Automated Behavioural Monitoring Systems in Metabolic Cage Studies: A Scientific Literature Review

Abstract

Automated behavioural monitoring systems have changed metabolic cage studies by allowing continuous, objective, and high-throughput tracking of animal behaviour and physiology. This review consolidates findings from 30 key studies on how these systems operate, their applications, and their limitations. The technologies used include infrared beam-break sensors, RFID tracking, video analysis, and combinations of different sensors. These systems are mainly used in circadian biology, drug discovery, toxicology, disease modelling, and behavioural phenotyping . They offer clear benefits over manual observation, including reduced observer bias, on going data collection, and the ability to monitor animals in social groups. However, challenges remain, including accuracy during high activity, identifying individuals in groups, managing large amounts of data, and high costs. The field is moving quickly toward better computer vision, machine learning, and non-invasive monitoring, which could improve both research quality and animal welfare. Future developments will integrate multimodal data streams, cloud computing, and real-time analytics to enhance precision, scalability, and reproducibility in experiments..

Keywords

Introduction

A key technique in preclinical biomedical research, metabolic cage studies allow for a thorough evaluation of energy balance, metabolic function, and behavioural patterns in lab animals. Manual observation, which is labour-intensive, prone to observer bias, and restricted to discrete time periods, has been a major component of traditional approaches to behavioural monitoring in these circumstances. This environment has completely changed with the introduction of automated behavioural monitoring systems, which provide continuous, objective, high-resolution data gathering capabilities that go well beyond the capabilities of traditional techniques [1], [2], [3].

 To simultaneously record several aspects of animal behaviour and health, automated monitoring systems combine a variety of sensor technologies, such as infrared beam-break detectors, radiofrequency identification (RFID), video analysis, and physiological sensors [2], [4]. These technologies allow researchers to track variables in real time and over long periods of time, including body temperature, breathing patterns, feeding and drinking habits, locomotor activity, and sophisticated behavioural repertoires [1], [5], [6]. Unprecedented insights into the dynamic interaction between behaviour, metabolism, and physiological control can be obtained by combining automated monitoring with metabolic data [7], [8].A thorough analysis of the creation and use of home cage monitoring systems has shown how quickly technology is developing and how its use is growing in a variety of research fields [9], [10]. In psychiatric and neurological research, where naturalistic behavioural evaluation in familiar settings may better capture clinically relevant traits, these methods hold special potential for enhancing translational utility [11]. Additionally, by facilitating more effective experimental designs, lowering animal numbers through greater statistical power, and reducing stress related to handling and novel test conditions, automated monitoring is consistent with the 3Rs principles (Replacement, Reduction, Refinement) [12], [13]

With a focus on technological approaches, research applications, and system constraints, this review of the literature summarizes the state of the art for automated behavioural monitoring systems in metabolic cage investigations. This review attempts to give researchers a thorough grasp of the current state of the area and guide future technology development and experimental design decisions by looking at 30 extremely pertinent works including system development, validation, and application.

TECHNOLOGICAL METHODS AND APPROACHES

2.1 Beam-Break Infrared Systems

 One of the most popular methods for automatic activity monitoring in metabolic cages is infrared (IR) beam-break technology. These systems use arrays of infrared laser beams placed at different heights inside the cage, and activity counts are triggered when the animal moves and interrupts these beams [3], [7], and [14]. Multi-axis infrared beam arrays (x, y, and z axes) are used by commercial systems such the LabMaster (TSE Systems) and PhenoMaster to differentiate between vertical activity (raising) and ambulatory movement [3], [29].

