Google’s AI can now predict a patient’s death
Google’s Medical Brain team has developed a new algorithm to predict the death chances of patients in hospitals. The team has found some early results that indicate the accuracy rate of Google’s AI is more than hospital’s own death predicting systems.
Bloomberg report describes the health care manifestations of its finding and uses previously unusable data to reach the prediction of death risks among patients. AI uses the electronic health records and other patient’s information to get the likelihood of death, discharge, and readmission of the patient. Google’s new tool forecasts a variety of patients outcomes, including how long patients may stay in hospitals, their chances of re-admission and risks that they will soon die.
Google published its research in Nature where it describes this algorithm:
“These models outperformed traditional, clinically-used predictive models in all cases. We believe that this approach can be used to create accurate and scalable predictions for a variety of clinical scenarios”.
In a case study published in Nature, Google applied its algorithm tools to a patient having metastatic breast cancer. After 24 hours of being admitted to a hospital, Google gave her a 19.9 percent chance of dying in the hospital, on contrary with the 9.3 estimate with the hospital’s Early Warning Score. In less than 2 weeks later, the patient died in a hospital from her condition.
This is not the first time Google’s algorithm tools have been applied to predict healthcare. Earlier in May, Google’s Artificial Intelligence research division made an augmented reality microscope for cancer detection. The special AR microscopes (ARM) use machine learning to detect cancerous cells.
AI head Jeff Dean said,
“Google’s augmented reality microscope (ARM) combines both methods, it blend[s] the expertise of automated machine learning systems with human expertise.”
If Google can both smooth the process of entering data and improve the means by which that data is used, it could lessen the chances of human error in medical care.