A Semantic Framework for Logical Cross-Validation

A Semantic Framework for Logical Cross-Validation, Evaluation and Impact Analyses of Population Health Interventions Arash SHABAN-NEJADa1, Anya OKHMATOVSKAIAb, Eun Kyong SHINa, Robert L. DAVISa, Brandi E. FRANKLINc, David L. BUCKERIDGEb a University of Tennessee Health Science Center – Oak-Ridge National Lab (UTHSC-ORNL) Center for Biomedical Informatics, Department of Pediatrics, Memphis, Tennessee, USA b McGill Clinical & Health Informatics, Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Quebec, CANADA c Health Outcomes and Policy Research, Department of Clinical Pharmacy, The University of Tennessee Health Science Center, Memphis, Tennessee, USA Abstract. Most chronic diseases are a result of a complex web of causative and correlated factors. As a result, effective public health or clinical interventions that intend to generate a sustainable change in these diseases most often use a combination of strategies or programs. To optimize comparative effectiveness evaluations and select the most efficient intervention(s), stakeholders (i.e. public health institutions, policymakers and advocacy groups, practitioners, insurers, clinicians, and researchers) need access to reliable assessment methods. Building on the theory of Evidence-Based Public Health (EBPH) we introduce a knowledge-based framework for evaluating the consistency and effectiveness of public health programs, interventions, and policies. We use a semantic inference model that assists decision-makers in finding inconsistencies, identifying selection and information biases, and with identifying confounding and hidden dependencies in different public health programs and interventions. The use of formal ontologies for automatic evaluation and assessment of public health programs improves program transparency to stakeholders and decision makers, which in turn increases buy-in and acceptance of methods, connects multiple evaluation activities, and strengthens cost analysis. Keywords. Program Evaluation, Outcome assessment, Ontology, Inference, Public Health Intervention

1

Corresponding Author: Arash Shaban-Nejad, University of Tennessee Health Science Center – Oak-Ridge National Lab (UTHSC-ORNL) Center for Biomedical Informatics, Department of Pediatrics, 497R Le Bonheur Research Center, 50 N Dunlap St, Memphis TN 38103, USA, Tel: (901) 287-5366; Email: [email protected]