Google’s new AI system can predict patient’s death

Artificial Intelligence has progressed so much that now Google is enabling its AI to predict death before doctors and even more accurate than them.

New report published in Nature have suggested that Google’s AI can soon examine a person’s medical history and help doctors make more precise predictions regarding a patient’s health and provide according estimates on when they might die.

As an example, Bloomberg reported, in one situation a woman arrived at a hospital with cancer and had fluid in her lungs. Doctors estimated a 9.3% chance of her dying during her stay. Google’s AI predicted that the risk of her death were 19.9%. The woman died a few days later.

Google makes this possible by using neural networks that are effective at gathering data and then using it to understand and enhance analysis. Along with predicting the time of death, Google’s AI tech can also predict how long patients might stay in hospitals and their odds of re-admission.

US approves new algorithm that can predict deaths

As per Daily Mail, Google’s algorithm uses patient data including their age, gender, ethnicity, along with hospital information like comments, handwritten notes, prior diagnosis, current vital signs, lab results, notes buried in PDFs and scribbles on old charts and make according predictions.

This data benefit hospitals to discover new techniques for prioritizing patient care, catch medical emergencies before time, free healthcare workers, and adjust their treatment plans. These abilities help doctors make diagnosis than doctors.

The report also suggests that this system is faster and more accurate than other methods used to evaluate a patient’s medical history. Google’s AI is 95% accurate, which is 10% better than traditional models. Google’s research scientist Alvin Rajkomar explained, “These models outperformed traditional, clinically-used predictive models in all cases.”

Though the system is still in its initial stages, Google believes that it could one day predict death way before the actual occurring.