15

I read in The Guardian (mirror) (published on 2014-06-26):

Google: 100,000 lives a year lost through fear of data-mining

The search firm's CEO and co-founder, Larry Page, estimates 100,000 lives could be saved next year if mining of healthcare data was acceptable

Are there scientific studies supporting this number?

  • "Could" is the operational word here. As with all data mining, it very much depends on how you apply the intelligence gathered. I am pretty sure insurance companies would love such data to demand premiums from higher-risk patients, or denying them insurance outright... which would be somewhat counterproductive, from the patient's standpoint... – DevSolar Oct 18 '18 at 9:21
  • @DevSolar you could have laws preventing this. – Franck Dernoncourt Oct 18 '18 at 15:09
  • Well, we have. They are what Mr. Page suggests should be removed, making "mining of healthcare data acceptable". If there's one thing we should have learned from the last two or three decades, it's that the ways data could be (ab)used far outpaces the ability of legislation to control such activities. You open the box, the data is out. – DevSolar Oct 18 '18 at 15:40
  • @DevSolar if you rule that data can't be de-identified / linked to an individual, you remove most of the potential misuses. – Franck Dernoncourt Oct 18 '18 at 15:45
  • For context, about 63 million people died last year – De Novo supports GoFundMonica Oct 19 '18 at 6:08
13

Heart disease kills over 17 million people each year. According to a 1991 study, sudden death accounts for 28% of heart disease deaths for men and 14% for women, about 20% of all heart disease deaths. By reducing these figures in half through early detection, it would save many more than 100,000 lives per year, as Larry Page suggests. Can data mining do this?

Researchers Thenmozhi and Deepika write in 2014,

The healthcare industry gathers enormous amount of heart disease data which are not “mined” to discover hidden information for effective decision making.

They propose several methods for predicting a diagnosis of heart disease, ranging in accuracy from 74% for "K-Mean based on MAFIA" to 99.62% based on a 15 attribute "Decision Tree", and 100% for a 15-attribute "Neural Network".

Based on this preliminary finding, Yes: data mining of medical information could save 100,000 lives per year.

Caveat

In a 2005 paper in Journal of Healthcare Information, researchers Koh and Tan of Singapore write:

Data mining applications in healthcare can have tremendous potential and usefulness. However, the success of healthcare data mining hinges on the availability of clean healthcare data. In this respect, it is critical that the healthcare industry consider how data can be better captured, stored, prepared, and mined. Possible directions include the standardization of clinical vocabulary and the sharing of data across organizations to enhance the benefits of healthcare data mining applications.

While Larry Page may see a true benefit to releasing medical data for scientific use, it can be easy for people in tech industries, where everything happens in a fully auto-recordable way on computers, to misjudge the effort required to consistently learn facts in medical interactions and to make those facts into data.

It seems clear that opening up medical data for mining will not be as simple as putting a tracker on user clicks-through for search results like spyware was doing in the early 2000's; the effort will need to be organized and sustained, with full participation by medical professionals and indeed the patients---who will need to answer many more questions and tolerate many more standardized forms each time they see the doc.

  • 6
    Plus, in addition to data quality/collection issues, is Larry Page proposing more data mining of healthcare data by and strictly for healthcare use only? Please. That's not how he got to $50B+ net worth. For him to commercially exploit, you really wouldn't need to fix those issues, so I wonder if he's proposing they allow access and exploitation of data, with the claim that allowing that access will lead to better data quality, eventually. Not a criticism of the answer, by the way. – PoloHoleSet Oct 16 '18 at 17:21
  • "They propose several methods for predicting a diagnosis of heart disease...". There are a lot of steps inbetween being able to predict outcomes in a test data set and saving lives. About on the scale of difference between Wright Brother's first flight and getting to the moon. – Cliff AB Oct 29 '18 at 17:30
  • @CliffAB, I'm suggesting that sudden death will not happen if the disease has been predicted. Sudden death is the population of deaths we're proposing to reduce. – elliot svensson Oct 29 '18 at 17:35

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .