Artificial intelligence (AI) is already making an enormous impact in the global healthcare community. But Google has more ambitious goals in store for the future of MedTech development — it wants to accurately predict your health outcome at the time of your admittance to a hospital.
Connecting the Dots
On January 24th, 34 Google AI researchers released the latest findings of their EHR data study in a non-peer-reviewed research paper. In the publication, the researchers discuss their new software which can predict patient outcomes like discharge, readmittance, and death more accurately than current paradigms in use.
In fact, Google claims that it can predict deaths one to two days faster than the traditional means. That is a substantial amount of time that could allow doctors to improve the odds of patient survival.
The San Francisco developers in charge of the project tapped into seven years of data from University of Chicago Medicine and four years of data from California San Francisco Medical center. The de-identified information spanned 216,221 patients which eventually led to 46 billion extrapolated data point connections between them.
The Disparity in Data
Electronic health records can be complicated. Not only is there an immense amount of information, but the data and its organization can be radically different from one system to another. Not only this, but a great portion of the medical industry still takes notes by hand, a notorious conundrum that AI has difficulty comprehending.
To circumvent this, Google’s researchers employed three deep neural networks to not only parse the data but determine which parts are most pertinent to accurately predicting the patient outcome. By being trained with vast amounts of data, Google’s system has learned how to cut through the noise and pay attention to key aspects that could drive the most probable result.
League of Legends
This research coming from Google is a bit surprising when considering that Alphabet, Google’s parent company, already has quite a few entries under its belt in the MedTech space, like DeepMind, Calico, and Verily. But the company has declared itself “AI-first,” and the team it put together for this project seems to stand by that mantra.
Some heavy hitters listed as authors of the research include Greg Corrado, who focuses on “brain-inspired” computing; Jeff Dean, who also worked on TensorFlow, LevelDB, and Google Brain; and Quoc Le, whose work in recurrent neural networks is world-renowned. It will be interesting to see if this new software is implemented in a health facility anytime soon.Tags: AI, AI App Developer, AI App Development, AI apps, artificial intelligence, artificial intelligence app developer, artificial intelligence app development, MedTech, MedTech app, MedTech app developer, MedTech app developers, MedTech app developers San Francisco, san francisco, san francisco AI app developer, San Francisco app developers, San Francisco tech