Insights / The Value of Nurse-Verified AI
AI Can Read Medical Records. But Can It Understand Them?
By Breen LNC.
Short answer: AI can read the words on a page with speed and consistency. Understanding a chart takes more: knowing which text was typed that day and which was copied forward, what normally belongs in the record, and what a missing entry might mean. That part still takes a clinician.
A chart is not a story written in order
A reader might assume each note reflects what the clinician observed that day. Published research says otherwise. A UCSF study of 23,630 progress notes found that on average only 18% of note text was typed by hand, while 46% was copied and 36% was imported by the system. In an outpatient study covering a decade of notes, text identical to the patient’s last note grew from 47.9% to 58.8%.
So when a note says the patient denies pain, a human reader who knows charting asks whether anyone asked that day or whether the line came forward from last week. The words on the page look the same either way. Each of these studies covers one center, so percentages elsewhere will differ.
What comprehension looks like in the research
Studies of AI summaries suggest that fluency and completeness can move apart. In JAMA Internal Medicine, AI discharge narratives scored better on concision and coherence and worse on comprehensiveness than physician narratives, with more errors per summary. They read well, and they left more out. That study looked at hospital discharge writing, not litigation, so it is a signal to weigh and not a measure of any legal product.
What a nurse brings to the page
AALNC describes legal nurse consulting as a licensed registered nurse’s “critical analysis” of clinically related issues, drawing on education, training and experience to evaluate standard of care, causation and damages. A nurse who has worked those shifts reads a chart knowing how it gets built:
- Which entries are likely templates
- What a routine assessment would normally include
- When the timing between an order and a dose looks odd
- When silence in the record is itself worth a question
Why the distinction matters to you
An attorney doesn’t need a summary that sounds right. You need to know what the record shows, how far you can rely on it and what the defense will say about it. Software helps you get through the pages. Breen puts a nurse on every entry so that someone who understands the chart has decided what it means.
Frequently asked questions
How much of a medical note is typed by the clinician?
In one UCSF study of inpatient progress notes, 18% of text was manually entered on average, 46% was copied and 36% was imported. That is one center in 2016, so percentages will differ elsewhere.
Why does copied text matter in a medical record review?
Copied text may not reflect what was observed that day. A reviewer who knows charting can look for that and weigh the entry accordingly.
Sources
- Wang et al., Characterizing the Source of Text in Electronic Health Record Progress Notes, JAMA Internal Medicine (2017)
- Rule et al., Length and Redundancy of Outpatient Progress Notes Across a Decade at an Academic Medical Center, JAMA Network Open (2021)
- Williams et al., Physician- and Large Language Model-Generated Hospital Discharge Summaries, JAMA Internal Medicine (2025)
- American Association of Legal Nurse Consultants, LNC Awareness Week
General information for legal professionals. This is not legal or medical advice.