Using AI to Assist & Train Physicians
Our custom models are being used to improve physician performance through diagnostic technology and training tools.
With the increase in AI and machine learning capabilities, we are making groundbreaking advances. Recently a team at Stanford built a model that outperforms the average radiologist at diagnosing pneumonia from chest X-rays [1]. Another team recently built a deep learning based model that detects skin cancer from photos with an efficacy rate on par with tested Dermatologist experts [2]. And we here at Xyonix are building a number of AI models that help physicians.
In the video on the right, you can see a model of ours in action detecting papillary carcinoma lesions from a cystoscopy (procedure that allows your doctor to examine the lining of your bladder and urethra, or tube that exits urine from your body). Note at 0:21 when the lesion detector activates on seeing the carcinomas (the sponge-like entities). You can also get a glimpse into what parts of the image the AI thinks are important in determining whether a given frame has a lesion present or not.
At Xyonix, we're busy building multiple AI backed tools to assist and train physicians. Below are a few of our systems helping to improve physician performance.
If you have data that you suspect could help a physician make a more accurate diagnosis or improve, we can rapidly assess your hypothesis and help you get your tool into the hands of physicians where it can make a real impact.
On our podcast, we spoke to Dr. Tom Lendvay of Seattle Children’s Hospital about using AI to assist and train physicians. In the episode, Tom explains how AI can best be used for assisting physicians and what changes are needed to allow for machine learning to make a bigger impact. If you would like to listen to this episode please click below.
On Your AI Injection, you’ll learn from industry experts about how to leverage AI for health solutions:
REFERENCES
CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning, Rajpurkar et al, rXiv:1711.05225
Dermatologist-level classification of skin cancer with deep neural networks, Esteva et al, Nature 542, 115–118 (02 February 2017) doi:10.1038/nature21056
Photo by Adrian Clark / CC BY