Blogs

FLOWRx technology using AI to improve outcomes

Written by Jared Hancock | Jul 22, 2024 2:53:02 PM

Jared Hancock, Director of Software at Centred Solutions, explains how the company’s cutting-edge FLOWRx technology is now harnessing artificial intelligence (AI) to improve performance.

Science, technology and AI are colliding to create transformative new opportunities to improve healthcare. At Centred Solutions we always have one eye on the future and that’s why our FLOWRx Hub solution is now using AI to support the way patient labels are automatically applied to medication packs.

Machine learning is now used to identify the text orientation on medication packs so that FLOWRx’s Automated Labelling Unit (ALU) can affix the patient label with its text running in the same direction as the pack. We were able to train the model with around 3,000 images which were classified as either:

  • Left or right: Where the writing on the pack is in a “portrait” orientation.
  • Normal or inverted: Indicating if the label should be put on the way it comes out of the printer (normal) or flipped around 180 degrees (inverted).

How Our FlowRX Hub & Spoke Solutions Work

While in operation, the system uses digital imagery of the medication pack when it is on the conveyor belt and identifies critical criteria about the box of medication. These images are then fed through the machine learning model and there are multiple classifications used to determine the orientation of writing ahead of labelling the medication pack. The AI model has been introduced over prior algorithms to improve the system accuracy rating for label orientation so that it is at or above 90%. This current AI solution is significantly faster, taking only six milliseconds to identify the text orientation of the pack compared to 450 milliseconds with the previously used methodology. It is being used by all of our FLOWRx Auto Hub pharmacy customers.

The machine learning model we now use is much more successful at determining the pack text direction so the patient label can be applied in a way that makes it easier for patients. The automated results closely align with how the pharmacy staff apply the labels by hand.

Ongoing learning 

In addition, the model allows us to take new images of medication packs being used by the automated labelling units out in the field. This helps the model learn to classify new medications and manufacturers of medication boxes that do not look like those we have previously trained against.

This is a really exciting time for the healthcare sector with AI innovation unlocking a host of new delivery models and significant potential to improve care and patient outcomes. At Centred Solutions we are always pushing boundaries to ensure our FLOWRx solutions are using the latest innovations so they can continue to optimise performance.

What's next 

We’re seeing incredible progress from technology companies when it comes to applying AI in a range of areas including medical imaging, drug discovery and tackling diseases. The dispensing of medication should be no different. That’s why we are now exploring further uses for machine learning models in our FLOWRx workflow. These ideas include:

  • Learning where to place the label on the medication pack so that in the future it can be could be placed specifically over the "affix dispensing label here" area.
  • Determining the height of the medication pack from the image captured by one of the barcode scanners on the automated labelling unit.
  • Learning to control a counting device to consistently count different kinds of pills.

All of this of course will take time and input, but it showcases Centred Solutions commitment to continue to push boundaries when it comes to innovation and providing sophisticated and reliable automated dispensing solutions for community pharmacy.

Get Started With Pharmacy Automation!

At Centred Solutions, our advice to pharmacies who are not considering automating their original pack dispensing process would be to act now to champion the change.

Book a free consultation with us today to discuss how you could automate your original pack dispensing process.