Medical AI may be more reliable than medical AI
Imagine that you go to a hospital to see a doctor. Once you enter the door of the clinic, a nurse will take pictures of you constantly. Then these photos will be uploaded to an AI device. The device will be based on your appearance in the photo. Diagnosis... And no professional human doctor will appear throughout the process.
Is it incredible? Even though AI medical development is very fast now, some AIs have reached different levels in the medical field, such as AI recognition medical imaging, drug development, and auxiliary diagnosis. However, these AIs are basically auxiliary, and ultimately responsible for decision-making. It is human. It seems that AI has been certified to work on a clinical basis and has never seen it before.
However, such a "care model" may have begun.
Recently, the US Food and Drug Administration (FDA) approved for the first time an artificial intelligence diagnostic device, IDx-DR, which can detect an ophthalmic disease by observing photos of the retina without the involvement of an expert doctor.
In other words, this AI device called IDx-DR actually got a job certificate and became a real "doctor"!
Scientists continue to break through one technical obstacle after another. While we are happy, we also have hidden concerns. The road to medical AI is getting smoother and smoother, but now it is appropriate to have an independent AI doctor.
AI doctors want to be "independent" and lack a complete industrial chain
The intelligent theory of relativity (aixdlun) analyst Yan Wei believes that if we want AI doctors to be independent, we must thoroughly study the industrial layout and the coordination and symbiotic development of each link of each industrial chain from the beginning. Otherwise, as long as there is a link in it. Poor development will lead to the breakdown of the structure of intelligent medical care between upstream and downstream, or by technical ethical issues.
1. There is no standard for the AI ​​doctor's landing.
From the perspective of the patient or other medical use, medical AI does not have a particularly large change in a short period of time. Because the certificate is not approved, and how to approve the certificate for an AI doctor is also vague. What is the standard for AI doctors to pass? Is the precision of the device or the correct rate of diagnosis? Even with the FDA-approved IDx-Dr, in a clinical trial using more than 900 images, the correct rate of detection of retinopathy was only 87%.
In the final analysis, whether AI doctors can land is not to think about whether the machine is a philosophical question, or whether the administration is authorized. In the medical field, the landing of a product must include licenses, medical rigor, and the appointment of an independent AI doctor, which may have a long way to go.
2. "After-sales" service is not easy to handle
In real life, the patient has encountered a doctor misdiagnosis, can ask the hospital to compensate or dispose of the doctor; the doctor will treat you, and it is almost possible to cure 3 in 5 cases. However, AI gives you a medical treatment. It may be wrong in 100. What would the only one who was mistaken? If you encounter a doctor who is not good enough, you can still laugh at the phrase "I don't know how to meet someone." If you encounter a misdiagnosed AI, I am afraid it is not so tolerant.
First of all, it is indispensable to chase the hospitals and manufacturers. Then, what should I do? The "culprit" AI has not been dealt with anything.
Is this AI doctor destroyed? Or format the "mind" of this AI doctor to show punishment? However, it seems inappropriate to erase it and see the merits of 99 patients. Moreover, in medicine, with the accumulation of electronic medical records and digital film, a large number of structured cases are used for machine learning. For AI doctors, the big data of this model training is at least 100,000 copies.
AI is the result of a collective delivery. From the development of programs and algorithms to mechanical installation, the cost of disposing of an AI doctor is difficult to calculate. Assume that the patient's family is angry and angry, and I am afraid that I will receive an expensive bill. Therefore, if the AI ​​fails, the powers and responsibilities can be borne by the hospital and the company, but what should be done to calm the patient's anger for this "mistaken" machine?
3. AI doctor turns "patient" into "consumer"
The medical industry has a feature: core services are provided by a single professional technician. When we go to the hospital to see a doctor, we care about which doctor visits, and will compare the doctor's word of mouth. The service location and the institution will affect our evaluation to a large extent. The hospital brand will guide patients' decision-making. Doctors in the top three hospitals and doctors in the dimethyl hospitals are more likely to choose who is the primary doctor?
Once you have an autonomous AI doctor, AI no longer uses high-tech medical equipment as a gimmick for hospitals and a tool to improve the efficiency of doctors. Instead, it enters an independent diagnosis and becomes a "professional." Although the core services are still provided by a single professional technician, the service location and medical institutions seem to be less important, and the innovator's voice will be even greater. The company that leads the patient's medical treatment will be the company that produces the AI ​​doctor. In this way, the patient's identity will be closer to the "consumer." After entering the consumer, the doctor-patient relationship will become cold.
Medical AI may be more reliable than medical AI
In fact, medical AI may be more in line with the development of today's society than smart medical care. We must make clear that medicine and medical care are actually two concepts, medicine is science, and medicine is a practice technology based on medical science.
Therefore, in the future, new technologies such as artificial intelligence and big data will drive the advancement of the entire medical science. However, medical care, especially independent AI doctors, involve “first-line†operations, and their lives may not be as fast as we think. Because in any case, medical care should be based on safety, maturity and stability.
For science, high quality data is a valuable engine for development. The medical health industry and the pharmaceutical R&D industry are a gold mine of data. Medical AI is a good research direction and development direction. More research fields will be quick and fruitful, and these will have better business opportunities.
According to McKinsey estimates, big data and machine learning in pharmaceuticals and medicine can generate up to $100 billion in value each year. These values ​​come from: better decision making, optimizing innovation, improving the efficiency of research/clinical trials, and creating new treatments for doctors, patients, and healthcare organizations.
For example, the oncology department of the IBM Watson Research Center and the Sloan Kettering Memorial Hospital in the study of personalized medicine, they are committed to using the patient's medical information and medical history to select the optimal treatment. Other than that. There are also many companies that are also working on such products.
Source: Prospective Economist
Former Google CEO Schmidt once said, "Computers can really play a role in analyzing a lot of useful information, such as predicting the outcome of a disease. But if you are sick, you still want to be responsible for a person, you will see a doctor, But doctors need to master the latest medical technology to help make decisions."
In essence, the concept of artificial intelligence is to enhance human intelligence. Just as steam engines save human energy, telephones strengthen the connection between humans, computers strengthen human computing power, and machine assistance does not replace human activities. It only expands people's skills and professionalism.
Therefore, in terms of medical AI, an independent AI doctor may not be able to match a virtual assistant who plays a supporting role. If the innovation of smart medical care is on the side of the doctor and empowers the doctor, the probability of successful innovation will increase greatly. Because this can bring real benefits to doctors, doctors will also actively cooperate. However, if the AI ​​doctor is placed in the hospital, the doctor will only feel that his occupation is threatened, and the initiative may not be high.
Image source: Prospective Economist
From the current trend, I am afraid that not many companies want to create an independent robot doctor. As reported in many news reports, artificial intelligence can now help build patient records, save doctors' time, and artificial intelligence can analyze X-rays and CT, but diagnosis and prescribing can only be done by doctors.
Even though many giants are working hard at the moment, Google has recently launched an AI+AR tumor diagnosis system, but our current medical ecosystem is still the "Stone Age", because many systems are not perfect enough, we must recognize The direction of force, to improve the speed of progress in the medical field, also requires more AI talents to accelerate the AI ​​research process in the medical field.
Hand And Feet Warmers,Hand and Foot Warmers,Winter Sports Hand & Foot Warmers,Cordless Hand and Foot Warmers
Ningbo Sinco Industrial & Trading Co., Ltd. , https://www.newsinco.com