The basic idea is to create a grid of detecting zones by sending infrared light from transmitters on one side of the cage to receivers on the other [2]. Specific beams are interrupted as an animal walks through the cage, and the frequency and pattern of these interruptions are noted to measure the animal's level of activity [7]. Triple-beam infrared technology is used in advanced implementations to improve spatial resolution and allow for more accurate movement pattern tracking [29]. Depending on the particular design and research needs, these systems' temporal resolution usually varies from continuous sampling to measurements every 15 minutes [3]. IR beam-break systems have a number of benefits, such as minimal interference with the typical cage environment, ease of integration with metabolic monitoring equipment, and comparatively low cost [2]. Researchers can account for activity-related fluctuations in metabolic measures by using the technology to synchronize activity data with metabolic parameters including carbon dioxide production and oxygen intake [2], [7]. Nevertheless, these systems are unable to capture complex behavioural repertoires beyond simple locomotion and rearing, and they offer rather limited spatial information when compared to video-based methods [14].

2.2 Tracking using RFID

A significant drawback of many automated monitoring techniques is addressed by radiofrequency identification (RFID) technology, which allows individual animal identification and tracking inside group-housed environments [1], [13], [21]. This strategy is demonstrated by the Rodent Big Brother project, which uses subcutaneous RFID transponders in conjunction with a baseplate that has a 3x4 array of antennae placed underneath the home cage [1]. Animals' RFID tags are picked up by various antennae as they walk across the cage floor, allowing movement patterns to be reconstructed and individual activity levels to be measured [1].The advantages of communal housing are preserved while subject-specific data gathering is made possible by RFID-based systems' special capacity to monitor individual animals inside social groupings [1], [13]. Studies looking at social behaviour, dominance hierarchies, or individual reactions to experimental interventions in a social setting will find this very useful [13]. An extra physiological parameter is provided by the technology's ability to measure subcutaneous temperature via the RFID transponder [1]. Another well-known RFID-based platform is the IntelliCage system, which has corner-mounted operant conditioning chambers with RFID readers that recognize certain animals as they approach drinking or feeding stations [21]. In addition to basic activity monitoring, this approach allows for automated evaluation of learning, memory, and decision-making behaviours [21]. To make it easier to analyse the complicated datasets produced by these systems, specialized software tools like PyMICE have been created [21].

Despite their benefits, RFID-based systems have certain technological drawbacks, such as sporadic missed detections while moving quickly, restrictions on spatial resolution (tracking is discretized to antenna positions rather than continuous), and comparatively high implementation costs [1]. Although this is usually well tolerated, the need for surgical implantation or subcutaneous injection of transponders also adds an intrusive component [1].

2.3 Monitoring with Video

By using computer vision and machine learning to extract rich behavioural information from video recordings, video-based monitoring systems offer a fast developing frontier in automated behavioural assessment [5], [12], [27]. These systems range from comparatively basic motion detection algorithms to complex deep learning techniques that can identify certain postures and behavioural patterns [5], [18].

Using side-view cameras installed on vivarium cage racks to continually record group-housed mice, the SCORHE (Spontaneous Behaviour in Cages Recorded from Home Environment) system is an example of a practical video monitoring application [12]. The system may be set up to identify particular behaviours using customized analysis pipelines and uses motion detection algorithms to measure overall activity levels [12]. In a similar vein, the Rodent Big Brother project incorporates infrared high-definition cameras (720p, 25 fps) for side-view video collection, allowing for automated vertical activity detection and producing visual records for behavioural validation [1].

Machine learning is used in advanced video analysis techniques to automatically classify behaviours. Through supervised learning on manually annotated training data, trainable vision-based systems can learn to identify complex behavioural repertoires such as walking, grooming, rearing, and social interactions [5]. Automated behavioural phenotyping from home cage video has proven to be a remarkable capability of deep learning architectures, especially convolutional neural networks [18], [25].

The focus of recent advancements is on long-term, continuous monitoring capabilities. Deep phenotyping systems can monitor behavioural patterns across weeks to months, recording developmental shifts, circadian rhythms, and subtle behavioural changes linked to the course of a disease or treatment interventions [25]. Physiological factors like heart rate, respiration rate, and body temperature can be extracted from video data using multi-camera configurations that include both top-down and frontal perspectives, as well as thermal and near-infrared imaging [30].

