HBCP Findings

Key findings from the HBCP are:

  1. It is possible to create an ontology covering all aspects of behaviour change interventions, their contexts and their evaluations.
  2. It is possible to annotate randomised controlled trials with high accuracy using this ontology.
  3. It is possible to make predictions about outcomes in novel smoking cessation intervention scenarios from manually extracted information from study reports.
  4. It is possible to automate the process of finding relevant studies in a domain of interest with high sensitivity.
  5. Intervention evaluation study reports miss out key information about interventions and contexts or present them in ways that severely limit the ability to 1) compare and synthesise evidence, and 2) predict outcomes.
  6. It is currently not feasible to fully automate the extraction of detailed information about interventions, their contexts and evaluations from pdfs of study reports because of variation and ambiguity in the way information is presented. However, at least some of the process can be automated, for example information in CONSORT diagrams that describe the flow of participants through a study.