It’s hard to believe that by the virtue of technology advancement Artificial Intelligence (AI) knows when you’re going to die. And in a futuristic break-through, this information can end up saving lives. Google (Alphabet Inc. (NASDAQ:GOOG)) AI and its noteworthy presence in the healthcare space has been steadily increasing, and importantly, rooting deep with innovations for mankind.
The latest research by a team of data scientists from Stanford, University of Chicago and UC, San Francisco, suggests that feeding electronic health record data to a deep learning system can now aptly predict a patient’s length of stay in the hospital as well as the mortality rates. What’s more astounding is the fact that it can predict the death of hospital patients, with 95% accuracy using Artificial Intelligence.
Like the quantum leap wasn’t enough already, Google’s ability to delve deep and scrutinize relevant window through heaps of data to predict the results is out-of-the-ordinary. The Medical Brain Team trains its AI in such a way that it efficiently filters health information of patients and then predicts it accurately than all other existing medical tools and techniques.
Google first divulged about this radical system in a paper published in the Journal Nature in May.
So How Did Google Do It?
In one major case study, Google incorporated its A.I. to evaluate a patient with breast cancer. In the following 24 hours, for which she was admitted in the hospital, the early warning score submitted 9.3% chance of her death. However, by using Google’s AI, it predicted 19.9% chance of her dying, which was a considerable difference in the Hospital’s prediction and Google’s AI.
The A.I. scrutinized about 175,639 data points found in the patient’s electronic medical records, including interpretation of handwritten notes. According to claims made by Google, the inclusion of all this information was responsible for the elevating difference between Google and previous conventional approaches.
Google has till now analyzed about 216,221 hospitalizations and 114,003 patients, which when accounted, stacks up to more than 46 billion data points from all EHRs and surprisingly, results are particularly promising for healthcare professionals. This can be a step forward for Google AI’s foot-wide entrance in the medical domain.
Google AI In Computer Predictive Systems
Google’s AI health care plans aren’t new. Apart from this recent addition of AI implementation in the medical field, Google’s research team has already introduced a deep learning model earlier, that can detect cancer.
This AI is trained in a way so as to when applied on microscope can detect cancer in real time.
Google introduced a paper “An Augmented Reality Microscope for Real-time Automated Detection of Cancer” at an annual meeting of the American Association for Cancer Research (AACR). It described a prototype Augmented Reality Microscope (ARM) platform that can possibly bolster acceleration of deep learning tool adoption, for pathologists, all around the globe. The ARM which was trained on images of human tissue resulted with AUC as high as 0.98. This can be implemented for the diagnosis of infectious diseases, including tuberculosis and malaria, in developing countries.
Taking Care of Health As Well As Privacy
Google uploads heaps of health data to its centralized computer systems, which are spread across various healthcare systems and government agencies.
What’s necessary for the better functionality of those AI, is the proper channelling of those private data into a single predictive model. However, harvesting such amount of personal data of patients, no matter how important, doesn’t seem that appealing given the privacy-breach by a chief tech giant, earlier. Electronic health records of millions of patients, available without any no-no, may give big-shot companies a lobby to exploit health industries and claim monopoly in healthcare.
Especially after the public brouhaha over data breach, Google’s access to private data even if for a noble cause, is sure to raise few eyebrows.
However, coming up clean, Google and its hospital partners claim that the data they fetch is completely anonymous and is used only after seeking patient’s permission.