Active Learning in DeepRod for Granular Behavioral Phenotypes

With only 0.5 percent of the annotation effort, our model detects the same medication-induced behavioral changes as a fully trained reference model.
by
Valeriia Dragan

Our paper,“Active Learning in DeepRod Enables Efficient Discovery of Granular Behavioral Phenotypes,”was published in *Current Research in Behavioral Sciences *.

What is the biggest struggle in applied ML research today? Validating methods in a way that shows real-life impact, not just significant metrics on benchmarks. In healthcare and drug development this is critical. We want the AI solution that doesn't just win benchmarks but actually helps patients, and we see that from the real outcomes: better survival rates, earlier diagnoses, fewer compounds failing late in development, real pharmacological effect.

This is exactly what we highlight in our recently published paper on advanced active learning for automated behavioral analysis. We have a computer vision system that automates rodent behavioral phenotyping in pharmacological studies.

The bottleneck of manual annotation for studies generating 14–24 hours of video per animal is already solved by DeepRod's core module, which automatically predicts behaviors at each timepoint. However, automated models still require training data, and active learning is the core approach we use to minimize this initial labeling effort for model training.

Advanced active learning methods solve the annotation burden by selecting only the most informative samples for labeling. We didn't just show it improves classification accuracy, we showed that the model trained on just 0.5% of a full labeling set, still detects statistically significant drug-induced behavioral changes - the same effects the fully-trained reference model detects. Including fine-grained distinctions like Apathy versus Sleep under Clonidine treatment, which coarser methods collapse entirely.

None of this happens without the collaboration. Boehringer Ingelheim's scientists didn't just provide data, they defined what a meaningful result looks like in a preclinical setting. That domain knowledge shaped the method.

This is what we build at Merantix Momentum: ML that's validated against outcomes, not just benchmarks, built with the people who understand the domain well enough to define what outcomes matter.

Subscribe to the Merantix Momentum Newsletter now.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

More articles

The latest industry news, interviews, technologies and resources.

Our publications

Discover the latest publications from our research team and more

Beyond Pilots: Why AI Strategizing in 2027 Demands a Forward-Deployed Mindset

Discover why AI pilots are no longer enough in 2027 – and how a Forward-Deployed mindset creates real value.

AI is now part of our culture. Communities will determine what comes next

How AI is transforming our society, and why it is not technology alone, but communities, values, and responsibility that will determine the future we shape with AI.

The Agentic Data Stack

​Join Merantix Momentum, ClickHouse, and Langfuse at the Merantix AI Campus in Berlin for an evening exploring what it actually takes to build, monitor, and scale agentic AI systems in production. Registration closes on September 2, 2026.

Building Better Medicines: Exploring AI-Driven Compound Optimization

An expert interview on research into AI-driven drug optimization.

Active Learning in DeepRod for Granular Behavioral Phenotypes

Our paper,“Active Learning in DeepRod Enables Efficient Discovery of Granular Behavioral Phenotypes,”was published in *Current Research in Behavioral Sciences *.

What is the biggest struggle in applied ML research today? Validating methods in a way that shows real-life impact, not just significant metrics on benchmarks. In healthcare and drug development this is critical. We want the AI solution that doesn't just win benchmarks but actually helps patients, and we see that from the real outcomes: better survival rates, earlier diagnoses, fewer compounds failing late in development, real pharmacological effect.

This is exactly what we highlight in our recently published paper on advanced active learning for automated behavioral analysis. We have a computer vision system that automates rodent behavioral phenotyping in pharmacological studies.

The bottleneck of manual annotation for studies generating 14–24 hours of video per animal is already solved by DeepRod's core module, which automatically predicts behaviors at each timepoint. However, automated models still require training data, and active learning is the core approach we use to minimize this initial labeling effort for model training.

Advanced active learning methods solve the annotation burden by selecting only the most informative samples for labeling. We didn't just show it improves classification accuracy, we showed that the model trained on just 0.5% of a full labeling set, still detects statistically significant drug-induced behavioral changes - the same effects the fully-trained reference model detects. Including fine-grained distinctions like Apathy versus Sleep under Clonidine treatment, which coarser methods collapse entirely.

None of this happens without the collaboration. Boehringer Ingelheim's scientists didn't just provide data, they defined what a meaningful result looks like in a preclinical setting. That domain knowledge shaped the method.

This is what we build at Merantix Momentum: ML that's validated against outcomes, not just benchmarks, built with the people who understand the domain well enough to define what outcomes matter.

Oops! Something has gone wrong.
Oops! Something has gone wrong.
Oops! Something has gone wrong.
Oops! Something has gone wrong.
Oops! Something has gone wrong.

Discover more whitepapers

Towards Tabular Foundation Models

About the status quo, challenges and opportunities

Leveraging the EU AI Act to your advantage

Using the EU AI Act to your advantage

Data-driven to the drug of tomorrow

Opportunities and barriers of AI in a GxP world.

The AI Canvas: Our tool for project evaluation

Discover the AI Canvas!

Data-driven to the drug of tomorrow

Opportunities and barriers of AI in a GxP world.

Towards Tabular Foundation Models

About the status quo, challenges and opportunities

Subscribe to the Merantix Momentum Newsletter now.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.