This snapshot from yesterdays' Cantor Fitzgerald Global Healthcare Conference is supposed to show how great Zebra Med is:
Answer: Revenues, even though Zebra Med boasts its own CPT code.
Note also that the slide below is a complete lie. Why? Because Zebra software can't tell whether a Nanox.Arc image is "normal" or not. Nanox.Arc is supposed to generate low-quality, short SID supine AP views, supposedly of chest or extremities, and artifact-filled tomosynthesis slices thereof with miniscule field of view. Zebra has no such views/data on which to train its algorithms and cannot obtain them for any money anytime soon. As of today, there is no evidence that any functioning Nanox.Arc device exists, so it will take years, thousand of working Nanox.Arc devices, and millions of dollars to generate the training data (millions of views) necessary to train Zebra's AI to perform even slightly better than a random coin toss.
That is something that both the current and future Nanox CEOs pretend to ignore.
Hello,
ReplyDeletethanks for sharing this.
Just read your post, and i have a question for you.
Do you think could be possible for Zebra to work and understand (AI) whats "normal" using data from old scan instead the one from new nnox sysem?
Regards
@Anonymous No (based on my review of Zebra's published research). Zebra's has access only to images collected from regular x-ray and CT scans, which means they cannot be used to determining "normal" images from Nanox.Arc. I mention chest images, for example. In the real world, chest images are properly-distanced upright PA images, and those cannot be done with an Arc at all. AI is really a misnomer - Zebra's algorithms do not understand what is on the images.
Delete