What Your VAI Score Actually Means

Your VAI Score is not what you think it is

Open the Vitality AI Health app and one of the first numbers you see is your VAI Score. It lives in a circular gauge on its own page and in a small widget in the sidebar on every other page. Most people assume it is a health score, a version of the Whoop Recovery or the Oura Readiness, calculated from biometrics.

It is not.

The VAI Score is a health data quality score. It measures how complete, how recent, and how high quality the data you have connected to the platform is. Not how healthy you are. How much the AI has to work with.

That distinction matters more than it sounds. This post explains why.

What the score actually measures

The VAI Score is a 0 to 100 number that answers a single question: how good is the data the AI is drawing on to analyze your health.

Under the hood, it evaluates nine data source categories. Each category gets its own score from 0 to 100, and those combine into the overall VAI Score.

The nine categories are: Blood Tests, DEXA Scans, Nutrition, DNA Results, Wearables, Environment, Supplements, Medications, and Health Profile.

For each category, up to three sub-metrics are evaluated: Completeness (how much of the available data for this category has been provided), Recency (how up to date is the data), and Quality (how reliable and granular is the data).

Not every sub-metric applies to every category, and that is on purpose. DNA Results only shows Completeness, because your genome does not change over time. Environment shows Completeness and Quality but not Recency. Medications and Health Profile show Completeness and Recency but not Quality, because those are user-entered fields where the platform trusts your input. Blood Tests, DEXA, Nutrition, Supplements, and Wearables show all three, because completeness, freshness, and data quality all meaningfully vary for those sources.

The result: a score that is honest about what it is measuring, category by category, instead of padding every category with irrelevant metrics to look more complete.

Why these nine categories

I picked these nine after mapping every type of consumer health data I could find and collapsing the redundancies. The goal was to cover the full picture of what people actually have access to, not what would exist in a hospital chart. Blood Tests and DEXA Scans stay separate because one is biochemical and the other structural; a blood panel tells you what is happening at the molecular level, while a DEXA tells you how your body is built. Collapsing them into a single "lab results" bucket would hide the difference, and that difference matters a lot when blood test markers and body composition data inform biological age signals. The other seven categories cover the behavioral, environmental, and genomic inputs that most longevity platforms either ignore or bury. If you want to understand why this platform was built around data depth rather than convenience, the reasoning behind the original design is here.

What a VAI Score actually looks like

Here is a real breakdown from the platform. The user has connected nine data sources, some more completely than others. Blood Tests score 96 (Completeness 100%, Recency 100%, Quality 85%). DEXA Scans score 100 across the board. Nutrition scores 87 with 77% Completeness. DNA Results score 100 on Completeness alone. Wearables score 80, held back by 50% Completeness. Environment scores 95. Supplements score 95. Medications score 100. Health Profile scores 100. Overall VAI Score: 92 out of 100. Rating: Excellent.

Notice what the breakdown tells you. Wearables scored 80 because Completeness is only 50%, even though Recency and Quality are both 100%. Translation: the wearable data this user has is recent and high quality, but the platform sees room to connect more wearable sources or fill in historical gaps. That is a specific recommendation, not a vague "improve your score" nudge.

The score is not just a number on a dashboard. When my wearable category dropped from 80 to 60 after I let my Oura subscription lapse, the dashboard surfaced the gap and Vaya Chat started prompting me about sleep questions it could no longer answer well. The score was telling me something specific: the AI's sleep reasoning had gone shallow, and I would know exactly what to do about it.

A beta user example

One early user, Marcus, came in with a VAI Score of 34. He had filled out his Health Profile and connected a Garmin, but nothing else. After uploading a recent blood panel and connecting his Oura ring, his score moved to 71 within a week. The change in Vaya Chat was immediate: responses that had been hedged with "based on limited data" started pulling in sleep recovery context alongside the metabolic markers, and the Healthspan Score shifted from Low Confidence to Moderate Confidence. He told me it was the first time a health app had explained its own limitations to him rather than just confidently getting things wrong.

Why build a data quality score at all

Because every AI health platform has a garbage in, garbage out problem and most of them hide it from the user.

If you have only connected your Garmin and nothing else, any AI running on your account can only analyze wearable data. It cannot cross-reference your sleep drop with a recent blood panel. It cannot flag a supplement interaction it has no visibility into. It cannot see the DNA marker that would change a nutrition recommendation. But most platforms will still give you a confident sounding answer. They just quietly narrow the question to what they happen to know.

