Using Deep Learning in Clinical Pharmacology for a Global Pharmaceutical Giant

  • Published: March 29, 2024



Explore how a leading pharmaceutical organization reduced the time and effort to identify PK/PD parameters with Reveal’s deep learning solution.


Determining Optimal Dosage for Patients

Our client used pharmacokinetic and pharmacodynamic (PK/PD) modeling in drug discovery to determine optimal dosing for patients. This was a very time-consuming process for our client’s expert clinical pharmacologists. They were looking for a solution that would support rapid experimentation and help speed up the drug discovery process.


Reveal’s Approach

Reveal experimented with several state-of-the-art deep learning models and evaluated the performance metrics across experiments. Powered with these insights, Reveal developed a deep learning model to infer coefficient estimates of PK/PD systems from drug response time series data. The model used a custom-built synthetic data generator to generate input/output data for viable regimes in PK/PD coefficient space.


Experimenting & Iterating to Success

The deep learning model reduced the time and effort taken by our client’s expert modelers to identify PK/PD parameters. It empowered their modeling and simulation teams with self-service workflows which facilitated rapid experimentation.