TLDR: Healthcare data, generated from every patient interaction, fuels the AI tools transforming medicine. Understanding the types of data—quantitative (numerical) vs. qualitative (descriptive), and how they’re categorized (categorical, continuous, ordinal) and stored (structured vs. unstructured)—is crucial for leveraging AI effectively in patient care. Structured data (e.g., lab values, checkboxes) is easily processed by AI, while unstructured data (e.g., clinical notes) requires advanced AI. High-quality, consistent documentation is key to maximizing AI’s accuracy and improving patient care.

What is healthcare data?

Healthcare data encompasses all information generated during patient care, flowing from various sources: clinical encounters, diagnostic tests, administrative systems, medical devices and patient interactions via portals. This continuous data generation allows AI to identify patterns, predict outcomes and support clinical decision-making.

Fundamental types: Quantitative vs. qualitative

Data falls into two main categories:

• Quantitative data is measurable and numerical (e.g., blood pressure, lab values, medication dosages). AI analyzes this for trends and predictions.
• Qualitative data is descriptive and non-numerical (e.g., clinical narratives, patient histories). This requires more sophisticated AI, like natural language processing, to extract insights.

Understanding variable types in your practice

The specific nature of your data points, or “variable types,” influences how AI uses them:

• Categorical variables represent distinct groups or categories (e.g., blood type, diagnosis codes). AI uses these to classify patients and predict group-based outcomes.
• Continuous variables can take any numerical value within a range (e.g., vital signs, lab results). AI excels at identifying subtle patterns in this data.
• Ordinal variables have a natural order but aren’t truly numerical (e.g., pain scales, functional status ratings). AI uses these for progression modeling.

Structured vs. unstructured data

• Structured data is organized and predictable (e.g., dropdown menus, checkboxes, numerical entries in EHRs). It’s easily analyzed by AI.
• Unstructured data includes clinical notes, dictated reports and free-text entries. This narrative information often holds nuanced insights but requires sophisticated natural language processing for AI interpretation.

Why this matters for your practice

Understanding these data fundamentals directly impacts your ability to leverage AI effectively:

• Quality documentation drives AI accuracy. Precise and complete information, whether in structured fields or detailed notes, provides high-quality input for AI systems, leading to more reliable insights.
• Consistency enables pattern recognition. Standardized terminology and consistent documentation help AI identify meaningful patterns to improve care for future patients.
• Awareness prevents misinterpretation. Knowing how AI processes different data types helps you recognize when AI recommendations might be incomplete and when human judgement is essential.

Looking ahead

As AI integrates further into your workflow, your role as both a data generator and interpreter become increasingly important. The information you document today forms the foundation for AI systems that will support tomorrow’s clinical decisions. Next month we’ll explore how data quality directly impacts patient care and discuss strategies for optimizing your documentation practices for both human and AI consumption.

Don’t forget to take the HonorHealth EHR Experience Survey

Your feedback directly improves the technology tools you use daily. HonorHealth is conducting an EHR Experience Survey to identify optimization opportunities and reduce documentation burden.

Why participate? Your responses will shape future EHR upgrades, training programs and workflow improvements. The survey takes 10-15 minutes, is conducted by KLAS Research, and your responses can remain confidential.

Get started

Want to shape how we use AI?

Join the Clinician Technology Experience Council (CTEC). Meetings are every other month on the fourth Wednesday at 5 p.m. Sign up here to have meetings added to your calendar.

Have questions or want to learn more? Reach out to Craig Norquist, MD.