Pharmaceutical Industry

Stringent regulation and process complexity limit optimization potential for pharmaceutical production chains. See how our non-invasive, AI-powered solution can improve productivity.

 

Yield prediction for biological process

Context:

Biological processes consist in producing a desired molecule by cultivating living organisms. Those organisms typically take as input nutrients, water and air and release as an output the desired molecule.

Deployment / Implementation:

Our solution is operational following a 3-stepped implementation process:

  1. Train initial algorithm on past process data (including both parameters and yield measures)

  2. Connect algorithm to production data sources

  3. Train production staff and monitor performance

Our biological process control solution can be up and running in 2 months.

​Problematic:

Many factors influence the ability of the organisms cultivated to produce the desired molecule, making biological processes difficult to model, and therefore, control, leading to a high variability in associated yields.

Solution:

We provide biological process managers with a yield prediction solution, able to determine future yields from process parameters (quantity and frequency of nutrient, water, air input, temperature and pH of the solution) and suggest adjustments to maximize output.

​Problematic:

While no batch is released if it is not homogeneous, the optimal parameters (mixing time and rotating speed) required to reach this homogeneity depends on the

Optimal parameters estimation for solid forms mixing

Context:

In most pharmaceutical production processes, mixing solid

forms, or powders, is necessary to obtain the final product.

Often, active principles (APIs) are mixed with excipients

in large rotating tanks to obtain the desired concentration.

In order to ensure every pill or tablet has the right

concentration, the mix must be homogeneous.

properties of the powders (quantity, granulometry, density) and can be highly variable. This translates into longer than necessary mixing processes that can add up to 30% to a batch production time.

Solution:

We offer a tool taking as input the specific characteristics of the powders to mix for each batch and computing the optimal mixing time and rotation speed to reach homogeneity as soon as physically possible.

Deployment / Implementation:

We can implement this solution in c.2 months.

Root cause analysis on presence of glass particles in injectable vials

Context:

The production of injectable drug involves washing, sterilising, filling and sealing thousands of vials daily. During this process, glass particles can detach from the container and integrate the solution, which is subsequently discarded.

Problematic:

It is often challenging for injectable manufacturers to rapidly identify the root cause of this defect, either from their process itself or from exogenous factors (quality of vials supplied, operators performance). Manufacturers’ ability to take corrective action is consequently limited.

Solution:

We provide injectables manufacturers with a solution linking every available process data to the results of the glass particle inspection to swiftly identify parameters deviations that could explain the presence of glass in the vial. Our

solution enables manufacturers to identify defects imputable to their process, and hereby those imputable to exogenous factors.

Deployment / Implementation:

We can implement this solution in c.3 months.