Advanced analytics is core to many of CIHI’s services and products. CIHI has a long-standing reputation for producing high-quality products and tools using advanced analytical techniques.
As the volume, breadth, richness and complexity of health care data continues to increase, and with advancements in chip architecture and research, the adoption of cloud computing and the availability of machine learning solutions and open-source offering, CIHI is expanding its efforts in predictive modelling and the use of machine learning both to change how we work and to support our stakeholders’ needs.
Strategic initiatives to support advanced analytics at CIHI
- Expand our in-house capacity to conduct machine learning and predictive modelling
- Modernize CIHI’s analytical toolkit and develop an open-source policy to support the adoption and use of open-source tools such as R and Python
- Explore privacy enhancing and synthetization technologies to enhance de-identification and protection of personal health information
- Incorporate machine learning into existing operational activities to create efficiencies and boost innovation in the development and operation of CIHI products
- Develop a data science learning program and deliver to CIHI staff
- Share and exchange experiences and explore collaboration opportunities with our stakeholders on the use of machine learning and predictive modelling to inform health care policy and the management of health systems
CIHI works with external partners to identify opportunities where machine learning and predictive modelling can be used to solve real-world problems and provide meaningful and actionable insight on our health care systems. Listed below are a few projects that we have undertaken, in collaboration with external partners.
Data Science Learning Program
To ensure that CIHI analysts continue to have the analytical skills and expertise to support CIHI’s strategic goals and priorities, CIHI has embarked on the development and delivery of the Data Science Learning Program for CIHI employees. The program includes coding and analysis using R and Python, big data and high-performance computing, statistical and machine learning techniques, and responsible data science. CIHI has partnered with the University of Waterloo to develop and deliver the program.
Modernization of CIHI’s analytical toolkit
With the evolution in the software used by the analytics community, including CIHI’s clients and peer organizations, CIHI has initiated a strategy to modernize its analytical toolkit and to leverage open-source technologies, including R and Python. This strategy will enable greater innovation and efficiencies in the development and delivery of CIHI’s products and strengthen CIHI’s analytical practice and engagement with the analytics community.
Tool to forecast demand for general internal medicine specialists
As the population ages and grows, hospital managers want to ensure appropriate staffing levels to meet the population’s growing health needs. CIHI developed an interactive tool that forecasts demand for general internal medicine (GIM) specialists to provide hospital planners with advanced warning about potential gaps in capacity to provide care.
The tool provides users with a 5-year projection on the overall demand for GIM specialists, as well as the option to create multiple scenarios to test the impact of changes in population morbidity and various health administrative decisions (e.g., staff mix, hospital mergers).
The tool is being piloted in a large teaching hospital in Ontario, testing the feasibility of using it to assist with long-term planning and budgeting for future staffing. Going forward, CIHI intends to make the tool available to other facilities to support their health workforce planning.
Case-mix products and predictive modelling
The use of predictive modelling and machine learning are an integral part in the development and ongoing enhancement of case-mix products. CIHI’s grouping methodologies use decision tree algorithms to convert clinical data, such as diagnosis and interventions, into patient classification systems. CIHI’s grouping methodologies include predictive models to estimate resource utilization, such as associated health costs, lengths of stay and expected use of health services like ED and primary care visits. To learn more about CIHI’s case-mix products, visit our Case mix web page.