Robot doctor Watson How to change the world?

Watson is the artificial intelligence system developed by the DEEPQA planning team, led by IBM's lead researcher David Ferrucci, since 2007, and it is the culmination of more than 20 IBM researchers ' four years of effort, with IBM founder Thomas? J. Watson's surname is named. Thanks to the efforts of scientists, Watson has the ability to understand natural language and answer questions accurately.

In 2011, Watson beat the top prize winner Brad Ruttel and the winning streak record holder, Ken Jennings, into the eyes of a veteran American variety show "on the brink of crisis".

According to IBM's data, Watson's hardware is a clustered server of 90 IBM Power 750 servers, with a total of 2880 Power7 processors and 16TB of memory. On the software side, Watson is written in Java and C + + and uses the Apache Hadoop framework for distributed computing, as well as the Apache UIMA (unstructured Information Management architecture) framework, IBM Deepqa software and SuSE Linux Enterprise Server 11 operating system.

Why is Watson so clever?

How could Watson have been so clever in defeating humans in the race? Watson stores millions of of documents, including dictionaries, encyclopedias, journalism, literature, and other reference materials that can be used to build a knowledge base. Watson's hardware configuration allows it to process 500GB of data per second, equivalent to 1 seconds to read 1 million books.

Watson, after getting a problem, will perform a series of calculations, including syntactic semantic analysis, searching the various repositories, extracting alternative answers, searching for alternative answers, calculating and synthesizing evidence, and so on, according to Dr. Lei of the IBM China Institute. Watson's main technical principle is to search a lot of knowledge sources, from many angles using a lot of small algorithms, the various possible answers for comprehensive judgment and learning. One of the key steps is to evaluate the reliability of alternative answers, and to examine and quantify the reliability of the answers from a number of dimensions, such as keywords, geographical location, type, etc.

Watson's journey?

In 2011, IBM said that, given that "Watson" understood human language, it was possible to quickly give diagnostic tips and treatment advice by asking about the patient's symptoms and medical history, by using artificial intelligence techniques, natural language processing and analysis techniques, and by relying on information and data collected from various sources. After that, the AP journalist personally realized how robot doctor Watson diagnosed the patient, researchers provided "Watson" with a virtual eye disease patient, blurred vision, a family history of arthritis, living in Connecticut State, pregnant ... "Watson" presented different diagnostic results, including uveitis, Behcet's disease, Lyme arthritis. It is estimated that Watson's diagnostic accuracy rate reached 73%.

In 2011, the American Health service provider, WellPoint, signed an agreement with IBM, the first job that Watson received. Watson's main task is to help WellPoint nurses who are responsible for complex cases to complete their work while examining medical requests from medical providers. And the later application in the clinical trial of cancer. Wellpoint also said in the future, Watson may also be able to obtain patient records and other information, and then integrated feedback to the doctor to improve the doctor's diagnostic speed.

In 2014, the University of Texas Anderson Cancer Center worked with IBM to build the "Moon Landing Project", which uses IBM Watson technology to eliminate cancer. Anderson Cancer Center is America's first and globally recognised cancer hospital, and the center's oncology expert Advisor (Oncology Expert Advisor) is driven by Watson's cognitive computing system to integrate the knowledge of clinicians and researchers at Anderson's Cancer Center. The OEA system will help clinicians develop, observe and adjust treatment options for cancer patients. IBM Watson technology will also help simplify and standardize the collection and integration of patient records, laboratory data, and research data, enabling the collection of data to be integrated into the patient database at Anderson Cancer Center, and then linking can be deeply analyzed by advanced analysis techniques.

In addition to the Anderson Cancer Center, the Mayo Clinic (Mayo Clinic) is also experimenting with the concept of IBM Watson to provide patients with appropriate clinical trials faster and more efficiently. At any given time, the Mayo Clinic was able to carry out more than 8000 human research trials. But because there are not enough people involved, many clinical trials are not done, either at the Mayo Clinic or elsewhere. IBM and Mayo Clinics are expanding Watson's knowledge corpus to include the Mayo Clinic and clinicaltrials.gov common databases, while training the knowledge base to analyze patient records and clinical trial conditions to provide a suitable match.

The Kang International Hospital in Thailand uses IBM Watson's cognitive computing to improve the quality of cancer treatment in the Bangkok Research Centre and to conduct case assessments in 16 countries. The hospital is committed to using Watson for Oncology technology developed in conjunction with the Caitlin Cancer Center (MSK) over the next 5 years. The system will help physicians develop effective treatment options for cancer patients using medical evidence, academic research, MSK's extensive clinical techniques and the records of each patient.