The flexibility to examine recordings for actions that were not initially targeted and the unmatched depth of behavioural information provided by video-based techniques [12], [18]. However, they produce enormous amounts of data that need a lot of storage and processing power, and automated analysis algorithms need to be carefully validated. They may also have trouble in situations like low lighting, occlusion by nesting material, or overlapping animals in group-housed environments [12], [30].

2.4 Integrated and Multimodal Systems

In order to capture complementary elements of behaviour and physiology, modern automated monitoring platforms are increasingly using multi-modal techniques, incorporating different sensor types [1], [6], [30]. These integrated systems provide more thorough and reliable datasets by combining the advantages of several technologies while reducing their respective drawbacks [14].The PhenoMaster system, which combines IR beam-break activity sensors with indirect calorimetric (O? consumption, CO? production), food and water intake sensors, and body weight tracking on a single platform, is a prime example of commercial multi-modal integration [29]. This integration facilitates the study of intricate metabolic behavioural relationships by allowing researchers to concurrently evaluate metabolic rate, energy balance, activity patterns, and feeding behavioural [7], [29]. By combining several non-invasive monitoring techniques, emerging integrated home cage designs seek to replace invasive telemetry [30]. These systems use near-infrared cameras to track activity in low light, thermal imaging to measure body temperature, and video-based algorithms to extract vital signals like heart rate and breathing rate [30]. Combining frontal and top-down camera perspectives improves resistance to occlusion and allows for more precise physiological parameter extraction [30].

Another integrated technique is the Cage View system, which was created especially for phonotypical research purposes and combines automated feeding control and monitoring with video recording capabilities [24]. These technologies facilitate research on circadian feeding patterns, dietary preferences, and metabolic regulation by enabling exact adjustment of feeding schedules while continually monitoring  behavioural  reactions [24]. Data synchronization, storage, and analysis become more complicated as a result of multi-modal integration [14], [30]. To manage the many data streams produced by numerous sensors, effective implementation necessitates careful consideration of sampling rates, data formats, and computational infrastructure [30]. Despite these difficulties, integrated systems offer the most complete method of automated behavioural and physiological monitoring in metabolic cage investigations, and they reflect the current direction of the field [14].

APPLICATIONS IN BIOMEDICAL RESEARCH

3.1 Circadian Biology & Chronobiology

Because they allow for the continuous evaluation of  behavioural  and physiological rhythms over long periods of time, automated monitoring systems have become essential instruments in circadian biology research [1], [4], [17]. These devices' 24-hour data collecting capabilities are especially useful for assessing circadian patterns in metabolic parameters, body temperature, eating behaviour, and locomotor activity [1], [7].Certain biochemical pathways are involved in circadian control, according to studies that use automated monitoring. For instance, studies using home cage monitoring showed that Toll-like receptor 2 (TLR2) controls the consolidation of active and inactive phases in mice, with TLR2 mutant animals displaying fragmented activity patterns [4]. It would be challenging or impossible to acquire such results using traditional activity testing or manual observation [4].Complex circadian  behavioural  analysis, such as measuring period length, amplitude, phase, and rhythm stability, is made possible by automated systems [17]. Real-time classification algorithms that can differentiate between several  behavioural  states (such as active, inactive, and feeding) and monitor changes between these states during the circadian cycle are examples of recent advancements [17]. These analytical skills facilitate the evaluation of chronotherapeutic therapies and the study of circadian disturbance in illness models [4], [17]. Energy expenditure, respiratory exchange ratio, and substrate utilization have all shown diurnal patterns when activity tracking and metabolic measures are combined [7], [29]. Researchers may evaluate therapies meant to restore normal circadian-metabolic coupling and investigate how circadian misalignment impacts metabolic health thanks to this comprehensive method [7].

