The sheer amount of data that Google collects from its estimated 1.17 billion users is quite staggering. From your online browsing and search habits, places you go, what groceries you buy, and even your psychological and political leanings.
This massive repository of data is mostly used for marketing, consumer research and more importantly, for predicting our behavior. From auto-completing your search terms to accurately guessing what consumer goods you’re most likely to buy, Google’s predictive algorithms are getting more powerful over time.
With the ever-increasing role of artificial intelligence and deep machines in our lives, Google is increasingly applying its predictive expertise to the medical field. This could, quite literally, spell the difference between life and death.
Google is training AI to predict when a patient will die
According to a new report from Bloomberg, Google is looking to take its predictive models and machine learning expertise to clinics and hospitals to have artificial intelligence predict if a patient is likely to die, among other things.
By combining information from patients such as their vital signs, age, ethnicity and gender with other raw data like prior hospital records, notes, and old charts, Google’s tools can predict a variety of patient scenarios with stunning accuracy.
These predictions include how long people may stay in hospitals, their odds of readmission and the chances of death.
What amazed medical experts the most was Google’s ability to go through data that was previously out of human reach. Notes buried in PDF files, for example, and scribbles on old charts are likewise parsed by Google’s AI and are included in its predictive models.
This means that Google’s system can quickly process almost any type of data that are thrown at it and with deep machine learning, it can be significantly more efficient and more accurate than humans over time.
Speaking of AI, will we create a super-intelligent AI and forget to give it a conscience? How far is too far when it comes to AI? Listen to this free Komando on Demand podcast as I explore the real risks that face all of us as the AI world gets bigger and smarter.
Google’s Medical Brain
The heart of this project is at Google’s Medical Brain, a division of the company that focuses on advancements in the medical and health sector.
Hoping to pioneer a whole new market for Google, the Medical Brain team constantly works on a variety of machine learning and AI tools that can predict symptoms and disease with a stunning level of accuracy.
According to Google’s research paper, even in its early stages, the initial results of Medical Brain’s AI deep learning approach already outperforms traditional predictive models in all scenarios – hospital stay length, in-hospital deaths and readmission.
In the future, deep learning can even guide doctors in diagnoses and in prescribing certain medications. Google’s learning algorithms may even incorporate health records with other data like local weather and traffic to fine-tune its predictions even further.
Over time, Google could then license these predictive systems to hospitals and clinics as a Cloud-based diagnostics-as-a-service.
And that’s not all, Medical Brain also has plans of going beyond hospital medical records as AI systems for radiology, ophthalmology and cardiology are already in the works.
The price to pay? More data access for Google
Using Google’s deep learning expertise for medical care does have its advantages but at what cost? Granting the company more access to our digital medical records, of course.
Amidst the public’s growing awareness about online privacy and the data collection practices of these big tech giants, securing medical records from various health providers and hospitals will be a big hurdle.
For this ambitious project to reach its lofty goals, Google will have to sell its vision to the medical industry and convince new partners to grant access to sensitive records.
Obviously, the big concern here is that we’ll just be adding more critical information to the already extensive repositories of data that Google has on all of us.