"Medical Revolution" is steadily progressing, such as medical information on cancer treatment organized by "machine learning" and used for treatment


Iteratively learn from a large amount of data and find a pattern "Machine learning"Is beginning to be used in various fields. Researchers at MIT have begun to use machinery learning in medical fields where there is no relation to machine learning, thereby making it useful for treatment.

Putting data in the hands of doctors | MIT News

Professor Regina Berryley who studies electrical engineering and computer science at MIT was diagnosed with breast cancer in 2014. Professor Bailey was trying to find information useful for treating her own illness but was unable to find information that could be confident that "it can be trusted based on data and heuristics", "guessed He said that he had to rely on. He seems to realize that information in the medical field is not overwhelming enough.

Although it is a machine learning technology which became common now in computer science, it is said that it is unclear state in the medical field. The doctor writes the information obtained by examining the patient manually and examines the correlation based on the basic statistical analysis. However, such a method seems to be a primitive method when it is said to Professor Berylley.

According to the American Society of Clinical Oncology, 1.7 million people are diagnosed as cancer each year in the United States, but only 3% of the total is registered in the clinical trial. In other words, in the current medical research, it depends only on data obtained from only a few% of patients. Professor Bergeley said, "We need to analyze information on the treatment of the remaining 97% of patients undergoing cancer treatment."

Professor Bergeley thinks that the possibility of a new medical treatment method and better medical practice is buried in medical information which does not appear in the world, and we are starting work of digging medical data with the student's student. First of all, in cooperation with the doctor of mammary tumor theory at Massachusetts General Hospital, he is working on machine learning work on cancer treatment.

Professor Bergeley 's research team extracts clinical information from a pathology report on as many as 108,000 cancer treatments using a natural language processing tool to create a database. The precision of database creation is very high, 98%, and we are doing a huge amount of work that can not be done manually by using computers. In addition, we try to make a neural model that can infer from machine learning by applying machineized learning of databaseized information.

In addition, Professor Bailey is trying to apply machine learning to preventive medicine. It is trying to find subtle signs leading to breast cancer by applying deep learning technology to mammography that contains a lot of information that is hard to decipher with human eyes. I hope to find early breast cancer patients that can not be judged by doctors' eyes and use them to predict which patients are likely to develop cancer recurrence.

Professor Bergeley thinks that data scientists often can cooperate with physicians and biologists, and by thinking that by providing various data analysis techniques such as machine learning, medical care will be developed and many patients will benefit. It is.

in Software,   Science, Posted by darkhorse_log