3.2 Safety Pharmacology and Drug Discovery

Automated  behavioural  monitoring systems allow objective, high-throughput evaluation of drug effects on behavioural and physiology, making them useful tools for safety pharmacology and drug development applications [1], [14]. These technologies make it possible to identify potential negative  behavioural  or metabolic effects of medication candidates in addition to the intended therapeutic effects [1], [14]. By evaluating medication effects in the comfortable home environment, automated home cage monitoring provides advantages over traditional  behavioural  experiments in central nervous system (CNS) drug development, potentially enhancing translational validity [1], [11]. Characterization of drug onset, duration of action, and dose-response relationships with great temporal resolution is made possible by the continuous monitoring capacity [1]. This is especially useful for substances that influence circadian rhythms, activity levels, or arousal, as their effects may change during the light-dark cycle [1], [14]. Applications in safety pharmacology include automated monitoring to identify negative effects on body temperature, eating habits, or locomotor activity that may point to toxicity or off-target effects [1], [22]. Continuous monitoring of animals in their home cage setting may highlight minor  behavioural  changes that would go unnoticed in quick observational evaluations [1], [23].  behavioural  abnormalities frequently precede more severe clinical indications, and historical data from home cage monitoring has proven useful in toxicity evaluation [23]. Finding translational digital biomarkers—quantitative  behavioural  or physiological metrics that may be evaluated in preclinical models and clinical populations—is the focus of emerging applications [14]. Complex  behavioural  metrics like activity fragmentation, circadian amplitude, and  behavioural  variability can be extracted using automated monitoring systems and may be used as translational biomarkers for neuropsychiatric and neurodegenerative diseases [14], [25].

3.3 Disease Modelling and Phenotyping

For the  behavioural  phenotyping of genetically engineered animals and disease models, automated monitoring methods are now indispensable [13], [18], and [25]. The home cage habitat lessens stress-related confounds, and the continuous, objective data collecting makes it possible to identify minute  behavioural  changes that could be overlooked in traditional testing paradigms [13], [18]. Automated monitoring is used in neuropsychiatric research applications to evaluate disease-relevant behaviour in mouse models. Because automated home cage monitoring can record naturalistic  behavioural  patterns over long periods of time, it has the potential to improve the translational utility of psychiatric research by better portraying the chronic nature of psychiatric illnesses [11]. These systems have been used in studies to describe  behavioural  characteristics in models of autism spectrum disorders, schizophrenia, anxiety, and depression [11], [18]. Long-term automated monitoring makes it possible to trace the course of neurodegenerative diseases and evaluate therapeutic approaches [25]. In order to find windows for therapeutic intervention, deep phenotyping procedures that use continuous home cage monitoring over weeks to months might reveal early  behavioural  alterations that precede more visible motor or cognitive deficiencies [25]. Models of Parkinson's disease, Alzheimer's disease, and other neurodegenerative diseases have been used with the systems [25]. While retaining social housing, automated monitoring makes it easier to analyze individual mouse activity in group-housed animals of various inbred strains, allowing for genetic and strain-specific  behavioural  characterisation [13]. When phenotyping genetically engineered animals, where  behavioural  effects may be subtle or context-dependent, this skill is especially useful [13], [18]. Quantitative trait locus (QTL) mapping and other genetic investigations of  behavioural  variance are supported by the capacity to gather substantial  behavioural  datasets [13]. Automated surveillance during experimental colitis, when home cage activity patterns offer sensitive indications of disease severity and recovery, is one example of a disease-specific use [22]. These applications show how useful automated monitoring is in a variety of illness domains outside of neurology [22].

3.4 Nutritional and Metabolic Research

 Investigating energy balance, dietary control, and metabolic disease is made possible by the combination of automated  behavioural  monitoring and metabolic data [3], [7], [8]. These technologies allow for the simultaneous evaluation of changes in body composition, energy expenditure (activity and metabolic rate), and energy intake (feeding behaviour) [3], [7].

