Using data mining methods with standardized terminology data sets for home visiting intervention effectiveness research

Home visiting is a widely accepted intervention for addressing psychosocial and health needs of disadvantaged families. However, a recent Cochrane Review reported there is no evidence that home visiting improves maternal psychosocial health or parenting for disadvantaged women, or outcomes for their children. The authors called for improved outcome studies to discover intervention models that effectively address these needs. Due to adoption of computerized clinical documentation systems and availability of data mining methodologies, there is new potential to expediently advance home visiting intervention effectiveness research. Standardized terminology data sets offer rich content regarding client characteristics, interventions, and outcomes. The Omaha System is a complex, multi-axial, hierarchical, relational classification system developed through federally funded research; and is commonly used for computerized documentation of home visiting practice. It has three components: the Problem Classification Scheme (416 problem descriptors), the Intervention Scheme (12,600 intervention combinations), and the Problem Rating Scale for Outcomes (126 Likert-type scales). Omaha System terms broadly describe health problems in environmental, psychosocial, physiological, and behavioral domains; and can define groupings of clients, determine problem-specific or client-specific outcomes, and describe intervention approaches. A data mining methodology for exploring complex data necessary for intervention effectiveness research in large data sets is Knowledge Discovery in Databases (KDD). KDD combines domain expertise with statistics, machine learning (artificial intelligence), and computer science to generate predictive models from data. KDD uses multiple steps for discovering patterns to inductively build models relating processes and outcomes: selection, abstraction, preprocessing, transformation, data mining (analysis), validation, and knowledge. Home visiting data exemplars will be given. Elements of Omaha System standardized documentation provide the necessary granularity and specificity to address intervention effectiveness questions using KDD methods. Hypotheses generated through KDD can be tested with conventional statistical methods to reveal new patterns of intervention effectiveness for disadvantaged families receiving home visiting services.