Method and design
This will be a mixed methods project combining systematic scoping reviews to inform the design, modelling and testing of a complex intervention (a prototype tool for improving database searches, using machine learning techniques with a dissemination strategy) to promote evidence informed practice.
Phase I: Investigation
Explore, through reviewing the literature, clinician beliefs and attitudes towards evidence to understand barriers and enhancers to receipt and application of information (Scoping review).
Explore new and novel ways of encouraging the adoption of evidence informed practice in particular reference to behaviour change theories (scoping review).
Phase II: Development
- Develop IT solutions (including machine learning approaches) to target and source information relevant to osteopathic practice and care. This will probably take the form of a pilot system for conducting semi-automated literature searches and topic analysis. This will feature elements of machine learning, including query expansion techniques using word2vec (a neural network that maps words from a given corpus into a numerical 'vector space') and unsupervised topic allocation using latent Dirichlet allocation (a Bayesian approach to modelling the probability that any topic represents a given document), or similar approaches.
Phase III: Model and finalise complex intervention
Develop conceptual models based on phase I and II for potential strategies for dissemination and utilisation of research in practice.
Develop and devise material and strategies suitable and acceptable for dissemination and uptake by clinicians (feasibility study).
Phase IV: Implement and evaluate the intervention
Evaluate, on a sample of osteopaths: receipt and application to practice and impact on patient care (prospective cohort study).
Assess the impact of intervention on clinical practice, through reassessment of clinician attitudes towards evidence based practice and analysis of patient experience and care.