Movement Health Blog | Sparta Science

Provider-Delivered, Data-Driven Movement Health | Sparta Science

Written by Greg Olsen Ph.D. | Nov 29, 2022 7:00:00 AM

By Greg Olsen, Ph.D. and Jessica Hinman Ph.D.

Overview

Increasingly available and affordable digital technologies (e.g., instrumentation, data collection, data storage, and machine learning capabilities) have enabled a generation of “smart” operational systems that learn from the massive data they produce.

Examples are everywhere:

  • Smart industrial monitoring systems collect sensor data and use models to predict failures
  • Media companies like Netflix collect viewing data and use it to predict what you’re likely to watch
  • eCommerce companies like Amazon use browsing and transaction data to predict what you’re likely to buy
  • Visa & Mastercard use transaction history data to detect potential fraud

Each of these cases shares common elements:

  • Aggressive instrumentation & frequent data collection: Measure more than you may need and accumulate large volumes of data.
  • Learning: Use machine learning to recognize patterns in the data and how those patterns relate to outcomes of interest.
  • Continuous improvement: Frequently update models as more data is collected to improve analysis accuracy.

Examples of this approach are starting to appear in the health & wellness world. The most visible cases are in the growing numbers of individuals who are instrumented with wearable technologies that collect data and feed machine learning analyses to support a host of health interests (e.g., arrhythmia detection, fitness assessment, blood sugar management). 

Provider-delivered healthcare, however, has yet to see the systematic application of big data and machine learning capabilities. Providers generally don’t collect biometric data suitable for large-scale active monitoring of patients, nor do they employ machine learning on that data. Frequently measured data (e.g., temperature, heart rate, BP, and weight) are captured only a few times a year for most people. Even when patients are undergoing a routine care regimen, like orthopedic rehab, the availability of rich, objective tracking data is limited. For the most part, instrumentation, data collection, and machine learning are things done by device vendors or in research laboratories. They are not integrated into clinical operations.

The digital health environment, however, is changing rapidly. The incredible growth in scalable, low-cost, health-relevant sensor technologies is creating a path to a more data-driven future.

New Sources of Data and Opportunity

Over the last 10+ years, health-related sensor technology has become ubiquitous. Health instrumentation is now present in nearly every small clinic, drug store, and home. Examples include:

  • Blood pressure monitoring devices
  • Continuous glucose monitoring devices
  • Wearable devices (e.g., watches, bands, rings) that measure:
    • Heart rate, HRV, ECG
    • Blood Oxygen
    • Temperature
    • Movement (Steps, Distance, Cadence)
    • Sleep patterns
  • Exercise equipment instrumentation (Bikes, Treadmills, Resistance Training):
    • Power
    • Work
    • Force
  • Video motion capture
    • Movement patterns, range of motion, biomechanical issues
  • Smart force plates & scales
    • Balance
    • Force development
    • Body composition


 

This path to more significant health instrumentation has been led by consumer applications that provide immediate insights into specific conditions (e.g., high blood pressure, blood sugar, arrhythmia, or fitness level). Downstream use of this data for machine learning is typically limited to device makers or large consumer personal health platform vendors like Apple, Google, Garmin, or Amazon.

These types of digital technology and data are making their way into provider-delivered healthcare. Within the recent wave of digital health startups, many seek to use instrumentation (e.g., wearables or motion capture via consumer devices) to improve the provision of care to patients. Though the pace of change has been slow, more traditional health providers are also beginning to embrace the expanded use of patient instrumentation and data analysis technologies.

The bottom line is that instrumentation, data collection, and machine learning represent tremendous opportunities for providers to improve their care delivery and their businesses. This use case for digital technology will ultimately find its way into every aspect of healthcare practice. One of the areas to see the early focus is a category of needs we refer to as “Movement Health.” 