In addition, the Sloan-Caitlin Hospital in New York and the Cleveland Clinic also have business cooperation with Watson.

On the corporate side, Johnson & Johnson uses the results to develop and evaluate drug regimens and other treatments by providing IBM Watson's reading and understanding of research papers detailing the results of clinical trials. With this knowledge, Watson Discovery Advisor can help scientists determine whether there are any bad genetic files for drug samples. Prior to the comparative study, 3 individuals were required to spend an average of 10 months to collect the preparation data before they could begin the analysis. The Johnson & Johnson team wants to be able to quickly synthesize useful information directly from the medical literature by Watson and start asking questions directly from the data.

According to the arterial network business Circle friend Gamp revealed that the IBM United States this month just restructured, the company divided into 7 major departments, one of which is the medical department, the department is Watson as the core.

The development of artificial intelligence in medical profession

In fact, before Watson, the relevant medical institutions were developing the relevant "computer Doctor" project.

In the 70 's, researchers at Pittsburg University in the United States developed software "quick medical references" for diagnosing complex illnesses in general medicine, which collected 4300 clinical representations, capable of diagnosing more than 600 diseases, and improved the likelihood of rapid diagnosis through systematic procedures.

1972, Stanford University in the United States began to develop the mycin system, 74 basic completion and put into use. Mycin is the suffix of many antibiotic drug names, so this is an expert system to help doctors diagnose hospitalized patients with blood infection and use antibiotics to treat them, which is still quite representative.

In the 80 's, Massachusetts General Hospital began to develop and improve the Dxplan project. Dxplan covers a wide range of areas of knowledge, including most of the diseases and clinical manifestations of internal medicine, with IBM's personal computer as a program development tool. Users can consult the computer on the next Test and test to get the most information at the lowest cost.

In the fall of 2010, the Isabel health care system was used by the Orlando Health Care Hospital in Florida to provide doctors with reliable diagnosis and treatment advice, and some doctors with less experience and fewer clinical practices were able to get more help from the system. However, Isabel can only be used by networking with the multi-purpose hospital health system, which is slow and expensive.

By 2013, computer Doctor Watson had boarded the medical arena, which was both a cancer diagnostic expert and a professional in the management of medical services. Since then, the computer-aided diagnosis opened a new page, formally entered the "Watson era."

The U.S. health care industry will spend $69 billion trillion in the information industry over the next 6 years, according to research firm Insight. Sources say Intel and SAP have started working with researchers at the University of California, Berkeley, to develop competitive medical supercomputers.

Is the computer doctor really reliable?

Watson, as a computer doctor, has an innate advantage-its knowledge stores are far more than human beings and will never be forgotten, diagnosed with high accuracy and can be kept on call without rest, although it seems encouraging, but there are a variety of questioning voices. "Watson may be a smart companion for us," said Abraham Vignesse, a New York Times writer and a Stanford doctor, "But what I hear from patients and relatives and friends is not the lack of technology, but the spread of technology." "The Watson Project, Marty Cohen, thinks that Watson is just an aid, and that if the doctor is unwilling to change, then Watson cannot change the medical profession." "Some of these technologies do change medical care, they offer treatments that were not there before, but it's not, and I think it's just a catalyst," he said. "Klaus-Peter Adrason is a computer scientist at the Medical University of Vienna and editor-in-chief of the Journal of Artificial Intelligence Medicine". In his view, a computer doctor like Watson, in essence, is actually a search engine that can answer various questions raised in natural language. As time goes on, computers will be able to sum up their mistakes, but the knowledge gained from medical literature and case studies is very empty and the knowledge is not necessarily very valuable in the clinical environment for the medical staff.

Practicing medicine is far from handling the data so simple, the patient and the family member's emotional comfort, grasps the subtle difference in the practice, the study grasps the uncertainty, no matter which point, all can not leave the human doctor. The most extensive medical data and the most cutting-edge processing skills do not teach a computer how to love a patient as a human doctor does. 1

Watson is not omnipotent, although it now has a reputation outside, but it is only a doctor assistant role, computer Doctor's vision still need to walk a long way. Perhaps in the future, human-machine collaboration is the final direction of development.

(Responsible editor: Mengyishan)