The use of integrated monitoring systems in metabolic research is demonstrated by studies on leptin signalling. Acute interruption of leptin signalling causes elevated insulin levels and insulin resistance, with thorough temporal characterization of metabolic and  behavioural  alterations, according to automated surveillance of food intake, activity levels, and indirect calorimetry measures [3]. The intricate connections between hormonal signals, food behaviour, energy expenditure, and metabolic consequences can be dissected thanks to such integrated datasets [3]. Automated monitoring has been used in research on exercise-induced metabolic regulators to evaluate physical ability and metabolic fitness throughout life [29]. PhenoMaster systems were used in studies looking at MOTS-c, a mitochondrial-encoded peptide, to show that this factor improves physical performance in young, middle-aged, and old mice. The effects on locomotor activity, metabolic parameters, and circadian patterns were thoroughly characterized [29]. Nutritional adjustments and dietary interventions are investigated with the help of automated monitoring systems. Studies of time-restricted feeding, calorie restriction, and macronutrient-specific effects are made possible by the capacity to carefully regulate and monitor food access while continually monitoring  behavioural  and metabolic responses [24]. A thorough evaluation of how dietary treatments impact energy balance and body composition is provided by integration with body composition studies (e.g., NMR-based fat and lean mass measurement) [29].

Investigation of metabolic- behavioural  connections in illness contexts is also made possible by the systems. In models of obesity, diabetes, and metabolic syndrome, for instance, automated monitoring has been utilized to describe metabolic and  behavioural  phenotypes, demonstrating how metabolic dysfunction impacts activity patterns, circadian rhythms, and  behavioural  responses to metabolic challenges [7], [8].

VALIDATION AND  PERFORMANCE CHARACTERISTICS :

Rigorous validation of automated monitoring systems is critical to ensure data accuracy and interpretability. Validation studies typically benchmark automated measurements against gold-standard manual observations or established reference methods to evaluate performance and reliability  [1], [12], [18]

The Rodent Big Brother project performed extensive validation of its RFID-based tracking system against manual video analysis [1]. Validation showed that ambulatory activity measured by the baseplate correlated well with manual tracking (r = 0.85) and even more strongly with whole-cage video pixel movement (r = 0.91) [1]. However, the system exhibited limitations: correlation decreased at higher activity levels due to tracking truncation, and rapid movements occasionally resulted in missed detections, indicating reduced accuracy under conditions of intense or fast locomotion [1].

Video-based monitoring systems require validation of both motion detection algorithms and behavioural classification models. The SCORHE system was validated by comparing automated activity measurements with manual scoring, reliably detecting activity changes induced by pharmacological manipulations [12]. For machine learning–based  behavioural  classification, validation involves comparing system outputs to expert human annotations, with state-of-the-art models achieving over 90% accuracy across many  behavioural  categories [5], [18].

Validation of physiological parameter extraction from video requires comparison with invasive reference methods. Studies validating video-based heart rate and respiratory rate measurement have compared automated estimates against ECG and respiratory plethysmography, though challenges remain in achieving consistent accuracy across different conditions [30].

Inter-system reliability is a critical aspect of validation. Comparisons between automated monitoring platforms generally show good agreement for basic activity measures; however, variations in sensor placement, sensitivity, and algorithm design can produce notable quantitative differences [18]. This underscores the need for consistent methodology within studies and cautions against directly comparing absolute values across different systems [18].

Systematic reviews of home-cage monitoring development and applications highlight the critical need for standardized validation protocols and comprehensive reporting standards to enable reliable comparison across studies and platforms [9], [10]. Recommended practices include detailed documentation of validation procedures, quantitative accuracy metrics, and explicit reporting of system limitations for all automated measurements [9], [10].

 LIMITATIONS AND TECHNICAL CHALLENGES

5.1 Technical and Hardware Limitations

Despite advances, automated monitoring systems still face technical limitations. RFID tracking can miss detections during rapid movement, leading to gaps and underestimation of activity [1]. Its spatial resolution is also limited by antenna density, yielding discrete rather than continuous position data [1].

Infrared beam-break systems are robust and widely used, but they provide coarse spatial information and cannot capture detailed  behavioural  repertoires [2], [14]. They are sensitive to beam alignment and cage configuration, which can affect measurement consistency [1]. Additionally, these systems can only categorize basic movements, such as ambulatory versus vertical activity [2].