The VAI Score flips that. Instead of the AI quietly delivering a shallow insight, the score tells you up front: here is how much data we have, here is where the gaps are, here is how much to trust what comes next.

This is consistent with what the research shows. A 2022 scoping review in NPJ Digital Medicine found that multimodal data fusion increases predictive accuracy by an average of 6.4% compared to single-modality approaches, and separately, a framework study published in NPJ Digital Medicine by Soenksen et al. found that adding data sources improves AI model performance by 9 to 28% across clinical prediction tasks. More complete inputs produce more reliable outputs. That is not a marketing claim; it is a pattern that holds across the literature.

It also sets a clean internal rule for the AI itself. A user with a VAI Score of 40 gets responses caveated accordingly. A user at 92 gets the full analysis. The confidence level the platform shows on outputs like the Healthspan Score is directly downstream of the VAI Score. More complete data, higher confidence.

Why the three sub-metrics exist

Completeness, Recency, and Quality each map to a different user behavior, which is why splitting them out matters.

Completeness answers: have I connected everything I could. It drives breadth. Connect more sources.

Recency answers: is my data current. It drives ongoing engagement. Keep syncing.

Quality answers: is the data detailed and reliable enough. It drives depth. Upload the full panel, not just the summary. Use the more granular wearable export.

Lumping these into a single score would let a user get to 90 by over-indexing on one dimension. Splitting them makes it clear that all three matter and that the strategy to improve each one is different.

How the score shows up in the product

The dedicated VAI Score page has a circular gauge that animates from 0 up to your score. Below the gauge is the rating label, a description, and the last calculation timestamp. Further down is a Score History section so you can track the trend over weeks and months.

The sidebar widget appears on every page of the app, showing your current number with a green progress bar and the label Health data quality. This keeps the number present without pulling you out of whatever you are doing.

The data source grid sits below the main gauge: nine cards in a 3x3 grid show each category with its score and sub-metrics. Every card is clickable. Drill in and you see exactly why that category scored the way it did and what you can do to improve it.

How it connects to the rest of the platform

The VAI Score is not a standalone vanity number. It is the substrate under every other insight the app gives you.

Vaya Chat shows how many sources are currently active. When Vaya generates a Health Summary Report for your doctor, it pulls from every active source and cites them explicitly. Higher source count and higher per-source quality means a more thorough report.

The Healthspan Score on the Overview Dashboard shows a confidence level next to it, for example Moderate Confidence. The note underneath reads: Confidence reflects how complete your connected data is. Add more history or connect more sources to improve it. That confidence tracks the VAI Score.

The Why we think this explainability feature on every insight shows the specific sources it used. The breadth and depth of those citations depends directly on how many sources are connected and how recent and complete they are.

Each Longevity Pillar (Cardiovascular Health, Sleep and Recovery, Body Composition, and others) shows a per-pillar data completeness badge. That is a downstream view of the VAI Score's per-source scoring, rolled up to the pillar level.

How to use your VAI Score well

Look at the overall number, then ignore it. The category breakdown is where the decisions live.

Find the lowest scoring category. It is almost always the highest leverage place to act. A wearable at 50% Completeness, a lab panel that is 18 months old, a supplement log that has not been updated in six weeks.

Check which sub-metric is dragging. Completeness, Recency, and Quality each have a different fix. Do not try to address all three at once. Pick the dimension that is furthest from 100 and work on it.

Watch the trend, not just today. The Score History on the VAI Score page shows how your data quality is moving over time. A score that has been climbing means the AI is getting more to work with every week. That is the story that matters.

Use it to calibrate trust. When the VAI Score is low, read Vaya Chat outputs as directional, not definitive. When the score is high, the same outputs carry more weight. The score is honest about its own limits, and you should be too.

The one number that makes every other number more useful

Most health apps show you dozens of scores. Readiness, recovery, sleep, strain, HRV, activity. The VAI Score is not another one of those. It is the meta-score that tells you how much the others can be trusted for you specifically.

That is why it lives in the sidebar on every page. That is why it animates when you open its dashboard. And that is why, when we talk about health signal, not noise, the VAI Score is the first tool in the kit. Before you can separate signal from noise, you have to know how much signal you actually have.

If you want to raise your score, the practical path looks like this: Marcus went from 34 to 71 by uploading one blood panel and connecting one ring. That single step changed how the AI reasoned about his sleep and his metabolic markers together. Connect the next source on your list. Check the category that is dragging your score. The improvement is usually closer than it looks.

If you have specific questions about your VAI Score or a category that is not behaving the way you expect, write to us at [email protected]. We read every message.