The Medical Brian team owned by Google has developed an AI technology that can obtain information that the past AI could not obtain, such as notes annotated on the chart, notes in the PDF, etc. Now all can be analyzed and judged. The new system is obtaining data, The required analysis time and other aspects are all more efficient than in the past. In addition to predicting the possible length of hospital stay for the patient, the probability of re-admission can even predict the probability of death in the short term.
The Google Medical Brian team worked with Stanford University, the University of Chicago, and the University of California, San Francisco, in May this year to study the AI model for predicting patient mortality. In one of these cases, a patient with an advanced breast cancer disease Suffering, after going to the hospital after an examination, the hospital computer interpretation of life signs, estimated that the patient's chance of death during hospitalization was 9.3%, but the Google AI analysis found that the risk of death was as high as 19.9%, and later the patient was admitted to the hospital. Died after the week.
It has been pointed out that Google AI program predicts the accuracy of human death rate as high as 95%, while the accuracy of traditional medical prediction methods is 86%.
In traditional medical diagnosis, doctors spend a lot of time dealing with documents and data, and AI is expected to help doctors to increase efficiency. However, the ability to further predict mortality has also led to many ethical discussions, including the timing of the use of AI technology, the scope of the right to use, etc. And whether it will cause panic, the impact on medical insurance, etc.