Video-based systems face challenges related to lighting conditions, occlusion, and image quality [12], [30]. Nesting material, bedding, and cage enrichment items can obscure animals from camera view, impeding accurate tracking and  behavioural  classification [30]. In group-housed settings, overlapping or closely positioned animals create ambiguity for tracking algorithms, making individual identification and behavioural attribution difficult without additional identification methods [30].

Physiological parameter extraction from video remains technically challenging, with accuracy varying based on animal size, fur colour, movement, and environmental conditions [30]. Current video-based vital sign monitoring cannot fully replace invasive methods for applications requiring high precision, such as ECG or EEG recording [30].

Hardware reliability and maintenance requirements present practical challenges. Sensor failures, camera malfunctions, and software errors can result in data loss or corruption [12]. The complexity of integrated multi-modal systems increases the potential failure points and maintenance burden [30].

DISCUSSION

Automated  behavioural  monitoring systems have fundamentally transformed metabolic cage studies, enabling continuous, objective, and comprehensive assessment of animal behavioural and physiology. This review has synthesized findings from 30 highly relevant studies, revealing a diverse technological landscape, broad research applications, and persistent challenges that shape current practice and future development.

The technological diversity evident in current systems reflects different priorities and trade-offs. Infrared beam-break systems offer robustness, ease of integration, and relatively low cost, making them the most widely adopted approach for basic activity monitoring [2], [3], [7]. RFID-based systems enable individual tracking in group-housed animals, addressing a critical need for social housing while maintaining individual-level data [1], [13], [21]. Video-based approaches provide the richest  behavioural  information and greatest flexibility but require substantial computational resources and sophisticated analysis methods [5], [12], [18]. Multi-modal integrated systems represent the current frontier, combining complementary technologies to provide comprehensive  behavioural  and physiological datasets [14], [29], [30].

The reviewed applications show the broad utility of automated monitoring across research domains. In circadian biology, continuous monitoring has uncovered molecular mechanisms of rhythm regulation and detailed circadian disruption in disease models [4], [17]. In drug discovery, automated monitoring supports high-throughput screening, safety pharmacology, and identification of translational biomarkers [1], [14]. Disease modelling and phenotyping use these systems to detect subtle  behavioural  changes and monitor disease progression [11], [13], [25]. Metabolic research benefits from integrated assessment of feeding, activity, and energy expenditure [3], [7], [29].

Despite significant advances, important limitations persist. Technical challenges together with missed detections, occlusion, and limited spatial resolution constrain measurement accuracy [1], [30]. Methodological challenges related to data volume, algorithm validation, and statistical analysis require on going attention [9], [10], [30]. Practical constraints contain  cost, species limitations, and implementation complexity, which influence accessibility and adoption [9], [12].

The field is moving toward increasingly sophisticated approaches that leverage advances in computer vision, machine learning, and sensor technology [5], [18], [25], [30]. However, several key challenges require continued focus. Standardization of validation protocols and reporting standards is essential to improve reproducibility and enable meaningful comparison across studies [9], [10]. Development of more accessible, lower-cost systems would democratize access to these powerful technologies [12], [15]. Enhanced integration with other phenotyping modalities and development of translational biomarkers may improve the predictive validity of preclinical research [14].

The ethical dimension of automated monitoring deserves emphasis. These systems align with refinement principles by enabling monitoring in familiar home environments, reducing handling stress, and potentially detecting welfare issues earlier than conventional observation [9], [13]. However, the ability to collect vast amounts of  behavioural  data raises questions about appropriate use, data interpretation, and the balance between comprehensive monitoring and animal privacy [9].

CONCLUSION

Automated  behavioural  monitoring systems have become indispensable tools in metabolic cage studies, offering capabilities that extend far beyond traditional manual observation. This review has examined the technological approaches, research applications, and limitations of these systems based on an analysis of 30 highly relevant studies. The field encompasses diverse technologies, including infrared beam-break sensors, RFID tracking, video analysis, and multi-modal integrated platforms, each with distinct advantages and constraints.