Movement Health

The term Movement Health captures a broad set of health concerns relevant to many different perspectives, including medical practice, physical therapy, sport & exercise, and occupational health & wellness. Example concerns that fit into this category include:

  • Musculoskeletal (MSK) injury prevention and recovery
  • Balance and mobility improvement
  • Athletic performance optimization
  • MSK pain management
  • Traumatic brain injury diagnosis and recovery management
  • Neurodegenerative disorder diagnosis & treatment

The limited availability of objective assessment data is a critical challenge to movement health practice. The situation fits the familiar “you can’t manage what you can’t measure” axiom. Indeed, diagnostic tools have been applied to movement health issues (e.g., x-rays, MRI, ECG, EEG, range-of-motion tests, and “functional tests”). However, the ability of these tools to address needs has been limited in two primary ways:

  • Granularity of measurement: What aspects of movement health are being assessed, and how precisely can differences between individuals be measured? Many current assessments focus only on a narrow element of movement health and only supply coarse-grained bad/not-obviously-bad distinctions.
  • Scale and reach of measurement: How many people can be assessed feasibly at high frequency? Apparatus, expertise, cost, & complexity limit the scaling of legacy assessment approaches.

What if providers had high-quality movement health metrics and associated analysis throughout the course of each patient’s care? The potential benefits are manifold:

  • Better patient screening & care assignment 
  • Better recovery progress tracking & care adjustment
  • Better care effectiveness & quality control

Movement Health Intelligence Platforms

To take advantage of this opportunity, health providers need technology platforms and applications beyond the current scheduling, billing, and EHR-focused systems they have deployed. These needs are analogous to those that drove the deployment of data-centric operational capabilities into other domains such as manufacturing, eCommerce, and finance – one key difference being that those were far less limited by a lack of instrumentation.

Consumer-driven investments in fitness-related technologies have resulted in radically improved instrumentation for movement health-related issues. Several instrumentation technologies are particularly well-suited to movement health assessment needs, notably:

  • Force plates: Tests performed on force plates can provide rich and reliable data for movement health assessment. Various test protocols are available (e.g., balance and jump variations) to serve the needs of different populations and conditions of interest. Force plates can also be pragmatically deployed in a manner similar to weight scales and used with minimal training.
  • Wearables: Wristband or ring-based wearables are exceptionally convenient for passively collecting activity data (e.g., exercise output, general movement, sleep) over extended periods. From wearable devices, providers can get information about what transpires between visits or checkpoints in therapy.
  • Video motion capture: Video systems can capture complex movement patterns and reduce them to usable data. High-precision systems have significant setup/apparatus needs, but some basic capabilities are available via cameras on consumer phones. 

This instrumentation capability, combined with available low-cost, scalable data collection, storage, and analysis technologies, provides the foundation for health intelligence platforms to be a standard tool for nearly every provider – not limited to research laboratories or the Apples and Googles of the world.

 

Providers should look for the following capabilities in a movement health intelligence platform:

  • Reliable tests that positively engage patients and that can be administered frequently: Testing is only useful if results are accurate and patients and clinicians are willing to do it- time, convenience, and integration into existing workflows are vital considerations. Tests need to extract sufficient reliable information to differentiate patients and provide immediate results feedback to assist with patient engagement.
  • Application functionality for clinicians: Clinicians need an application that enables them to track large numbers of patients through their care efficiently. Application functionality should include comparative analysis and trend analysis. The application should support the administrative needs of providers with complex organization and access control requirements.
  • Application functionality for patients: Giving patients the ability to see their progress over time makes them more engaged. Application capabilities can also include guidance to patients and feedback from those patients.
  • Scalable data capture & management: Data is an accumulating asset for the provider. Instrumentation results need to feed into a centralized repository for downstream analysis and potential recalculation in the event of test calculation updates.
  • Rich data analysis & machine learning capabilities: The platform needs to facilitate new insights for the provider based on the data they collect. Insights can come from statistical properties, pattern recognition, clustering, feature importance analysis, classification, or other machine learning techniques.

Movement health providers that arm themselves with these platforms will be able to deliver superior care, operate more efficiently, and build a data asset that grows with time.