An important aspect of identifying and reducing health disparities is data collection, which is an area of ​​focus for Arizona health care leaders. State health officials met at the 2023 Arizona State of Reform Health Policy Conference last month to discuss improvements in data collection across the state.

Siman Qaasim works as the assistant director for policy and intergovernmental affairs at the Arizona Department of Health Services (ADHS). He said it’s important to understand how health departments—including public health departments—collect population data, and to refine it to understand health conditions at the local level.

“What I would like to see, and we’re working on this a little bit in our Data Advisory Committee that we have in the department, which is the real champion of data collection, more diverse race and ethnicity (data), and of course. SOGI data —that’s sexual orientation, gender identity, and expression data. We need that data to know who’s experiencing what.”

— Kasim

ADHS data analytics section lead, Wesley Kortuem, discusses how artificial intelligence is being used to improve health equity. Kortuem said the department is working on a software called Tamr that allows ADHS to take data and use a machine—which they learned in training—to identify when an individual is in a database. same as an individual in a separate database.

“Having machine learning in place, that allows us to bring two datasets in—multiple datasets, actually—and then we can tell the machine what the layout of the two tables is and give it a certain variables-address, city, state, zip (code), race, ethnicity, name, all these things, whatever we have-and then let the machine look at these different records and will try to match it,” said Kortuem.

Pairs can be looked up and determined if they are the same individual in separate databases. Regardless, Kortuem says the machine learns as a result. This process must be continued until the machine is fully trained to make that distinction among individuals.

Kortuem explains how from there, a master person index can be created, identifying each individual in each siled dataset. An additional modernization initiative came from apparent areas of improvement during the COVID-19 pandemic.

“We’re making improvements from where the data is entered, to the final analysis, so, we can get data from multiple sources—death certificates and so on,” Kortuem said.

Matthew Isiogu, chief revenue officer of Contexture (the leading health information exchange in Arizona and Colorado), noted how important it is for individuals to have multiple options for submitting identifying data about themselves.

He said he was at a meeting earlier this year where the team looked at data on race and ethnicity from all hospitals and health systems. Isiogu said that when looking at the individual level, the team noticed how some chose more than one race, and that the option to choose multiple careers was the basis of the data. On the other hand, this data pattern may not be the way hospital papers, for example, are designed, which may force individuals to choose between one race option or another.

“I think we are at a critical time when there is interest in sectors to do this work,” Isiogu said. “State agencies are very creative to create funding opportunities and strategic alignment among agencies, but it’s just as important—if not more important—that when we make these decisions, the structure, the data about the governments of that data, that we have different views and representations on the table.”