Applications span circadian biology, drug discovery, disease modelling, and metabolic research, with automated monitoring enabling continuous, objective assessment of  behavioural  and physiological parameters in naturalistic home cage environments. Validation studies demonstrate generally good performance for basic activity measures, though accuracy varies across systems and measurement contexts. Persistent limitations include technical challenges related to tracking accuracy and occlusion, methodological challenges in data analysis and validation, and practical constraints including cost and implementation complexity.

The field is rapidly evolving toward more sophisticated computer vision approaches, machine learning integration, and non-invasive physiological monitoring. Future progress will depend on continued technological innovation, standardization of methods and reporting, improved accessibility through open-source platforms, cost reduction, and enhanced integration with other phenotyping modalities. As these systems continue to advance, they promise to further enhance the quality, efficiency, and translational relevance of preclinical biomedical research while supporting improved animal welfare through refined experimental approaches.

REFERENCES

1. W. S. Redfern et al., "Automated recording of home cage activity and temperature of individual rats housed in social groups: The Rodent Big Brother project," PLOS ONE, vol. 12, no. 9, 2017, doi: 10.1371/journal.pone.0181068.
2. J. R. B. Lighton, "The Varieties Of Activity Detectors," in Measuring Metabolic Rates: A Manual for Scientists, 2nd ed., Oxford University Press, 2018, doi: 10.1093/oso/9780198830399.003.0018.
3. J. Levi et al., "Acute disruption of leptin signaling in vivo leads to increased insulin levels and insulin resistance," Endocrinology, vol. 152, no. 9, pp. 3385-3395, 2011, doi: 10.1210/en.2011-0185.
4. N. W. DeKorver et al., "Toll-Like Receptor 2 Is a Regulator of Circadian Active and Inactive State Consolidation in C57BL/6 Mice," Frontiers in Aging Neuroscience, vol. 9, 2017, doi: 10.3389/fnagi.2017.00219.
5. H. Jhuang et al., "Trainable, vision-based automated home cage  behavioural  phenotyping," in Proceedings of the 1st ACM International Conference on Multimedia Retrieval, 2010, doi: 10.1145/1931344.1931377.
6. F. Oberhauser et al., "Towards Continuous Home Cage Monitoring: An Evaluation of Tracking and Identification Strategies for Laboratory Mice," arXiv preprint, 2025, doi: 10.48550/arxiv.2507.07929.
7. P. Jenny et al., "Additional file 5 of A systematic review of the development and application of home cage monitoring in laboratory mice and rats," 2024, doi: 10.6084/m9.figshare.26638905.v1.
8. K. Hohlbaum, "Development and Application of Home Cage Monitoring in Laboratory Mice and Rats: a Systematic Review," bioRxiv, 2023, doi: 10.1101/2023.03.07.531465.
9. P. Jenny et al., "Additional file 3 of A systematic review of the development and application of home cage monitoring in laboratory mice and rats," 2024, doi: 10.6084/m9.figshare.26638899.v1.
10. P. Jenny et al., "Additional file 3 of A systematic review of the development and application of home cage monitoring in laboratory mice and rats," 2024, doi: 10.6084/m9.figshare.26638899.
11. A. Mingrone et al., "The Promise of Automated Home-Cage Monitoring in Improving Translational Utility of Psychiatric Research in Rodents," Frontiers in Neuroscience, vol. 14, 2020, doi: 10.3389/fnins.2020.618593.
12. G. H. Salem et al., "SCORHE: A novel and practical approach to video monitoring of laboratory mice housed in vivarium cage racks," BehaviouralResearch Methods, vol. 47, no. 1, pp. 235-250, 2015, doi: 10.3758/s13428-014-0451-5.
