ODHS Research and Analytics collaborates with ODHS programs, leadership, other state agencies, community stakeholders and policy-makers to develop and implement research and disseminate findings. Research and analytics help ODHS learn more about the efficacy and limitations of the programs and services it provides and the people and communities that it serves, including a more fundamental understanding of their needs and experiences.
In 2015 ODHS’s internal research and analytics capacity was expanded through the addition of the Oregon Enterprise Data Analytics (OEDA) contract. The Oregon State Legislature created OEDA to produce evidence-based, actionable information through inter-agency research to improve the lives of Oregonians. People experience intersecting issues across health, employment, poverty, education and other aspects of their lives. OEDA projects explore individual and community issues that involve programs or services that fall under multiple state agencies (for example ODHS, Oregon Health Authority and Oregon Department of Education). Essential direction in prioritizing our projects is provided by the Oregon Enterprise Data Analytics Advisory Committee, whose members include representatives from nine state agencies.
- Collaborate in the development of research agendas for each ODHS program area;
- Develop research-based tools to help inform decisions for both policy and field operations;
- Support the development of data-informed decision-making in each ODHS program area;
- Evaluate program effectiveness and service matching;
- Ensure research and analytics are replicable for other, related ODHS or state agency research;
- Enhance cross-system data sharing and collection to improve service delivery across the continuum of each service delivery area.
Social Complexity in Medicaid Youth
Literature has shown that social and environmental factors present in a child’s life greatly influence health outcomes as an adult, and that the effects of these experiences are cumulative. Some of these social determinants of health are identifiable through administrative data on service utilization within DHS. This project, in partnership with Oregon Health Authority and the Oregon Pediatric Improvement Partnership, seeks to identify available administrative data sources for social risk factors and provide an index of social risk to Oregon’s coordinated care organizations so they can provide additional services to children most at-risk.
Learn more about research and analytics across DHS programs and state agencies.
Child Welfare Screening Predictive Analytics Tool
ORRAI’s Research and Implementation units partnered with Child Welfare to launch an innovative new tool that utilizes predictive analytics to support child abuse allegation screening decisions. The Screening Predictive Analytics Tool uses Oregon historical child welfare outcomes and predictive models to generate probability scores that assist in decision-making about whether to assign a report to the child abuse hotline for further investigation. The predictive models combine the current report information with historical data from OR-Kids, the statewide DHS Child Welfare data system. The models use machine learning to leverage more than a hundred distinct variables (i.e., individual pieces of information) into thousands of combinations to generate the estimates.
Implementation convened Child Welfare leadership and staff and members of the Office of Equity and Multicultural Services (OEMS), the Office of Information Services (OIS), and the Office of Training, Investigations, and Safety (OTIS) in a series of workgroups that informed the development of the predictive model, implementation of the tool into OR-Kids and the screening process, troubleshooting and testing, communication to stakeholders, and staff training. The process was also supported by partnerships with Action for Child Protection
and Casey Family Programs
. Read more about the model and pilot implementation of the tool at the Oregon Child Abuse Hotline (ORCAH) in the Safety at Screening Research Brief.pdf
and the Safety at Screening Tool Development and Execution Report.pdf
Because predictive equations are created with historical data, they have the potential to perpetuate biases embedded in the data.
An additional, important component of the Safety at Screening Tool is the fairness correction procedure that minimizes bias while preserving accuracy. Read more in the Fairness Correction Manuscript.pdf