cases early, while getting a relatively small number of false positives.

              Diagnostic and screening tests trade off sensitivity (detecting cases) and specificity (positive tests indicating the specific condition). This test has a very low sensitivity in order to attain its high specificity.

            • by AK Marc ( 707885 )
              So a false negative rate of 85% to 95% is so high that you find they must be lying. You have unreasonably high standards. They tuned to minimize false positives, at the expense of false negatives. They could get the false positive rate you demand, and greatly improve the false negative rate. And, then with the same data and analysis, meet your false positive rate requirements.
            • But if very few people get pancreatic cancer, then with a false positive rate of that magnitude, it could still be that the majority of detections are false positives.

          • by Dunbal ( 464142 ) *
            This only means that one in 10,000 turned out to have pancreatic cancer without querying for the symptoms. That is what false positive means. I am pretty sure they didn't follow up on every single patient and rule out say acute diarrhea.
            • 1/10000 of people who searched for those symptoms didn't have pancreatic cancer. Around 90% of people who did have pancreatic cancer didn't search for those symptoms.

        • by AK Marc ( 707885 )
          Nope. The researchers don't know or care about "symptoms". If everyone with lung cancer googles for "left handed mouse" and "pineapple and red bean ice cream" 7 months before being diagnosed with cancer. It's a correlation, not a diagnosis, and not "symptoms". If a correlation is strong, it's useful. Though, as discussed, the better correlations happen among more logical things, like possible cancer symptoms.
      • Actually, if you had patience to read the summary, they did something very clever. They did find the people who had been actually diagnosed with cancer. Then they went back months to their previous searches, and found that BEFORE they had any idea they had cancer, they were searching for information on their symptoms... symptoms of the cancer that would be discovered much later. How much would it be worth to you to find out you have cancer when it can be treated, rather than too late?

        If you have any symptoms from Pancreatic cancer, its already too late.

        • If you have any symptoms from Pancreatic cancer, it's already too late.

          Yes, a diagnosis of pancreatic cancer is usually a death sentence, but many people would still appreciate a few extra months to wrap up life issues, reach out to old friends, maybe arrange one last family reunion, and knock a few items off their list of life goals. Earlier diagnoses are good, even if they don't provide a cure.

          • If you have any symptoms from Pancreatic cancer, it's already too late.

            Yes, a diagnosis of pancreatic cancer is usually a death sentence, but many people would still appreciate a few extra months to wrap up life issues, reach out to old friends, maybe arrange one last family reunion, and knock a few items off their list of life goals. Earlier diagnoses are good, even if they don't provide a cure.

            My Mother in law died from Pancreatic cancer. Funny, but she didn't feel much like doing a bucket list. Kinda laid around, then died. Here's an even better approach. We all die. Seems like a bitched up idea that you wait until you have a couple months left to do the things you suggest. Do them now, when it isn't so depressing.

            • by AK Marc ( 707885 )
              When you ignore the symptoms until you are too sick to do anything, you don't do a bucket list. Because the health care system in the US is so bad, many people in the US wait until they are too sick to have any options. I know more than one person who was having deadly symptoms before the first time they sought care.
              • When you ignore the symptoms until you are too sick to do anything, you don't do a bucket list. Because the health care system in the US is so bad, many people in the US wait until they are too sick to have any options. I know more than one person who was having deadly symptoms before the first time they sought care.

                Most times, Pancreatic cancer does not have symptoms until it is too late. That is why I already wrote that if you have symptoms, and it is pancreatic cancer, your outlook is nil.

                How many family members you have die from it? My Stepmom - in law first showed some diabetes like symptoms and went to the doctor in a couple days. She was a RN.

                Pancreatic Cancer does not do it's work because of the US health care system or the expense of treatment. Unless caught pretty much by luck, you are out of luck. I

        • Unless you have symptoms that are just in common with pancreatic cancer symptoms, in which case they are just symptoms of something else. The study has a biased sample, since they where studying the history of "known cancer" patients (or someone who lies to search engines... for example, I just now searched "I think I have pancreatic cancer"... Shit! - oh, wait, I just searched for "medical student syndrome", too! Whew! That's what I have.)
  • by Anonymous Coward

    And your search engine mining your search history to figure out which diseases you might have is not creepy at all! No sir.

    And surely there is nothing else that could be done with this technique.

    I'd suggest this one instead. [duckduckgo.com]

  • STOP watching/tracking what I'm doing with my computer -- it's creepy as fuck!

    (Switched to Mac years ago, not going back to windows and it's "telemetry". Not so sure of Apple either, but at least they claim to not track users and collect data on them -- their business model (currently) is selling shiny toys -- not selling data. And they have been pushing back on surveillance in the courts and their encryption is good. Filevault should be on by default.)

  • cancer

    or

    dissident ideology

    which one do you think is likely to be acted on to promote stability for the economy from which microsoft profits?

    Seriously us people who aren't herd animals need to get it together. There is clearly a massive conspiracy against us all. Are you going to wait around to see what the top of the hierarchy does with all of this information they have on us once the economy becomes majorly automated?

  • Clippy: It looks like you may have cancer! Would you like me to schedule an appointment with an oncologist?

    • ... It looks like you may have cancer; your insurance company & employer have been notified. Have a nice day !!!
      • ... It looks like you may have cancer; your insurance company & employer have been notified.

        The ACA has some issues, but one of the things it does right is make it illegal for your insurance to dump you for things like cancer, and make it illegal for an insurance company to refuse coverage if you have a pre-existing condition.

        Okay, TWO of the things it does right.

  • Just what I need, a pop up window that says "CONGRATULATIONS YOU'RE THE 1 MILLIONTH VISITOR WITH PANCREATIC CANCER!"

  • This is where a good doctor comes in. If you are vomiting and have abdominal pains and DO NOT go to the Doctor then you should have no expectations of survival once this fast-moving Cancer is revealed. A good Doc will run tests when presented with these symptoms and hopefully catch it early. A good Doc with EXP will know what's up when presented with your symptoms.

    Do yourself a favor, stop searching on WebMD, Google and Bing. Visit your Doctor instead.

  • And how many of their research subjects had been diagnosed with hypochondria? Searching for symptoms and eventual disease isn't unlikely pattern, whereas someone actually suffering from it would be more likely to only ask a doctor. Didn't bother to read the article, of course, but hopefully they did also check whether they did search indicating diagnosis also before, and possibly for other diseases.

    I also have to join those questioning the "false positive" rate there. People are perhaps even more liable to

  • by Anonymous Coward

    I wonder if my search history would have shown I was likely to have it. My symptoms were so vague that I waited about 6 weeks before going to the doctor - general discomfort in the abdominal area (2 different spots that were separated), a little more gas and diarrhea than normal, possibly slight weight loss, maybe a bit more tired than normal. The doctor ordered tests that showed gall stones and positive for H. Pylori (main cause of ulcers). She put me on antibiotics for the H. Pylori but my impression

  • What I took away from this is that people who use Bing get cancer.

  • One in ten thousand is a lot of erroneous diagnoses when you're doing web search scraping.

Sigmund Freud is alleged to have said that in the last analysis the entire field of psychology may reduce to biological electrochemistry.

Working...