SoylentNews is people
SoylentNews
An AI and machine learning model tuned to decode hastily scrawled notes from doctors:
Google is developing an AI model that can decipher difficult-to-read handwriting, with a focus on notes and prescriptions written by doctors. The search giant announced during its annual conference in India on Monday that it was working with pharmacists to create a tool in Google Lens that can decode messily written medical notes (via TechCrunch).
Google showcased the feature during the event, demonstrating its capability to specifically detect medicines in a handwritten prescription. There's no detail yet on when the new text deciphering feature is expected to launch, only that "much work still remains to be done before this system is ready for the real world."
From the TechCrunch article:
[...] The feature, currently a research prototype and not ready for the public yet, allows users to either take a picture of the prescription or upload one from the photo library. Once the image is processed, the app detects and highlights the medicines mentioned in the note, a Google executive demonstrated.
"This will act as an assistive technology for digitizing handwritten medical documents by augmenting the humans in the loop such as pharmacists, however no decision will be made solely based on the output provided by this technology," the company said in a statement.
(Score: 1) by anubi on Friday December 23, @02:04AM
It's a database of supposedly every drug out there.
We need this in a local database, residing completely inside a phone or computer that has NO connection to the internet ( or cellphone network. No SIM card! ), To be personally kept by the physician.
The physician can select anything in the database to provide the QR code that links to the selected drug. That link, for that particular drug, containing no other information whatsoever, can be verified for it's privacy and correctness, can now be read and given to the pharmacist, along with printed copy of the prescription, verified by the doctor for accuracy.
The main database, is public. About as private as a copy of the PDR on Amazon.
The physician accesses it occasionally just to update his local copy in his private machine.
At no time does the physician access the public reference database for an individual drug. He gets the whole shebang, providing no information to the public other than he downloaded the file.
I am thinking this to be done similar to downloading WIKI files as .slob or similar as used in the AARD (F-droid) Wiki reader.
Except only the providers of this drug Wiki have write permissions. This Wiki itself should be public. Funded by the public, and suggest administration by an entity such as the one who presently publishes the PDR.
AARD seems to have a good reader. All that's needed is appropriate .slob files and an export link as QR code to save bookmarks to your own local copy.
Well, that's my thoughts on this. Make the whole thing public.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]