The new HIMSS Analytics Maturity Assessment Model supports smart AI implementations

“Analytics as a discipline has changed dramatically over the last five to 10 years – and certainly over the last five years,” said Anne Snowdon, chief scientific officer at HIMSS. “With the explosion of artificial intelligence – the ChatGPT era, if you will – large language models have really shifted the question of where and how these advanced analytics tools provide value to healthcare.”

In other words, it’s a good time to… HIMSS Analytics Maturity Assessment Modelwhich was first launched in 2016 as a benchmarking framework to help hospitals and healthcare systems sharpen their analytics programs and data management efforts.

Eight years ago, AMAM helped healthcare organizations track their use of analytics technology in eight steps, from Phases 0 and 1 (fragmented point solutions and early data aggregation efforts) through Phases 6 and 7 (clinical risk intervention and predictive analytics; personalized medicine and prescriptive analytics).

With the original model, healthcare systems like UNC Health Care and Children’s Hospital Colorado demonstrated the value of pursuing and achieving Phase 7 – achieving big gains in process efficiency and patient outcomes.

With artificial intelligence and automation poised to transform every corner of healthcare, the assessment model has been redesigned from the bottom up and made available to healthcare systems around the world.

‘What are you achieving?’

Officially launched earlier this month at the HIMSS APAC Health Conference & Exhibition 2024, the new AMAM is not just a benchmark for analytics adoption, but a way to measure the real impact of analytics, AI and data-driven decision making on enterprises . -wide business operations and quality of care.

An emphasis on patient outcomes is critical, Snowdon says.

“It’s not, ‘Do you have AI?’” she says. “The question is: ‘What can you achieve now as an organization or a system, given your advanced maturity or your analytical maturity? What are you achieving, for whom? That’s a fundamental shift from the previous model.’

The new AMAM is designed to measure the impact of analytics initiatives within a healthcare system: how they impact quality and safety, patient and population health, operational and financial performance, and more.

It now focuses on other areas such as governance, privacy and security, the analytics lifecycle and promoting a culture of responsible analytics – while also including capabilities for real-time prescriptive and predictive analytics, natural language processing and other advanced AI applications.

AMAM modernization “is not only about keeping pace with this rapid evolution of analytics technologies and their potential value, but also about potential risks,” Snowdon says. “As you advance your use or consider the use of things like artificial intelligence, do you have the data, the data quality, so that that AI tool or technology is going to be accurate? Will it be fair?

“Models can be trained on a lot of data from one sector, the large sector of the population, but it can actually be quite damaging to another sector of the population,” she adds. “In Canada, for example, we have a lot of data on Asian patients. We have much less data about our indigenous community. How is an AI model going to work for that indigenous community if the model has never been trained on data representative of them?”

And the risks of bias and inaccuracy due to bad data are not alone. The challenges of AI-based analytics are “very different now compared to what we’ve seen in the past, given the nature of these technologies,” Snowdon says.

“The risk here is multi-layered: from the perspective of infrastructure data, through the perspective of patient care and outcomes, to the perspective of accuracy, fairness and data integrity, AI tools are being used to inform decisions.

“It’s very layered and this model moves forward and supports organizations to understand all the variations and levels of risk as their analytics maturity evolves over time.”

The first few phases of the new AMAM – which joins other HIMSS models, including the Infrastructure Adoption Model and the flagship EMR Adoption Model, which has been refreshed in recent years – are focused on helping participating healthcare systems build foundational data management and quality measures, while data is collected repositories that build expertise in dashboards and data visualizations to support decision making that is strategically aligned with the organization’s objectives.

At the top of the ladder, phases 6 and 7, healthcare organizations will use predictive analytics to inform healthcare decisions and integrate AI and machine learning into their analytics processes, with real-time clinical decision support. They will also have systems in place for monitoring population health outcomes and establishing health equity programs.

HIMSS (parent company of Healthcare IT news), notes that AMAM is designed as a flexible framework, rather than a rigid checklist, and is intended to be used in all healthcare settings to help healthcare systems refine and improve their data strategies and decision-making.

“It’s really a strategic roadmap for advanced, very advanced analytics, which at levels 6 and 7 in this model is heavily focused on artificial intelligence,” Snowdon says.

“We have extensively tested this new AMAM model with our partners and organizations who are very familiar with the AMAM model,” she adds. “The overwhelming feedback we got from customers who have used the current AMAM model is, ‘This is what I need. It gives me my roadmap to go to my CEO and C-suite executives to help them see where we stand now and where we stand ‘we have to go there”.