Release date: 2016-09-12
A 60-year-old aunt lay in the ward of the Affiliated Hospital of Tokyo Medical University. She has cancer and has been treated for several months, but her condition has not improved. This makes the doctors feel a little helpless.
In desperation, the medical team thought of using IBM's Watson smart program to help - this is a supercomputer that defeated the human champion in the quiz show "Dangerous Situation". The doctors entered the patient's condition information into Watson and searched it in the tumor database. It was found to be a rare secondary leukemia. The medical team changed the treatment plan accordingly. It didn't take long for the patient to be discharged from hospital.
A doctor told the reporter of the Japan Times that Watson could do it in a few minutes, and it took a few weeks to change the doctor for diagnosis. "It would be a bit exaggerated to say that artificial intelligence saved the patient's life, but it did give us the data we wanted."
Is this the future of smart healthcare? Let's take a look at what the researchers can dream of with artificial intelligence machines:
Can diagnose human health
Advising the doctor about the treatment plan
It can even predict how the patient's health will change.
The biggest advantage is not the speed, but the accuracy.
A study published earlier this year found that medical malpractice was the third leading cause of death in the United States. Among them, a considerable number of people died of misdiagnosis.
Herbert Chase works at Columbia University's School of Biomedical Information in New York. He said:
People's health is too diverse, and the papers are updated very quickly, and it is difficult for a primary care physician to write them all down.
The machines we designed have been able to do what doctors can't do. A machine can diagnose dozens of conditions that doctors can't diagnose.
Chase has proposed the establishment of an IBM Watson team. Now, he is designing an algorithm that can find clues from the doctor's notes that may develop into multi-function sclerosis, and eventually establish a program to calculate how much risk each person has with multi-function sclerosis. He envisions that future software can automatically alert everyone's electronic health data to issue warnings or provide advice.
“The machine gives advice, people refer to the advice to make action, this is a partnership relationship.†But given the variety of human diseases, “algorithms need to be built step by step.â€
Other achievements and existing problems
Recently, a research team from Stanford University has released a new machine learning algorithm that can be used to examine lung tissue slides infected with cancer cells. Using this technique, the computer can display the individual characteristics of each slide, such as cell size, shape, structure, and so on. It also distinguishes from the sample the length of time a person has survived a medical diagnosis—in just a few months, or longer. By testing historical data, the algorithm proved to be fruitful. Therefore, in principle, artificial intelligence can be applied to diagnose human health.
Slide readers are just one of many medical artificial intelligence tools.
Last week, the Machine Learning and Healthcare Conference was held in Los Angeles. At the conference, researchers showed people new algorithms: they can detect epilepsy, predict the development of kidney disease and heart disease, and discover abnormal physical conditions in pregnant women and newborns. In a programming competition, participants used their own artificial intelligence to listen to the heartbeat frequency to distinguish between normal rhythm and abnormal rhythm.
However, the information resources used by other medical diagnostic projects are vague and indirect:
Microsoft released a new algorithm in July to guess who has pancreatic cancer through a web search.
Google's Deepmind uses a lot of anonymous data from the National Health Service to design new artificial intelligence to solve severe eye problems faster. However, the project has triggered a new problem - the price of commercial companies buying health data is not too low.
Although artificial intelligence diagnosis is of great help to doctors, the question is, are medical experts willing to use this new method? At present, people need more evidence to prove that computer prediction can really help people improve their health.
Some people worry that the development of artificial intelligence diagnosis technology will make doctors over-diagnose and over-test, and the result will be counterproductive. Even if the algorithm works properly, the problem of how to combine it with clinical practice still exists. Doctors are already tired of work, and they don't want artificial intelligence to add to their workload.
expert's point
Chase believes that artificial intelligence should eventually be combined with electronic health data records, so seeking treatment advice from the machine is just as much a part of everyday work as getting patient-related data.
“Doctors want artificial intelligence to help them get things done, which means they have to admit that they make mistakes occasionally.â€
In fact, there are apps on the market that can help with diagnosis, such as Isabel, which provides doctors with a networked list of accurate diagnoses after they enter symptoms and test results to prevent doctors from ignoring possible rare diseases and misdiagnosis. But Chase said that this method is not yet popular. The reason is that artificial intelligence can be successful in this field only if it does not put any pressure on the doctor.
Leo Anthony Celi is a doctor in the intensive care unit of the Beth Israel Deaconess Medical Center. He said that artificial intelligence is widely used in the medical field, and there is still some social resistance. With the help of artificial intelligence, the doctor will be more like a "captain." Most of the day-to-day work is done by the machine or by a trained nurse, medical electronics or physician assistant. In order to achieve this kind of work, the doctor must first take a step back and admit that the machine is indeed better than others in some areas. This is very difficult. Because both medical professors and patients, everyone wants the doctor to always give the most correct answer.
Celi said that people's ideas need to be changed, and the potential of big data and artificial intelligence in the medical field should be more respected. Only in this way can machines and human beings develop their strengths and do their best.
“The communication skills of doctors and patients are irreplaceable. Doctors should focus more on what they can do better, for example, communicating with patients and guiding them to express their values ​​and advance directives. As for complex decisions, We are handing it over to the machine, and we are really not good at it."
Source: Internet
YT-S350
YT-S350
Shenzhen Sunshine Technology Co.,Ltd , https://www.yatwinsz.com