From Data to Action: Why Public Health Data Readiness Must Evolve in the AI Era

Artificial intelligence is rapidly becoming part of everyday public health practice. Much of the discussion, however, still focuses on AI models, algorithms, and predictive analytics. I believe we should first ask a more fundamental question:

What makes public health data ready for AI?

Many organizations still define data readiness primarily in terms of data quality: accuracy, completeness, timeliness, and consistency. These characteristics remain essential, but they are no longer sufficient.

As illustrated in Figure 1, public health does not stop with data. It transforms data into information, knowledge, situational awareness, decision support, and ultimately public health action.

Figure 1. Seven‑stage public health information lifecycle illustrating how AI‑enabled capabilities augment every stage, from data acquisition and semantic harmonization to information generation, knowledge synthesis, situational awareness, decision support, and public health action, while remaining anchored in governance, standards, human oversight, and continuous monitoring.

This entire information lifecycle determines whether surveillance systems, dashboards, risk assessments, and emergency operations successfully support population health.

Artificial intelligence is transforming every stage of this lifecycle.

As shown in Figure 2, AI is no longer limited to predictive modeling. It increasingly assists with data acquisition, semantic harmonization, information synthesis, knowledge discovery, situational assessment, decision support, and outcome monitoring, one capability for each stage of the lifecycle.

Fig 2. Why Public Health Data Readiness Matters in the AI ERA

In other words, AI is becoming an enabling capability across the entire public health information ecosystem.

This shift has an important implication.

If AI participates in every stage of the information lifecycle, public health data readiness must extend beyond data quality alone. Organizations must also consider semantic readiness, metadata quality, interoperability, documentation, standards, governance, continuous monitoring, and human oversight. These capabilities determine whether AI can operate safely, transparently, and effectively in real-world public health settings.

Rather than viewing AI simply as a consumer of prepared data, we should also recognize its emerging role as a contributor to data readiness. AI can accelerate semantic harmonization, improve metadata management, synthesize information, produce high-quality visualizations, and support situational awareness. At the same time, AI-generated outputs require governance, validation, and continuous evaluation.

I believe the next generation of public health informatics should move beyond asking whether data are AI-ready and instead ask a broader question:

How can we build public health data ecosystems that are truly ready to support trustworthy AI-enabled public health?

At the Center for Applied Medical AI (CAMA), we are working on exactly this problem.

I would welcome perspectives from colleagues working in public health, health informatics, emergency preparedness, data science, and AI governance. How do you define public health data readiness in the AI era?