How does AI become a "medicine" step by step?

"He just wants to live, what sin he has!"

Outside the morgue, after Huang Mao’s death, Cheng Yong shouted to the police officer.

For generics, in order to survive, how many people have made their lives.

High-priced genuine medicines allow patients to retreat. It is undeniable that the development of a new drug, especially the "special effect drug", requires over 100 million research and development costs and research and development cycles. It can be available to the market and is already the "gospel" of many patients. However, in the face of high prices, how to "cool down" "special effects" and disease treatment, AI may be able to become your "drug god" step by step.

The first step: AI predicts leukemia, so that leukemia no longer becomes a "sudden disaster"

Recently, Nature published a study that used a blood test and machine learning by a team of leukemia scientists from a number of national research institutions to predict whether a healthy individual is at risk for acute myeloid leukemia (AML). .

This means that we will have early warnings about the emergence of AML, and will be able to detect and monitor AML high-risk populations early, and at the same time, we can conduct research and development to find solutions to reduce the risk of the disease.

AML is called "acute myeloid leukemia". It is characterized by abnormal proliferation of primitive and naive myeloid cells in bone marrow and peripheral blood. The cancer cells of AML patients proliferate rapidly in the bone marrow and prevent normal blood cells from producing, leading to bleeding. And symptoms of infection, even life-threatening.

The researchers therefore developed a gene sequencing tool that sequenced blood DNA from 124 AML patients for known genes associated with AML and compared them with 676 people who did not have AML or related cancer.

Through big data monitoring, they found that many people with AML had genetic changes in their genes, and those who did not have the disease did not. Those who later develop AML have more mutations in their genes, and these mutations also have a higher proportion in their blood cells.

With further research, the researchers built a model of AML prediction through the robot learning model, supported by big data variables, which can achieve AML prediction within 6-12 months before diagnosis. Its sensitivity and specificity. They reached 25.7% and 98.2% respectively.

Earlier, Watson also diagnosed a rare leukemia of 60 women. Watson used only 10 minutes to compare the genetic changes of different patients in the 20 million cancer data reports - not only accurate conditions For diagnosis, Watson also provided an appropriate treatment plan.

The emergence of AI predictions is gratifying and many people are suspicious. Indeed, for example, AI predicts the occurrence of death time, which makes the application of AI no longer a technical issue, but also an ethical issue. It doesn't seem like a good thing when you know when you are sick and when you die.

Step 2: AI Pharma, changing the drug development model

In the movie "I am not a drug god", the focus of contradiction lies in the high price of "Gleining", the new drug is expensive, and the "trial and error" link in research and development and research and development, AI may be able to help.

From the current development of smart medical care , artificial intelligence that is good at pattern recognition can screen and screen from the vast amount of existing and new genes, metabolism and clinical information to solve the complex network behind various diseases. This, in turn, helps to identify drugs that are appropriate for a particular patient population, while guiding the drug companies to circumvent drugs that may fail.

In addition, the biological significance of artificial intelligence can help pharmaceutical companies to participate in clinical trials of innovative therapies that are most likely to be effective, depending on the patient's condition, which may be possible to increase the approval of new drugs, such as FDA approval.

In fact, the core of medical research and development lies in the knowledge map, which combines experimental information, data, clinical experiment results and data, and integrates scattered data to provide valuable data support for decision-making.

From the current point of view, there are seven main scenarios in which artificial intelligence mainly plays a role in drug development: target drug development, drug candidate drug discovery, compound screening, prediction of ADMET properties, drug crystal form prediction, auxiliary pathological biology research, and new drug adaptation disease.

According to Tech Emergence's research report, AI can increase the success rate of new drug development from 12% to 14%, which means that it can save billions of dollars in research and development costs and a lot of trial and error time for the biopharmaceutical industry.

However, it is undeniable that AI drug development must be a protracted war. At present, there is no successful case of AI drug research and development in the world, and drugs developed by artificial intelligence have not been approved for marketing.

The new drugs developed by foreign companies with good AI have entered the second phase of clinical trials, but the failure rate of the second to third phases is as high as 70-80%. AI technology has broad application prospects and still has a long way to go. Giants such as Pfizer, Roche, and GSK have "bet" AI companies, and the current development will take time to test.

But this does not mean that AI Pharmaceutical is not possible. If the technology can effectively shorten the efficiency of drug research and development, and increase the success rate of research and development, the cost of drug research and development will be greatly reduced, which can greatly reduce the burden of national medical insurance, and "parity drugs" will also become possible.

The third step: drug data becomes the key to AI pharmaceuticals

In fact, we can also see that AI in every step of the field of smart medical care, which can not open an important factor: drug data.

For example, in the field of new drug research and development, AI can help scientists complete literature searches from a large-volume compound database. Many companies are also studying how to use machine-simulated compounds to combine with specific targets, thus greatly speeding up the process of screening new drugs. Every year, hundreds of billions of dollars are used in new drug research and development, and the use of AI technology can improve research and development efficiency to some extent.

By learning machine learning, AI can not only speed up the time, but also increase the probability of success of drugs reaching the later stages of the trial. If AI can reduce the risk of drug testing, it can save large pharmaceutical companies a lot of money, allowing them to free up resources to focus on finding more potential opportunities.

Similar artificial intelligence applications are also promising in epidemiological statistics, clinical trial data analysis, and precision medical genetic testing. In the field of artificial intelligence precision medical projects, IBM also launched "Waston for Genomics" following "Waston Oncologist"

In addition to drug data, medical data has become an important basis for physician diagnosis and follow-up drug development. With the rise of health-smart hardware, the boundaries of medical data are constantly being expanded.

In April 2016, data recorded by a health-smart hardware saved the lives of a New Jersey man. The man had a heart attack at work, and the doctor extracted his daily heart rate data from his smartphone, which helped the doctor to eliminate unnecessary diagnoses and quickly find a suitable medical method with the doctor to save a life.

Medical data is more than just medical journals and medical records that doctors enter into computers. Our bodies generate massive amounts of potential medical data all the time. But for now, most of the data is in a "lost" state - how many steps we take every day, how heart rate is today, whether the temperature of the skin is high or low, what we eat today, and so on, They are only kept in local, solitary devices and apps.

For AI medical care, the importance of data is self-evident. Whether it is applied to drug development or diagnosis and treatment, there is considerable prospect, but the diagnosis or presumption made by AI on a small sample set is considered to be an unsustainable model, because once you expand a little more, you can change a disease. If you change one place, the result may be biased and the correct rate will drop.

In general, AI medical development has made great progress so far. Although many AI medical products have not yet landed, it is not the patient's "medicine god", we are walking around.

Product That Can Be Eaten Straight Away

Coffee Ice Cream Crunch Cone,Extra Large Waffle Cones,Purple Potato Wafer Cone,Square Wafer Ice Cream Tart

Tianjin Yongkang Food Co., Ltd , https://www.yongkangfood.com