Can Big Data Analysis Help Save Sick Lives?
Today, big data is in trend and an equally big favorite. Recently, Larry Page also noted in this area, who said that if there was more public information about the state of health in the public domain, then thanks to its analysis next year, it would be possible to save about 100,000 people. After the US National Security Agency, Google ranks second in terms of the amount of data stored. However, Page probably hurried up a bit with his statement, especially in light of the fact that the large Google Flu Trends program ( official website ) showed low efficiency . Big data is not a magic tool that can solve all our problems, and it is unlikely that Page will be able to save thousands of lives with their help.
This is hindered by several factors. The very idea of analyzing big data for the sake of preserving the health and life of people is very ambitious, but in no case can this be considered a simple task. Any question in which personal data of people is involved requires a very thoughtful scheme to ensure access to information from third parties. Providing big data for analysis requires mandatory depersonalization so that it is not possible to compare information with specific people.
Globalization and the comprehensive comprehensive collection of information about the population, in fact, have become almost taken for granted. Many public and private organizations accumulate gigantic amounts of information about users every day, including their behavior on the network. And the development of mechanisms for regulating the circulation of such information lies with state organizations. And the inevitable restrictions on the dissemination of personal data will be an obstacle in the good deed of the timely detection of diseases on the basis of medical indicators.
Machine intelligence as an analysis tool
With his statement, Larry Page illustrated the false view of big data that is characteristic of the powers that be. From his words we can conclude that it is precisely the lack of public information that leads to the death of people as a result of certain diseases. But these are just emotions. It is likely that some of the deaths in the health system can be prevented by providing the right information to the right people. However, it must be remembered that the discovery of such information implies providing access to large databases for various organizations. At the same time, there are no arguments in favor of the fact that the provision of access in itself will provide a “meeting” of the necessary information with the right people.
Disclosure of medical data may be useful, but, unfortunately, there is no reason to believe that automated analysis of big data alone will help reduce mortality for medical reasons. Speculation on this topic refers to emotions, not to reason. Without a scientific theoretical basis and further practical steps, such an analysis becomes meaningless.
Unfortunately, we are far from omnipotent, and our computer analysis tools are far from perfect. Even 60 years after the death of Alan Thuring, his test remains inaccessible to all applicants for the proud title of “artificial intelligence”. Even the last high-profile attempt involving the program " Eugene Gusman", Imitating a 13-year-old teenager, turned into a failure. The tactics laid down by the author of the program in his brainchild are similar to those used 42 years ago in the Parry program , which “pretended” to be a paranoid schizophrenic. You can also recall the Eliza program , which was difficult to distinguish in communication style from a doctor who adheres to Rogerian theory ( Karl Rogers Theory ).
For many decades, work has been relentlessly underway to create artificial intelligence, or rather, a trained machine. Many of the brightest minds participated in such developments. Computing power until recently has grown exponentially, and the worldwide network has given many examples of the interaction between people on which you can train a computer. And despite all this, scanty progress proves how difficult it is to transform large amounts of data into the likeness of human intelligence.
Therefore, returning to Page, it is better to avoid making such loud statements about the potential of big data. The Google Flu Trends project tried to get information about the spread of influenza by collecting data on cases where people used search queries with the word "flu." But again and again we were faced with the fact that computers are not able to understand people and reliably imitate the features of our behavior.
As an example confirming this thesis, one can cite the fact that by September 11, 2001, the NSA had enough intelligence information to prevent a catastrophe. But this organization simply could not connect all the parts of the mosaic in time . Edward Snowden's revelations confirmed suspicions that the NSA and the Center for Government Communications (British intelligence ) constantly collect a variety of information about the citizens of many countries. And not only them. Intelligence services have repeatedly stated that the analysis of this big data more than once allowed preventing serious terrorist attacks, but all these statements do not withstand criticism upon closer examination. Given the computing power available by the NSA, they would have to process information collected before the Internet era with artificial intelligence for 30 years. And there is no evidence that such studies yielded specific results. At the very least, such information is not publicly available.
Alas, the situation is such that the existing beginnings of the technology of “artificial intelligence” will not allow saving anyone's life by analyzing big data from the healthcare sector. Correct correlation of information requires the human ability to make connections, and computers have not yet acquired this ability, despite many years of efforts and financial injections from many corporations.
The diagnosis is not a simple comparison of the test results and the current state of the body. To do this, doctors must ask patients the right questions and make decisions that affect their lives. Therefore, the analysis of big data in the field of medicine will only make sense when the computer can independently “ask” the necessary questions and find answers to them in the information arrays.