13. . S. Bains et al., "Analysis of Individual Mouse Activity in Group Housed Animals of Different Inbred Strains using a Novel Automated Home Cage Analysis System," Frontiers in  behavioural  Neuroscience, vol. 10, 2016, doi: 10.3389/fnbeh.2016.00106.
14. S. W. Baran et al., "Emerging Role of Translational Digital Biomarkers Within Home Cage Monitoring Technologies in Preclinical Drug Discovery and Development," Frontiers in  behavioural  Neuroscience, vol. 15, 2022, doi: 10.3389/fnbeh.2021.758274.
15. T. Libey et al., "Open-Source, Low Cost, Free-BehaviouralMonitoring, and Reward System for Neuroscience Research in Non-human Primates," Frontiers in Neuroscience, vol. 11, 2017, doi: 10.3389/fnins.2017.00265.
16. . Kahnau et al., "Full title: Development and Application of Home Cage Monitoring in Laboratory Mice and Rats: a Systematic Review Short title: Systematic Review of Home Cage Monitoring in Rodents," unpublished manuscript.
17. C. J. Perry et al., "A circadian  behavioural  analysis suite for real-time classification of daily rhythms in complex behaviors," bioRxiv, 2024, doi: 10.1101/2024.02.23.581778.
18. V. Voikar et al., "Three Pillars of Automated Home-Cage Phenotyping of Mice: Novel Findings, Refinement, and Reproducibility Based on Literature and Experience," Frontiers in  behavioural  Neuroscience, vol. 14, 2020, doi: 10.3389/fnbeh.2020.575434.
19. P. Jenny et al., "Additional file 5 of A systematic review of the development and application of home cage monitoring in laboratory mice and rats," 2024, doi: 10.6084/m9.figshare.26638905.
20. K. R. Jacobsen et al., "Postsurgical food and water consumption, fecal corticosterone metabolites, and behaviouralassessment as noninvasive measures of pain in vasectomized BALB/c mice," Journal of The American Association for Laboratory Animal Science, vol. 51, no. 1, pp. 69-75, 2012.
21. J. M. Dzik et al., "PyMICE: A Python library for analysis of IntelliCage data," BehaviouralResearch Methods, vol. 50, no. 4, pp. 1807-1815, 2018, doi: 10.3758/s13428-017-0907-5.
22. E. Zentrich et al., "Automated Home-Cage Monitoring During Acute Experimental Colitis in Mice," Frontiers in Neuroscience, vol. 15, 2021, doi: 10.3389/fnins.2021.760606.
23. H. L. Evans, "Behaviors in the Home Cage Reveal Toxicity: Recent Findings and Proposals for the Future," International Journal of Toxicology, vol. 8, no. 4, pp. 475-491, 1989, doi: 10.3109/10915818909009092.
24. S. Saeedi et al., "CageView: A Smart Food Control and Monitoring System for Phenotypical Research In Vivo," 2023, doi: 10.32920/23737437.
25.  B. Jurek et al., "Long-term deep phenotyping of  behavioural  traits in mice using homecage monitoring," Neuroscience & Bio behavioural  Reviews, vol. 170, 2025, doi: 10.1016/j.neubiorev.2025.106453.
26. F. Nehchiri, "Investigating the Role of Exercise in Reducing Carbamylated Proteins: Implications for Health and Disease Management," unpublished manuscript.
27. P. J. J. Noldus et al., "EthoVision: a versatile video tracking system for automation of  behavioural  experiments," BehaviouralResearch Methods Instruments & Computers, vol. 33, no. 3, pp. 398-414, 2001, doi: 10.3758/bf03195394.
28. A. Milesi-Halle, "Sexual dimorphism in the pharmacokinetics and pharmacodynamics of (+)-methamphetamine and (+)-amphetamine in models of human drug abuse," unpublished manuscript.
29. . N. Reynolds et al., "MOTS-c is an exercise-induced mitochondrial-encoded regulator of age-dependent physical decline and muscle homeostasis," Nature Communications, vol. 12, 2021, doi: 10.1038/s41467-020-20790-0.
30. F. Mösch et al., "Towards substitution of invasive telemetry: An integrated home cage concept for unobstrusive monitoring of objective physiological parameters in rodents," bioRxiv, 2023, doi: 10.1101/2023.05.12.540546.