Qure.ai, a healthcare startup funded by Fractal Analytics, has launched an artificial intelligence based system to identify abnormalities in head CT scans.
The effort is the latest example of how AI and machine learning are working through the health care industry. Qure.ai released a clinical validation study showing its algorithms were nearly on par with radiologists in a sample of 21,000 patients.
In addition, Qure.ai is making a dataset of 500 AI analyzed head CT scans available for download.
Qure.ai is aimed at a key supply and demand choke point in the healthcare system. Images from MRIs and radiology are outpacing the humans available to interpret them. The general idea is that AI can be used to interpret results and free physicians up for patient care. Speed is also an issue when it comes to interpreting a head CT scan of a stroke victim.
Fractal Analytics buys Final Mile as AI, data science meld with behavioral science | Death and data science: How machine learning can improve end-of-life care
The Qure.ai model was trained with 313,318 anonymized head CT scans and their clinical reports. Out of that sample, 31,095 scans were used to validate the algorithms. From there, AI was clinically validated on 491 CT scans and compared against a panel of three radiologists.
According to Qure.ai, its algorithms were more than 95 percent accurate.
The results were published via Cornell University and the paper is publicly available along with the data set.
Fractal Analytics is planning to invest up to $30 million in Qure.ai in the next few years.
Here's what a Qure.ai AI-driven CT analysis looks like.