DR GOOGLE GETS SMARTER
The latest artificial intelligence technology can predict when people are likely to die
GOOGLE really does know everything about you – including, apparently, when you’re going to die. Using new artificial intelligence (AI) technology, the global tech giant can now predict when you’re most likely to kick the bucket.
Last month the journal Nature detailed how the new algorithm can use vast data sets to predict whether patients would survive in two different hospitals.
Google’s Medical Brain algorithm was 95% accurate in predicting patient mortality in the first hospital. In the second hospital the accuracy was at 93%.
“This was significantly more accurate than the traditional predictive model,” Nature reported.
“These models outperformed traditional, clinically used predictive models in all cases.”
The data used includes things such as patients’ age, ethnicity and gender. This information is then combined with hospital information that includes any prior diagnoses, current vital signs and lab results.
The system is unique in that it receives data that’s typically not within reach of other machines such as doctors’ notes and charts.
In one case study the algorithm gave a woman with late-stage breast cancer a 19,9% chance of dying, whereas the hospital’s early warning score gave her only a 9,3% chance. The patient subsequently died two weeks later.
The research demonstrates the potential role of AI in healthcare, predicting things such as how long patients might stay in hospital, chances of readmission and risk of death.
Google isn’t stopping there – the Alphabet-owned company plans to use the powerful tools not only to predict deaths but also symptoms and diseases.
The tech giant also has plans to provide doctors with voice recognition systems that will minimise the time-consuming task of writing notes, Digital Trends reports.