Peter Mork

Context: We describe a participatory framework that enhanced and implemented innovative changes to an existing distributed health data network (DHDN) infrastructure to support linkage across sectors and systems. Our processes and lessons learned provide a potential framework for other multidisciplinary infrastructure development projects that engage in a participatory decision-making process. Program: The Childhood Obesity Data Initiative (CODI) provides a potential framework for local and national stakeholders with public health, clinical, health services research, community intervention, and information technology expertise to collaboratively develop a DHDN infrastructure that enhances data capacity for patient-centered outcomes research and public health surveillance. CODI utilizes a participatory approach to guide decision making among clinical and community partners. Implementation: CODI's multidisciplinary group of public health and clinical scientists and information technology experts collectively defined key components of CODI's infrastructure and selected and enhanced existing tools and data models. We conducted a pilot implementation with 3 health care systems and 2 community partners in the greater Denver Metro Area during 2018-2020. Evaluation: We developed an evaluation plan based primarily on the Good Evaluation Practice in Health Informatics guideline. An independent third party implemented the evaluation plan for the CODI development phase by conducting interviews to identify lessons learned from the participatory decision-making processes. Discussion: We demonstrate the feasibility of rapid innovation based upon an iterative and collaborative process and existing infrastructure. Collaborative engagement of stakeholders early and iteratively was critical to ensure a common understanding of the research and project objectives, current state of technological capacity, intended use, and the desired future state of CODI architecture. Integration of community partners' data with clinical data may require the use of a trusted third party's infrastructure. Lessons learned from our process may help others develop or improve similar DHDNs.

To date, the principal use case for schema matching research has been as a precursor for code generation, i.e., constructing mappings between schema elements with the end goal of data transfer. In this paper, we argue that schema matching plays valuable roles independent of mapping construction, especially as schemata grow to industrial scales. Specifically, in large enterprises human decision makers and planners are often the immediate consumer of information derived from schema matchers, instead of schema mapping tools. We list a set of real application areas illustrating this role for schema matching, and then present our experiences tackling a customer problem in one of these areas. We describe the matcher used, where the tool was effective, where it fell short, and our lessons learned about how well current schema matching technology is suited for use in large enterprises. Finally, we suggest a new agenda for schema matching research based on these experiences.