TL;DR
Physical activity monitors (PAMs) offer objective insights into companion animal movement, crucial for understanding chronic pain like osteoarthritis. This review highlights the technical and biological factors influencing PAM data interpretation, emphasizing the need for a deeper understanding of PAM technology and its application in veterinary research to advance objective pain assessment.
What Problem Does This Paper Address?
Chronic pain in companion animals, particularly osteoarthritis, significantly impacts their quality of life by affecting movement and mobility. While physical activity monitors (PAMs) hold great promise for objectively measuring these changes, their application in veterinary research is hampered by a lack of understanding regarding their technology, data interpretation, and the biological factors that influence readings. This leads to challenges in accurately assessing pain severity and treatment efficacy. The paper aims to bridge this knowledge gap by providing a comprehensive review of PAMs, focusing on accelerometers, and discussing the technical and biological considerations for their use in companion animal chronic pain research.
Methods
This paper is a systematic and commentary review. The authors searched Google Scholar and PubMed using keywords related to physical activity monitors, activity monitors, chronic pain, veterinary medicine, osteoarthritis, and accelerometers. Relevant abstracts were read, and selected papers were reviewed in full. References from these papers were also checked for additional relevant reports. The findings were synthesized with the authors' expertise. The review delves into the technical aspects of PAMs, including:
- Accelerometer Technology: Explaining how accelerometers function, including piezoelectric and capacitance types, and the importance of multi-axial measurements (1, 2, or 3 axes).
- Data Acquisition and Processing: Discussing sampling frequency, data filtration, and the proprietary nature of "activity counts" and epoch settings.
- Device Calibration, Reliability, and Validity: Examining the importance of ensuring devices accurately measure what they intend to, the variability between devices, and the need for validation in companion animals.
- Data Summation and Analysis: Reviewing various methods for analyzing PAM data, from simple averages to more sophisticated techniques like functional linear modeling (FLM) and functional data analysis (FDA).
- Biological and Use Considerations: Addressing how factors like sensor placement, body conformation, age, and even leash attachment can influence data.
- Future Directions: Exploring integrated sensor units (IMUs), machine learning for behavior detection, and the concept of movement smoothness.
Datasets
Not discussed in the paper. The paper is a review and does not present new experimental data or datasets.
Key Insights and Ideas
- "Activity Counts" are Device-Specific: There's no universal unit for activity. Different PAMs use proprietary algorithms to generate "activity counts," making direct comparison between devices without correction equations impossible. This highlights the need for researchers to understand the specific device they are using.
- Biological Factors are Crucial: Age, body conformation, and even species significantly influence PAM data. For instance, older dogs tend to have lower activity counts, and body weight can inversely affect activity. Ignoring these factors can lead to misinterpretations.
- Placement Matters: Where a PAM is placed on an animal (e.g., collar vs. harness) can impact the data collected. While collars are convenient, leash attachment can interfere with readings.
- Beyond Simple Summaries: Traditional analysis methods (e.g., mean activity counts) can mask important nuances. More advanced techniques like functional linear modeling (FLM) are better at capturing complex patterns and identifying subtle effects of pain or treatment, especially when considering factors like time of day.
- Pain Doesn't Always Mean Less Movement: While intuitive, chronic pain doesn't always lead to a simple decrease in overall activity. It can manifest as changes in movement patterns or quality (e.g., less smooth motion), which are harder to detect with basic metrics.
- "Normalizing" Activity is Unrealistic: Analgesic treatments may not return an animal's activity to a pre-pain state. Instead, they might induce a novel set of behaviors. The goal should be to improve function and reduce pain, not necessarily to achieve a pre-pain "normal."
Key Results
As this is a review paper, it does not present quantitative experimental results. However, it synthesizes findings from numerous studies, highlighting:
- PAMs are Validated (to an extent): Many PAMs have been shown to be valid surrogate measures of spontaneous activity, though validation studies are crucial for each device and application.
- Inter-Device Variability is High: Some accelerometers can exhibit variability as high as 57% between units of the same make, underscoring the importance of reliability considerations.
- Activity Patterns are Informative: Studies show that analyzing activity patterns over 24-hour periods, rather than just aggregate summaries, can reveal more about the impact of pain and treatment. For example, changes in nighttime activity have been linked to pain and analgesic effects.
- FLM Shows Promise: Functional linear modeling has demonstrated its ability to reveal subtle effects of covariates (like age or pain) on activity patterns that might be missed by traditional statistical methods.
Limitations and Future Work
Limitations:
- Proprietary Technology: Many PAMs are "black boxes," with limited transparency regarding their data processing algorithms, making independent validation and interpretation challenging.
- Lack of Standardization: There's no standardized approach to data analysis or reporting of activity counts, hindering direct comparisons across studies.
- Limited Validation in Companion Animals: Many devices are designed for humans, and their suitability for companion animals requires specific validation.
- Understanding Biological Meaning: The biological significance of PAM data is still in its infancy, requiring more research to define clinically meaningful changes.
Future Work:
- Integrated Sensor Units (IMUs): Combining accelerometers with other sensors (gyroscopes, GPS) can provide more comprehensive data on movement quality and specific behaviors.
- Machine Learning (ML): Applying ML algorithms to PAM data can help detect specific movements and behaviors, offering objective insights into pain-related changes.
- Implantable Accelerometers: Future advancements may lead to implantable sensors for continuous, real-time monitoring.
- Networked Approach: Integrating PAM data with other measures (clinical signs, owner reports) will provide a more holistic understanding of pain's impact and enable personalized treatment.
- Defining Clinically Meaningful Changes: Further research is needed to establish thresholds for successful treatment outcomes based on PAM data.
Summary updated