Webinar Recap: Advancing Translational NHP Research Through Digital Behavioral Phenotyping

For those unable to attend the live session, this recap summarizes the scientific themes, methodological approaches and representative applications discussed during the webinar.

From Observational Scores to Quantitative Behavioral Endpoints

NHP behavioral assessment has traditionally relied on manual observation, video review and experience-based scoring. While these approaches remain useful in many research settings, they may be time-intensive and subject to inter-observer variability. They can also provide limited temporal and spatial resolution when subtle or prolonged behavioral changes need to be characterized.

The scoring problem is partly arithmetic. A hand-scored ethogram samples behavior in fixed intervals, so a short event that falls between intervals may be missed or counted twice, and two observers watching the same clip can assign different scores without either being careless. Continuous tracking changes what can be asked of the data.

During the opening session, James Song, General Manager and Co-founder of Prisys Biotech, discussed the increasing importance of functional behavioral endpoints in CNS, pain and other translational research programs. As therapeutic mechanisms become more complex, behavioral data can complement established readouts such as PK/PD, imaging, pathology and molecular biomarkers.

NHP models are particularly valuable for translational research because of their anatomical, physiological and functional similarities to humans. To make full use of this translational potential, however, behavioral observations need to be collected and interpreted with appropriate objectivity, resolution and reproducibility. Digital behavioral phenotyping provides a framework for characterizing behavior continuously and quantitatively while minimizing the need for physical markers or repeated handling.

NHP BehaviorAtlas: A Framework for High-Fidelity 3D Behavioral Analysis

A central part of the webinar focused on NHP BehaviorAtlas, a digital behavioral analysis framework developed to support high-resolution characterization of NHP movement and behavior.

The approach combines multi-view video acquisition, markerless three-dimensional pose estimation, behavioral kinematics, automated behavior classification and data visualization. By reconstructing movement and posture in three dimensions, the framework can help researchers characterize behavioral features that may be difficult to assess consistently through conventional observation alone.

The technical discussion covered several capabilities:

• Markerless 3D pose estimation: Reconstructing key anatomical landmarks without attaching physical markers, allowing behavior to be assessed under more naturalistic conditions and supporting the principles of refinement within the 3Rs framework.
• Behavioral classification: Identifying and quantifying behaviors such as locomotion, grooming, scratching, posture changes and activity patterns over defined observation periods.
• Behavioral kinematics: Measuring movement-related parameters, including trajectory, velocity, posture, gait characteristics and limb coordination.
• Facial expression analysis: Exploring the use of three-dimensional facial modeling to characterize subtle changes associated with pain, discomfort or other treatment-related responses.

The webinar also addressed the importance of study-specific validation. Algorithmic outputs should be evaluated against appropriate human-reviewed references and interpreted in the context of the animal model, study design and research question.

Validation deserves the emphasis it received. A pose model trained on one camera layout, lighting condition or body size does not transfer automatically to another, and agreement with human scoring has to be demonstrated for the behaviors that matter to the study, not only for overall tracking accuracy.

Representative Applications in Translational Research

The webinar presented several application areas in which digital behavioral phenotyping may support NHP research.

Pain and Pruritus Models

Automated analysis of scratching behavior can help quantify the frequency, duration and temporal distribution of scratching episodes in cynomolgus macaque studies. These measures may provide additional efficacy endpoints for anti-pruritic and anti-inflammatory programs, particularly in long-term studies where manual video review can become resource-intensive.

Three-dimensional tracking may also assist in distinguishing scratching from other activities, such as grooming or routine movement, by considering limb trajectory, body posture and contact with specific anatomical regions.

Neurological Disease Models

In Parkinson's disease, stroke and other neurological disease models, movement-related changes may involve subtle alterations in gait, posture, limb use, coordination and spontaneous activity.

Quantitative kinematic analysis can help characterize these changes over time and may support evaluation of functional recovery following treatment. Longitudinal measurements within the same animal can also provide a more detailed view of disease progression and response dynamics, while helping researchers account for inter-animal variability.

Pain- and Discomfort-Related Facial Responses

Facial expression analysis is being explored as a complementary approach for assessing pain and discomfort in NHP studies. Multi-view imaging may help capture dynamic changes that are difficult to evaluate from a single viewing angle.

Such analyses should be considered alongside established clinical observations, physiological measures and other study endpoints rather than interpreted in isolation. Their value depends on appropriate model characterization, annotation quality and validation in the relevant research context.

Safety and Tolerability Assessment

Continuous or repeated behavioral monitoring may also provide useful information for safety pharmacology and tolerability studies. Changes in spontaneous activity, posture, locomotion, grooming or other behavioral patterns could serve as supportive indicators of treatment-related effects.

Digital behavioral data may complement clinical observations and other safety assessments by providing a more granular temporal record of behavioral changes. Sedation, for instance, often appears first as a shift in activity distribution across the light cycle, before it becomes obvious in a spot check.

Integrating Behavioral Data within a Translational Research Workflow

The final session discussed how quantitative behavioral endpoints can be incorporated into a broader NHP research workflow at Prisys Biotech.

Behavioral phenotyping can be combined with NHP disease models, in vivo imaging, PK/PD analysis, pathology and molecular biomarkers to provide complementary perspectives on treatment response. This multimodal approach may help researchers connect molecular exposure and target biology with functional outcomes.

A representative translational workflow can be expressed as:

Behavior → PK/PD → Imaging → Pathology → Biomarkers

Each modality addresses a different aspect of the research question. Behavioral analysis contributes functional and phenotypic information, while PK/PD, imaging, pathology and biomarker data help characterize exposure, mechanism, tissue-level effects and biological response.

The scientific value of this integrated approach lies in combining complementary evidence. Behavioral readouts rarely replace established endpoints; their contribution is a fuller interpretation of disease biology and treatment effects.

Access the Webinar Recording and Resources

To review the technical presentations and application examples, please access the following resources:

• Watch the Full Webinar Recording
• Download the Presentation Slides
• Learn More about NHP Translational Research at Prisys

About Prisys Biotech

Prisys Biotech is a translational research organization specializing in non-human primate models and preclinical evaluation. Its capabilities include NHP disease models, CNS research, behavioral studies, in vivo imaging, PK/PD analysis and integrated efficacy assessment.

By combining model development with quantitative behavioral and imaging approaches, Prisys supports the generation of structured, reproducible and translationally relevant evidence for global biopharmaceutical research programs. Programs that need a behavioral readout defined before dosing begins can raise the question during study design, when camera layout, observation windows and endpoint definitions are still open.

Contact Prisys Biotech

Related Reading

• Global Webinar | Quantifying Complex NHP Behaviors: Introducing NHP BehaviorAtlas For High-Fidelity 3D Behavioral Phenotyping
• AI-Powered 3D NHP Behavioral Analysis | Markerless Behavioral Phenotyping
• Advancing Parkinson's Disease Research With The NHP Model
• Prisys Biotechnologies' NHP Research And AI-Based Behavioral Analysis Platform