How CareNav Uses AI to Help Families Compare Nursing Homes

At CareNav, we believe families deserve clear, balanced, and honest information when choosing a nursing home. That is why we built an AI-powered analysis layer on top of the federal data and family reviews already available for every facility. This article explains how the system works, what it produces, and why we designed it the way we did.

The problem we are solving

When a family is evaluating a nursing home, they typically have two sources of information: the CMS star ratings (which are numerical and can be hard to interpret without context) and Google reviews from other families (which can number in the dozens or hundreds and are time-consuming to read through individually).

Neither source alone tells the full story. A facility might have a 4-star CMS rating but a pattern of complaints about communication in its Google reviews. Another facility might have a 3-star rating but overwhelmingly positive reviews praising its staff's compassion and attentiveness. The ratings and the reviews together paint a more complete picture — but only if someone takes the time to synthesize them.

That is what our AI does. It reads and analyzes the public reviews available for each facility and produces a structured summary that highlights both the strengths and the areas for improvement. The goal is to save families hours of reading while giving them a balanced, honest overview.

How it works

Our AI analysis pipeline operates in three stages. First, we collect and cache public reviews from Google for each nursing home in our database. This happens through an automated background process — reviews are refreshed periodically to keep the data current. At no point are reviews fetched live during a page visit, which keeps costs predictable and page loads fast.

Second, the collected review text is passed to a large language model with a specific set of instructions. The model is asked to identify recurring positive themes (things families frequently praise), identify recurring concerns or areas for improvement (things families frequently mention as negatives), and generate a sentiment score from 0 to 100 based on the overall tone and content of the reviews.

The model is instructed to be balanced. It is not designed to make facilities look good or bad — it is designed to surface the themes that appear most frequently in what real families have written. If most reviews praise the staff but several mention issues with food quality, both themes will appear in the summary.

Third, the output is stored as structured data — an array of strengths, an array of improvement areas, and a numeric sentiment score. This data is displayed on the facility page in the AI Insights panel, clearly labeled as AI-generated and showing the number of source reviews analyzed.

What you see on a facility page

The AI Insights panel on each facility page shows two columns. The left column lists strengths — the things families tend to appreciate about the facility, prefixed with a plus sign. The right column lists areas for improvement — things that come up as concerns, prefixed with a minus sign.

At the top of the panel is a sentiment score, which represents the overall positivity of the reviews on a scale from 0 to 100. A score above 70 indicates predominantly positive sentiment. A score between 40 and 69 indicates mixed sentiment. A score below 40 indicates predominantly negative sentiment.

At the bottom of the panel is an attribution line that reads "AI-generated by CareNav" along with the number of reviews that were analyzed. This transparency is deliberate — families should always know that this summary was produced by AI, not written by a person, and they should know how much data it is based on.

What the AI does not do

Our AI does not modify, filter, or remove reviews. The original Google reviews are displayed separately on the facility page, unedited and in full. The AI summary is a supplement to the reviews, not a replacement.

The AI does not change or influence CMS star ratings. Those ratings come directly from the federal government and are displayed exactly as published. CareNav has no ability to modify CMS data, and neither does any facility.

The AI does not generate fake or synthetic reviews. Every insight in the AI summary is derived from real reviews written by real people. If a facility has no reviews, the AI panel will indicate that insights are not yet available.

Why transparency matters

We label every piece of data on CareNav with its source — CMS data is marked as coming from CMS, Google reviews are marked as coming from Google, and AI summaries are clearly marked as AI-generated. This is a deliberate design choice.

Families making nursing home decisions are often under stress and time pressure. They need to be able to trust the information in front of them. By being explicit about what is government data, what is from other families, and what is AI-generated analysis, we help families calibrate their trust appropriately.

We also allow facilities that subscribe to CareNav's Pro or Premium plans to submit a response to the AI summary — similar to how a business can respond to a Google review. This response is displayed alongside the AI summary in a visually distinct block, clearly labeled as provided by the facility. Facilities cannot modify or override the AI summary itself.

How facilities can participate

Facilities that claim their listing on CareNav can subscribe to a plan that gives them the ability to respond to their AI-generated summary, add photos and descriptions to their profile, and track how families interact with their listing. This does not change the AI analysis — it gives facilities a voice alongside it.

We believe this balance — objective data and AI analysis that cannot be edited by the facility, combined with the facility's ability to provide their own context — serves families best.

Continuous improvement

Our AI analysis is not static. As new reviews are posted and cached, the summaries are regenerated to reflect the most current feedback. If a facility improves (or declines) in a way that is reflected in new reviews, the AI summary will update accordingly over time.

We are also continuously refining our analysis methodology. As we gather more data on how families use CareNav and what information they find most valuable, we will evolve the AI insights to better serve their decision-making process.

If you have questions about how our AI works or feedback on the insights displayed on a facility page, reach out to us at support@carenav.io. We are always listening.