Beating High-Throughput with ML: Optimizing Categorical Variables in Cross-Coupling Chemistry for Ghent University

SuntheticsML matched the performance of a 768-run high-throughput campaign—using just 48 experiments. The platform identified optimal solvent, base, and catalyst combinations with up to 94% experiment reduction.

94%

Fewer experiments

48 vs 768

Experiments vs HTE

2

Iterations to optimum
Key Findings:
  • 94% fewer experiments, same optimal results—machine learning outperformed expensive automation.
  • Categorical ML breakthrough: SuntheticsML optimizes directly from labels without complex parameterization or one-hot encoding.
  • New scientific insight: Base identity—not catalyst—was the key variable, guiding smarter future experimentation.
  • Lower barrier to entry: SuntheticsML achieves elite-level results without the infrastructure or cost of high-throughput robotics.
  • Repeatable & reliable: All five ML seeds converged to the correct optimum, confirming robustness.
Company Name

Ghent University

Category

Pharmaceutical Chemistry

Published Date

April 19, 2025

Highlighted features / use cases
Categorical Optimization
Active Learning
Client

This study was conducted by a joint team from Ghent University and Sunthetics. The goal: optimize a Suzuki–Miyaura cross-coupling reaction using real-world, industrial-grade datasets—where categorical variables like solvent and catalyst dominate the design space​.

Challenge
  • Identify the best combination of catalyst, solvent, and base for a complex cross-coupling reaction.
  • Achieve this using minimal experimentation.
  • Tackle the high-dimensional complexity of categorical variables, which are often not easily optimized by standard ML or DoE tools.
  • Avoid the 768 experiments typically required by High-Throughput Experimentation (HTE) to explore the full design space.

Goal
  • Identify the best combination of catalyst, solvent, and base for a complex cross-coupling reaction.
  • Achieve this using minimal experimentation.
  • Tackle the high-dimensional complexity of categorical variables, which are often not easily optimized by standard ML or DoE tools.
  • Avoid the 768 experiments typically required by High-Throughput Experimentation (HTE) to explore the full design space.

Approach & Solution
  • SuntheticsML was deployed using proprietary:
    • Supervised Learning (SL) and
    • Active Learning (AL) algorithms
  • No variable encoding or parameterization was required—SuntheticsML directly optimized categorical inputs like catalyst identity.
  • Each optimization was initiated with 36 randomly selected experiments.
  • The platform iteratively recommended 6 experiments per cycle: 5 likely to yield maximum performance + 1 random for exploration.
  • Five independent "seeds" were used to evaluate repeatability and robustness.

Results & Metrics
  • HTE benchmark:
    • 768 experiments required to explore full combinatorial space and find optimal conditions.
  • SuntheticsML outcomes:
    • Average performance:
      • 6 iterations, 72 experiments, 91% experiment reduction
    • Best-case scenario:
      • 2 iterations, 48 experiments, 94% reduction
    • Worst-case scenario:
      • 9 iterations, 84 experiments, 89% reduction
  • Insight generated:
    • Variable importance analysis showed the base had the strongest effect on yield—surprising the research team and shifting future optimization focus away from catalysts and solvents​.
Why it worked

"SuntheticsML matched the output of a fully combinatorial HTE campaign in a fraction of the time and cost—while revealing insights HTE didn't, like the true variable driving reaction performance."

Looking for similar results?
Contact us today

Contact us