The Role of Learning in Chronic Musculoskeletal Pain

Aims and objectives:
Despite its prevalence, we still don't know why some people get chronic pain, and others do not. One influential idea is that the processes in the brain that normally allow us to adapt to an injury and recover from it, are used excessively, meaning that pain is exaggerated and prolonged beyond what is necessary. This 'maladaptive brain learning' hypothesis, in its various forms, is a popular model of chronic musculoskeletal pain. However, evidence is currently limited by the lack of sufficient tools required to measure and quantify learning. The project addresses this by implementing a novel set of experimental tools based on a basic science understanding of how learning works in the brain. We will use these tools to study outcomes in two complementary longitudinal studies: i) recent onset lower back pain and ii) fibromyalgia patients embarking on a multidisciplinary treatment (MDT) program. These tools will also be disseminated openly across the APDP and further afield.

Over the last year, we have established a unique partnership with people with lived experience of pain, physiotherapists, clinicians and engineers at Oxford and Cambridge, to create a core toolkit for pain-related learning evaluation. This is a set of tablet games that people play at home. People can also use their tablet camera to record specific physio exercises which we use to reconstruct and quantify the impact of pain on movement. By using a patient-led design at the outset, the tools are user-friendly, engaging, and maximise accessibility. This provides the means to test the maladaptive learning hypothesis in a broad range of clinical cohorts.

Data to be collected:
We will pursue 3 integrated workstreams:
i) Development and dissemination of an open data analysis platform to accompany the experimental toolkit.
ii) Data collection in a cohort of 140 patients with recent onset (acute) lower back pain, presenting via NHS GP services, to predict outcomes at 12 months
iii) Data collection in a cohort of 80 patients with fibromyalgia undergoing an NHS MDT program.
Both clinical cohorts aim to look for the ability of learning metrics to predict clinical outcomes, and validate the findings with neuroimaging. All data will be made open via the ADPD datahub.

Potential benefits:
The main scientific outcome will be the identification and characterisation of how learning correlates with chronic pain outcomes. This is important because such mechanisms directly imply treatment targets, which can be realised using the non-pharmaceutical interventions (cognitive and physical rehabilitation). This therefore provides a springboard for treatment innovation across the APDP and partners.
These benefits are facilitated by the patient-led design of the tools. Practically, they are easy-to-implement, require minimal expertise, and come with open data analysis pipelines and collaborative support if required. This builds a UK-based network that will capitalise on the innovation and expertise across the whole APDP. Furthermore, the data generated by this infrastructure opens up new opportunities for bioinformatics and related applications (e.g. in clinical stratification and outcome prediction).

Legacy and sustainability.
The very nature of data collected, being based on a common set of tools, will seed a database that will grow over time. This is self-sustaining because the statistical 'power' of the database increases as more data is added, permitting comparative analyses of individual datasets to a wide-range of other conditions. The tools also provide a technological backbone that can be developed and refined over time, as new insights, tools and techniques become available. In effect once the systems are in place, they grow organically over time. This is likely to be realised in new therapies, potentially as early as 5 years, given its potential to be integrated with existing technology-based treatment methodologies.

We address whether maladaptive learning systems contribute to the maintenance of chronic pain. This has been difficult to answer because learning comprises a set of complex, interacting processes, and hence difficult to evaluate and quantify. Based on computational models of learning, we have designed a suite of tasks and analysis tools that probe domain-general value-based and sensorimotor learning. These are implemented online as a set of tablet-based computer games that can be applied easily and widely in observational or longitudinal clinical studies in domestic settings.

We will study whether learning metrics can predict clinical outcomes in: i) a longitudinal study of patients with recent onset low back pain, and ii) fibromyalgia patients undergoing an NHS MDT program. We will measure neural changes (using resting-state fMRI) and physical outcomes using a novel video-based tool to quantify movement during physiotherapy exercises.

The tools, analysis pipeline and data will all be made free and openly available to the research community. Within the APDP open Datahub, data from our and other partners can be pooled across various conditions, allowing further comprehensive characterisation of how learning is linked to chronic pain.

We adopt a systems engineering approach to the development of our tools, in which patients have and will continue to have a fundamental role in the design process itself, necessary to ensure the usability and relevance of the framework to the lived experience of pain. This builds the joint vision of researchers and patients to develop an open platform for integrated multidisciplinary evaluation and treatment.

The proposed project will have a lasting impact: building a self-sustaining data generating infrastructure focused on learning mechanisms; providing a technological platform to support innovation in non-pharmacological treatment; and building a UK pain neuroscience and engineering community focused on chronic pain