Google’s work in artificial intelligence is moving at a remarkable pace in the health sector. In a recent breakthrough, Google decided to compete with hospital’s old machines to predict a patient’s death and came up with astonishing results, subtly hinting us about the future of A.I.
As first reported by Bloomberg, Google tested out its brilliantly performing A.I algorithms on a patient who walked in with late-stage cancer in a city hospital. Fluids were already building up in her lungs, amid which the hospital’s computer predicted 9.3% chance of dying based on her vital signs.
Whereas, Google using its A.I neural network algorithms was able to predict an acute percentage of 19.9 chance of death. Unfortunately, the doctors weren’t able to save to the women, but her name surely got registered in what’d we call an A.I breakthrough.
A journal published by Google’s Medical brain team stated: “Patient records vary significantly in length and density of data points, so we formulated three deep learning neural network model architectures that take advantage of such data in different ways: one based on recurrent neural networks (long short-term memory (LSTM)) one on an attention-based TANN, and one on a neural network with boosted time-based decision stumps.”
The neural algorithm analyzed 175,639 data points from her medical records and considered nearly all variables like blood pressure, heart rate, respiratory rate, temperature, white blood cell count and came up with the most accurate assessment of her death risk.
What made the AI efficient was the ability to decipher large volume of messy data, including errors in labels and bring out patterns by interacting with the data itself.
After the phenomenal revelation of Google cancer detection microscope, A.I penetration in the health sector is undoubtedly going to enhance the ability of doctors and caretakers, ultimately saving more and more patients.
But A.I also brings in the question of privacy for availing new world services. Last year, Google received a lot of scrutiny after a revelation that its subsidiary DeepMind had access to nearly 1.6 million records of patients from three hospitals without their consent.