Customer Case Study: Reaction modeling using AI/ML on HTE data for a large BioPharma company

About the customer

A leading top-5 biopharmaceutical company with annual revenue exceeding $58 billion, recognized for its innovative R&D pipeline addressing critical areas such as immunology, neurology, and oncology

Executive summary

Context

At a large biopharma, Process R&D team carries out High Throughput Experiments (HTE) to maximize yield of chemical reactions

>80% of the experiments done to identify the right combination (of base, solvent, ligand) for a given set of reactants yield <10%

Objective

Mine the historically generated HTE data to identify low-performing & high-performing substrates

Build models that predict & prescribe the experiments to be conducted

Results

Deployed predictive and prescriptive models that are used by chemists to complement their intuition

Automated pipelines for HTE data pre-processing & Interactive visualizations to slice & dice the data

Outcomes

Automated insights from HTE data using a custom-built analytics tool in conjunction with LLMs

Cost, Material & Labor savings from avoiding unnecessary experiments

Problem

One of the major problems is avoidable experiment expenditure as more than fifty percent of the experiments yield less than ten percent.

Multiple experiments to be done to identify right set of conditions

Identifying suitable reaction conditions for yield optimization is time-consuming

Underutilized corpus of data though it is a well-studied reaction

How did we solve it

Outcomes 1/2

Outcomes 2/2

Discover our offerings across the biopharma value chain

Learn more about our platform, solutions, and services