Smartphone manufacturers may have changed the meaning of Artificial Intelligence (AI) for the worst. But unbeknownst to many, it has a way deeper meaning for other fields of science.
AI helps scientists to forecast the weather or prompt pilots for the incoming planes. In the business side, AI aid investors in making trading decisions faster than any human being capable of. It also assists manufacturers for accurate and faster production rates.
Now, in a rather spine-chilling (but certainly useful) turn, artificial intelligence can now accurately predict our own death.
The recent study published in Digital Medicine highlights the use of AI to predict patient’s mortality rate 24 hours after hospital admission. The new Artificial Intelligence (AI) is currently under the development of Google Medical Brain team.
In one instance reported in the study, a woman with late-stage breast cancer was admitted to hospital. Upon arrival, they found out that her lungs were filled with fluid, and so several doctors were called to evaluate. The hospital’s assessment claimed that she had 9.3 percent chance of dying during her stay. But this is solely based on her vital signs, disregarding all the previous data taken.
Google’s AI ran its own evaluation on the same patient, analyzing 175,639 data points on her record. Some of the data points include records doctors don’t normally consider during patient evaluation. But the AI was able to access it and add it on its assessment. These records include PDFs of notes made by doctors and nurses that indicate evidence of malignant pleural effusions (or fluid build-up around the lungs) and potential risk of pressure ulcers.
“In general, prior work has focused on a subset of features available in the EHR [electronic health record], rather than on all data available in an EHR,” the authors wrote in their study. “Which includes clinical free-text notes, as well as large amounts of structured and semi-structured data.”
Google’s new AI system is capable enough to analyze and process large amounts of data, both structured and unstructured, with more accurate predictions. In fact, up to 93-95 percent accuracy rate versus hospital’s current predictive model which is only about 85-86 percent accurate.
Taken into account all the data available for the patient, the AI was able to lift the patient’s risk of dying during her stay to 19.9 percent. Sadly, she died 10 days after admission.