Data Analytics

The Building Blocks of Data Literacy for Healthcare

To use artificial intelligence effectively, healthcare organizations need to foster a culture focused on data governance and quality.

Dean Burrill

Dean Burrill is a Healthcare Strategist for CDW.

Data literacy is the organizational ability to capture, evaluate, normalize and transform available data sources into actionable business insights. It is as much about culture as it is about processes and tools.

To understand whether your organization has good data literacy, ask yourself these questions. Do you know which data sets you have? Do you know the quality of those data sets? Do you know which ones are sensitive or not? Do you know which buttons or levers that bit of data pushes when you’re making a decision about something? That is what data literacy means. The next step is figuring out how to make your entire organization care about the answers to these questions.

Data literacy is becoming more important in healthcare with the growing use of artificial intelligence. AI tends to be seen as an authority. Humans have been conditioned to take whatever the machine tells us as truth and to act on it accordingly. With that in mind, it’s vital that an organization ensures the truth really is the truth when it comes from AI.

Consider the “garbage in, garbage out” nature of AI. These models are trained on data, and the decisions they make are based on that data. It’s not that bad data is going to create a bad output; it’s going to create a bad model. That model is going to make bad decisions. It’s imperative that organizations understand the data they’re feeding AI models so they can get desired, accurate outputs.

Healthcare AI is pervasive, and people might not know when they’re interacting with it. In healthcare, it has the potential to improve workflow efficiencies — the pace at which people work, the volume they can handle and the quality of what they produce. However, if they don’t understand the data used to train the AI, how will they know when the output is suspect? Teams working with AI models need to understand the data being fed into these systems and how they work to avoid adversely affecting patient care. To achieve that, an organization needs to make data literacy their cultural norm.

Click the banner below to find out how to take advantage of data and AI for better healthcare outcomes.

ht-dataandai-animated-2025-prepare-desktop

ht-dataandai-animated-2025-prepare-mobile

Who Is Responsible for Data and AI Literacy?

Defining roles is a long-standing challenge when rolling out any form of governance, but particularly data governance. In the minds of many business leaders, “data” belongs in IT. And that may be true for moving and processing data. But the “how” of moving and processing data is a business responsibility. Unfortunately, in most organizations, IT ends up with all the data that’s been filtered into the enterprise, and it’s required to make sense of that and deal with any related challenges with sporadic business input.

In an ideal state, data governance is seen as a business process. It lives under a COO, CIO or (even better) a chief data officer — someone at a high level in the operational side of the organization. That person is responsible for ensuring that data management and data quality are everybody’s job.

Everyone has a responsibility to make sure the data is complete, from nurses and providers entering patient data in the electronic health record to the quality staff that changes the definition of a bed-day based on government program requirements. It’s important that the correct code sets and vernacular are used. And if there’s bad data somewhere, clinicians need to be empowered to act on that and raise the issue. There should be a portal available for them to look up the correct definition of the data and to mark bad data for review. When someone sees a red flag, they must be able to act on it, and that capability doesn’t exist in a lot of organizations.

Data governance ROI can be difficult to prove because it’s a broad initiative that touches all areas of an organization and can take years to fully put in place. The ROI also depends on management buy-in. If leadership isn’t on board, it’s going to be hard for the benefits of data governance to trickle down. ROI can be an obstacle for data governance initiatives, and IT leaders may have to get creative in monetizing the benefits.

EXPLORE: This AI and data readiness checklist can help guide your organization toward AI adoption.

What Can Healthcare IT Leaders Do To Improve Data Literacy?

The single biggest thing IT leaders can do is create a culture that is data-centric, data-aware and data-driven. It’s hard to do. It has to start at the very top. Leaders need to walk the talk; they must exemplify the use of data in every decision. For every recommendation they receive or decision they make, leaders should ask to see the data, analysis and ROI, and they should probe the assumptions and quality of the data.

Data quality and management need to be written into job descriptions and evaluated as part of performance reviews. Often, a focus on data isn’t absent, but it needs to be called out and exist as a specific goal across the organization, and be actively measured.

Proper data assessments are crucial to understanding the organization’s data ecosystem and measuring the quality of the data. Do you know where your data is and how much of it you’ve got? Do you know how frequently it arrives? How much does the company rely on each data set? What is the lineage of the data — where does it come from, what is done to it and where does it go? This isn’t something an organization needs to do all at once, but it should have a roadmap that includes all of those steps prioritized for each data domain to be able to drive change.

Click the banner below to sign up for HealthTech’s weekly newsletter.

There are several tools available to help healthcare organizations improve their data governance and quality. Collibra is a data governance tool that helps organizations build a catalog to drive workflows around the governance itself. Anomalo is a data quality tool that uses AI to validate and scrub data. It finds data quality patterns, surfaces them and helps organizations correct them. CDW has partnerships with both solutions and many others.

The important thing to remember is that tools only work when there’s a culture in place to use them. The culture needs to drive requirements that drive tool selection. Organizations often think buying a tool solves their problems. But these are complicated tools that require a lot of configuration, administration and care. Their licenses can be expensive, and you don’t want to invest in them unless you’re going to use them fully.

An investment in a tool such as Collibra or Anomalo implies that you have a cultural framework, processes, committees and key roles identified — a community of practice. A healthcare organization’s community of practice makes these tools effective. But a community of practice takes it further. It makes caring for data a passion, even a social occasion. The governance process becomes self-governing, and the stewards involved hone the processes, call out achievements, recognize outstanding contributions and even have fun. One organization’s data culture is so strong that its governance committees have an annual picnic where staff give presentations on data initiatives they’ve undertaken and what’s worked versus what hasn’t. This excitement helps to break down the silos and create a culture of sharing around data and data best practices.

Achieving Data Literacy With Strategic Partnership

CDW has provided training and consulting in data governance for over a decade. It is clear that AI governance is a new layer of the onion of data governance, and we’ve extended our existing toolsets to include AI governance.

We partner with healthcare organizations and conduct five-day workshops in which we train the organization on what governance is, the major concepts and what the roles are, and then we help identify the individuals within the organization who will steward data governance.

Minimum viable data governance is a new focus for CDW. Rather than having to boil the ocean and attack the entire enterprise for governance, we suggest picking the two or three most strategic projects, identifying data domains and stewards for those areas, and setting them up to catalogue and define their data. We look at existing tools to see what can be leveraged to help with the metadata and process flows. Products like Databricks and Snowflake have capabilities that can be used for governance. The methodology is also helpful for gaining leadership buy-in, breaking governance into smaller pieces and requiring fewer expensive tools up front.

This approach can reduce the barriers to governance and provide some quick wins, which will help create momentum and give birth to a new community of practice and a data-centric culture.

This article is part of HealthTech’s MonITor blog series.