AI powered Tablet Design an update by TaBlitz

Over the weekend (6/1/24) a vast expansion to the capabilities of the TaBlitz AI engine was released. TaBlitz introduces three new prediction models to Guided Mode, further simplifying the drug development process for Oral Solid Dosage forms. Learn more about these features in this edition of "The Excipients Week"

Prediction Model Overview

Ratio - Constrained by Aspect Ratio

The Ratio prediction model is the original prediction model that came with the first release of TaBlitz Guided Mode. It allows a user to control the different ratios of a given tablet. "Length to Width", "Cup to Thickness", "Thickness to Width" and "End Radius to Width". When associated to "Round" shapes the "Width" is the "Diameter" of the tablet. By defining the ratios, a user controls the physical proportions of what the prediction model produces for results. Controlling the appearance of your design is extremely important especially when in context of patient acceptance. Having a tablet shape that is familiar and appealing is preferred.

As you adjust the ratios on the Ratio tab (Image 1) you are provided with a live preview that helps you visualize the constraints being defined. This preview allows for fine tuning and a comprehensive method for adjustments as you go.

Die Profile - Targeting Existing Tooling

This prediction method allows for users to predict tablets based on die profile constraints. With this method also comes the ability to define a die profile and also a library for these profiles to be stored. This allows users to define and select an array of different profiles to predict to a target volume (Image 2).

Once the desired profiles are selected, TaBlitz AI will predict the most optimal cup configuration for each associated cup type that best suites the die profile. This method allows for users to target existing tooling. This is helpful as it allows customers to utilize dies already setup in a machine and also allows for maximizing the life of your die inventory.

?Volume / Surface Area - Single Value

The new Volume and Surface Area prediction models we are going to present together, as they rely on very similar principles and have equal capabilities. These prediction methods require minimal inputs. As the names imply the Volume model only requires volume to be specified and the Surface Area model requires surface area. You can also constrain the shapes that these models will predict against. What makes these models so powerful is how they were trained. These models utilize over 130,000 tablet designs to predict the most optimal associated geometry based on volume or surface area inputs. It is quite fascinating to watch a single number be fed into our AI and then produce an entire tablet in return.

Comprehensive Data Output

When we first discussed the extent of the additions to TaBlitz AI and Guided Mode we knew that we were going to produce so much data and that it would be difficult for users to digest so many data points. With this release of Guided Mode, we are also releasing interactive reports (Image 3).

Not only do these charts summarize all the critical information but they are also interactive in that you can target a given chart element and hover to see the details (Image 4).

Taking it a step further, you can click on the chart element and it will take you to a live preview of that dataset (Image 5).

The Data

All data provided in guided mode is in a pareto format to easily identify top ranking designs.

Total Design Score

The first tab is a stacked column chart that summarizes the design score calculated for each tablet. This is the same score in the main design window but now you can see it across the array of predicted tablets (Image 6).

Tablet MFG / Swallowability / Tool MFG / Tool Life

The next 4 tabs are an individual break down of each index that drives the design score (Image 7).

Max Compression Force

For all the designs predicted TaBlitz calculates the max compression force for each associated geometry. With this chart you can now review which design will yield the highest compression force (Image 8). Note that this is based on the tool steel selected on the tooling tab.

Surface Area

TaBlitz also provides a chart of the Surface Area for all predicted tablets. This helps when you are trying to maximize the surface area around a set of design constraints (Image 9).

Volume

When using the Surface Area prediction model, TaBlitz also provides a chart for volume. This helps you compare which shape closest fits your required volume, while also meeting the need of the surface area you require (Image 10).

Summary

In summary, this release provides many new features and functionality intended to speed up the drug development process. With this release our goal was to reduce the iterations through the development process and at the same time provide the formulators and staff closer to the manufacturing floor with a faster approach to adjustments or product change overs while maximizing the utilization of tooling.

If you are interested in seeing these features in action, please watch the tutorial posted on the Tablitz YouTube channel.