The Google AI team brings a new algorithm for predicting human death time! How does this discovery save more lives?
Release date: 2018-06-20 In the previous article, we reported that Stanford University developed an AI algorithm earlier this year to predict when a hospitalized patient will die. And this technology has recently made another leap: Recently, researchers at the Google AI team used the AI ​​algorithm to successfully predict whether a patient is at risk of death after 24 hours of hospitalization, and the accuracy of the prediction results. Amazing 95%! The results of this study have been published in the journal Nature, npj Digital Medicine. In this study, researchers used neural networks and a large number of data on patient vital signs and medical history to predict the risk of death in hospitalized patients. This new algorithm combines the prior events recorded by each patient into a timeline, allowing the deep learning model to predict future outcomes such as time to death. This neural network records a wide range of data types, including handwritten notes, comments, and graffiti on old charts for more comprehensive and complete predictions. Throughout the study, the researchers analyzed more than 210,000 hospitalizations and more than 46 billion data points in the Electronic Health Records (EHR) of approximately 110,000 patients. Schematic diagram of the AI ​​algorithm prediction process (Source: npj Digital Medicine) In one of the main case studies, Google researchers applied the algorithm to a patient with metastatic breast cancer. According to the doctor's diagnosis and radiographic results, the hospital gave the patient a 9.3% death rate during hospitalization. Google’s algorithm, after analyzing 175,000 data points in the patient’s electronic health record, found that the patient’s death probability was 19.9%. In fact, the breast cancer patient died in the less than two weeks after admission because of his illness. This result verifies the accuracy of this new algorithm. The researchers say that this study allows machines to parse data independently, as well as data that was previously unavailable, including data in PDF documents and information recorded in old notes and charts, compared to other previous studies. Wait. Neural networks can incorporate this data into the analysis, while predicting results more quickly and accurately than existing technologies, and even presenting which data and medical records lead to the final conclusion. Dr. Jeff Dean, head of Google's AI department, said that the next step for researchers is to apply this research to the actual diagnosis. To this end, Google's Health Research department is developing a series of AI tools that accurately predict disease and corresponding symptoms. At the same time, this research is expected to eliminate the need for doctors to spend a lot of time integrating patient health data into a standardized format. In this way, the doctor can interact and communicate with the patient to adjust the treatment plan according to the actual situation of the patient, and can also find the emergency in the treatment and take corresponding first aid measures. Reference materials: [1] Scalable and accurate deep learning with electronic health records [2] Google Is Training Machines to Predict When a Patient Will Die [3] Google says its AI is better at predicting death than hospitals [4] Google's AI Can Predict When A Patient Will Die Source: WuXi PharmaTech AI Royal Jelly is also called Bee Milk. The fresh royal jelly is slightly ropy milk paste substance; it is the excretive mixture of nutrition gland and maxilla gland of the head of little worker bee. The worker bees use this to feed the 1-3 days` worker bee larva and drone larva, 1-5.5 days` queen bee larva and queen bee in the oviposition period. The royal jelly is the biologic product containing very complicated active elements which contains almost all the nutrition elements required by the growth of the human body. Royal Jelly,Natural Royal Jelly,Healthy Royal Jelly,Organic Fresh Royal Jelly Easy Food (Jiaxing) Co., Ltd. , https://www.jxeasyfood.com