Behavioral chaos in Alzheimer’s disease mice decoded by machine learning

Cutting-edge machine learning reveals hidden patterns in the behavior of mice in Alzheimer’s disease, opening the door to innovative treatments targeting neuroinflammation.

Study: Machine learning reveals prominent spontaneous behavioral changes and treatment efficacy in humanized and transgenic models of Alzheimer’s disease. Image credits: nobeastsofierce / Shutterstock

This is evident from a recent study published in the journal Cell reportsresearchers used the machine learning (ML)-based Variational Animal Motion Embedding (VAME) segmentation platform to analyze behavior in mouse models of Alzheimer’s disease (AD) and test the effect of blocking fibrinogen-microglia interactions. They found that AD models exhibited age-dependent behavioral disruptions, including increased volition and impaired habituation, largely prevented by reducing neuroinflammation, with VAME outperforming traditional methods in terms of sensitivity and specificity.

Background

Behavioral changes, which are central to neurological disorders, are complex and difficult to measure accurately. Traditional task-based tests provide limited insight into disease-induced changes. However, advances in computer vision and ML tools, such as DeepLabCut, SLEAP, and VAME, now enable the segmentation of spontaneous mouse behavior into postural units (motifs) to reveal sequence and hierarchical structure, allowing for scalable, unbiased measurements of brain dysfunction are provided.

AD, characterized by amyloid accumulation prior to tau pathology and neurodegeneration, often presents subtle behavioral or neuropsychiatric changes such as agitation, depression, and loss of motivation decades before the onset of dementia. These early changes provide a promising window to study AD pathogenesis and therapeutic interventions.

Humanized amyloid precursor protein (App) knockin and transgenic APP mouse models replicate key AD hallmarks, such as amyloidosis and neuroinflammation. Despite these advances, analyzing non-task-oriented spontaneous behavior in AD models has remained technically challenging until the emergence of ML-based behavior analysis methods. Recent refinements of VAME enable the integration of advanced kinematic and network analyses, providing deeper insights into behavioral organization and disease progression.

In the current study, researchers enhanced their VAME ML pipeline to analyze spontaneous behavior, disease progression, and treatment effects in AD mouse models to validate the model’s sensitivity for detecting neuroinflammation-related therapeutic outcomes.

About the study

This study examined behavior and pathology related to Alzheimer’s disease using two mouse models: AppNL-GF and 5xFAD. The AppNL-GF mice, which express humanized amyloid-β (Aβ) with familial AD mutations, were assessed at young (six months), middle-aged (13 months), and advanced age (22 months).

Behavioral experiments included spontaneous activity in an open arena and assessment of spatial memory using the Morris water maze. Histological analysis examined amyloidosis and gliosis. The 5xFAD mice, which overexpress human App and presenilin 1 (PS1) with multiple familial AD mutations, were studied after nine months. To evaluate a potential therapeutic intervention, 5xFAD mice were crossed with Fggγ390-396A mice, a model that targets fibrinogen-microglia interactions.

Behavioral data was captured using DeepLabCut, a video-based pose estimation tool, and analyzed using VAME ML that identifies different behavioral motifs and sequences. Motive use, transitions, and behavioral community structures were examined to identify disease-related changes.

Spatial learning deficits and increased behavioral randomness were observed in AppNL-GF mice, while 5xFAD mice showed significant motif changes, with females showing increased sensitivity to these changes. Many of these abnormalities were restored by the Fggγ390-396A intervention. Classifier analysis was used to compare the sensitivity and specificity of VAME with conventional open field statistics. A comparison with keypoint-MoSeq was also performed to validate the results of VAME.

Results and discussion

Old AppNL-GF mice (22 months) showed mild memory impairment in the Morris water maze, along with severe amyloidosis and gliosis. Using VAME, age- and genotype-related changes in spontaneous behavior were identified. These mice showed significant changes in the use of eight of the thirty identified behavioral motives, including walking, rearing and exploring. Higher order behavioral communities were also disrupted, with reduced habituation, abnormal sensitization, and increased randomness in behavior. Motive transition analysis revealed reduced predictability and premature transitions from active to static behavior.

In 5xFAD mice, 17 behavioral motifs were found to be significantly affected, with significant abnormalities in transitions and reduced behavioral predictability. Blocking fibrinogen-microglia interactions using the Fggγ390-396A mutation partially or completely rescued these behavioral abnormalities, restoring motif usage, rates, and transitions. Therapeutic effects were especially evident in rapid exploratory and ambulatory behavior. Importantly, the therapeutic intervention demonstrated disease-specific effects, as fibrinogen blockade did not alter behavior in non-AD controls.

Classifier analysis showed that VAME provided greater sensitivity and specificity than conventional open-field metrics in detecting behavioral differences between genotypes and therapeutic outcomes. Both VAME and keypoint MoSeq reliably identified disease-associated behavioral changes, but the VAME results were more comprehensive and specific. These findings underscore the utility of VAME in addressing the core disorganization of behavioral sequences observed in AD models.

Together, these results highlight VAME as a robust tool for quantifying complex behaviors and assessing preclinical disease phenotypes and therapeutic interventions with superior specificity and scalability compared to conventional methods.

Furthermore, the findings highlight fibrinogen-microglia interactions as a potential therapeutic target. However, the study did not explicitly assess cognitive functions, brain regions or neural systems. It remains unclear whether spontaneous behavioral disturbances directly reflect cognitive decline and could potentially serve as biomarkers.

Conclusion

In conclusion, unbiased ML approaches such as VAME enable rigorous quantification of disease-induced behavioral changes, improving construct and predictive validity assessments in mouse models of neurodegenerative diseases. The integration of behavioral community analysis and transition networks provides a scalable and sensitive framework for identifying disease-related disruptions. The method could potentially increase the translatability of preclinical testing by providing sensitive and scalable tools to evaluate disease progression and therapeutic interventions, addressing a critical gap in neuroscience research.