During the past two years or so, the pace of evolution within healthcare has been mind boggling. Perhaps the most notable achievement in which technology helped shape healthcare outcomes is the speedy rollout of viable vaccines for new diseases.
This was practically demonstrated when the vaccination for Covid-19 was prepared in record low times, saving millions of lives. In a similar breakthrough, we are a step closer towards Clinical Language Generators.
The academic health center of the University of Florida, called UF Health, and technology giant Nvidia have successfully built the world’s largest clinical language generator till date. The innovative and groundbreaking tools is being dubbed SynGatorTron.
The most notable achievement of SynGatorTron is its ability to generate synthetic clinical data. This in turn is an invaluable achievement in the highly regulated sector of public healthcare. Medical records are otherwise a very sensitive and confidential information.
However, researching diseases is also equally important, as precious human lives are involved. There is also the element of protecting the identity and personal health records of thousands of patients.
SynGatorTron addresses these limiting factors within healthcare research by generating its own synthetic data. This neural network has been trained on one decade of data, spanning more than two million patients.
Being a whopping five billion parameter model, SynGatorTron has become the world’s largest language generator within healthcare. By leveraging the data and findings of this tool, healthcare researchers can perform their tasks without compromising privacy.
This model was trained on the University of Florida’s in house Nvidia DGX SuperPOD system, which is ranked among the top thirty most powerful supercomputers across the globe. The synthetic healthcare data it generates can be invaluable for medical research.
With the introduction of the world’s largest and most powerful clinical language generator, we are entering a new era of healthcare research. Not only will the remedies to diseases become much quicker, but all this will be achieved without compromising patient privacy.
Great Job Nvidia and University of Florida!