Overview / Usage

The user or his/hers caregivers connect a online electronic health record (such as Lydia).
The prescription medication is either already in EHR, or using computer vision is added to the EHR.
When bought new prescription medicine the number of pills is either detected by computer vision or by internet search.
As usage, at the times specified by EHR the user is reminded to take the medication. For some users using computer vison (checking if is the right box) it can ensure that they are using the right pills. The assitant knows when new pills are needed and can notify the user and/or the caregiver. As a second step the skill again using AI could "see" is the pill consumption is on the mark or over. And as a further development could touch the diabetics care, calculating from food calorie intake, read the insulin levels and recommend actions.

Methodology / Approach

The approach is snowballing to MVP.
First step is connecting to EHR, and just getting notification when and what medication to use.
Then see how to use computer vision to "read" prescriptions. Second how to read from recipients the name and the number of pills.
Explore AI to count the actual number of pills. As a further complex steps to see if AI can recognize from the image of the pills if they are the right ones (from the prescribed ones).

Technologies Used

Besides PC Skills (is a topic :))…
The differentiator of the assistant are his vision and AI skills, here I'm planning to look and pick the right combination (maybe one technology would be enough but now not sure), between Open Vino, Movidius NCS and AI DevCloud. As EHR would tart with Lydia.