Composite Health Scores

Why a single health score matters

Modern health tracking is fragmented. Your Oura Ring gives you a readiness score. Garmin calculates Body Battery. Whoop produces a recovery percentage. Each device sees only its own slice of reality, and none of them talk to each other. The result is conflicting signals: one app says you're recovered while another says you're strained. Omnio solves this by ingesting raw data from every source and computing composite health scores that reflect your whole picture. Instead of guessing which device to believe, you get one number grounded in all the evidence available. These scores are not black boxes. Every weight, every input, and every normalisation step is visible so you can understand exactly what is driving your number up or down on any given day. The multi-signal approach also preserves facets that single-number device scores tend to flatten: recent work in a 10-year, ~88,000-participant UK Biobank cohort found that sleep regularity is a stronger predictor of all-cause mortality than sleep duration^[3], which is why Omnio weights regularity and duration as distinct inputs rather than collapsing them into a single sleep figure.

How composite scoring works

Omnio builds your scores through a repeatable, auditable pipeline. The approach mirrors the multi-signal monitoring framework recommended by sports-medicine consensus on athlete training-load surveillance^[1]:

1. Raw metrics are pulled from every connected source -- sleep stages from Oura, HRV from Garmin, strain from Whoop, macros from MyFitnessPal, and more.
2. Each metric is normalised against population baselines and your personal history so that a 40 ms HRV and 7 hours of sleep land on the same scale. Personal-history normalisation is motivated by the established between-subject variability of HRV baselines^[2].
3. Weighted aggregation combines normalised inputs using a dependency graph. Sleep quality feeds into recovery, recovery feeds into readiness, and readiness feeds into your overall health score.
4. Confidence bands widen or narrow based on data freshness and completeness, so you always know how much to trust the result.

Key capabilities

Composite health scores in Omnio give you:

• Multi-source fusion across Oura, Garmin, Whoop, bloodwork, body composition, nutrition, and environment sensors
• Topological dependency ordering so upstream scores like sleep quality are computed before downstream scores like readiness
• Personal baselines that adapt over time -- your scores reflect your trajectory, not just a population average
• Backfill on demand so historical data gets scored consistently when you add a new data source or change weights
• Daily, weekly, and monthly trend views with sparklines on the dashboard

What makes this different

Most wearable scores are proprietary algorithms you cannot inspect. Omnio takes the opposite approach: every weight is documented, every input is traceable. The system adapts weights automatically based on your personal data patterns, so scores become more accurate over time. This transparency extends to the confidence model. When your Garmin has not synced in two days, the score does not silently degrade -- it tells you which inputs are stale and how that affects reliability. The result is a scoring system you can actually trust, one that improves as you connect more data and refines itself as your physiology changes.

What composite scores cannot tell you

Composite scores are informational tools, not diagnostic ones. They cannot identify or rule out a medical condition, replace clinical judgement during illness, or quantify psychological stressors that do not show up in HRV, resting heart rate, or sleep architecture. They are also sensitive to sensor artifacts: a loose ring, a missed Whoop strap sync, or an off-night of motion data can pull individual inputs around. Because the engine normalises against your personal history, the first few weeks after you connect a new device are the least reliable -- personal baselines fall back to population ranges until roughly 30 days of data are available. Scores are best read as one signal in a broader decision, not as an instruction.

Minimum data for a useful score

Every composite score has a minimum confidence threshold (default 0.4). Below that threshold Omnio suppresses the score rather than publishing a number dominated by a handful of available inputs. To clear the threshold reliably you typically want:

• Sleep score: a tracked night from Oura, Garmin, or Whoop -- ideally with stage data, HRV, and respiratory rate rather than just a duration estimate
• Readiness score: overnight HRV and resting heart rate plus a recent sleep score and some training-load context from the last 7 days
• Activity score: either wearable workout data (TRIMP, heart rate zones) or logged strength sessions, combined with daily step or active-minute totals
• Nutrition score: consistent meal logging in MyFitnessPal; a single logged meal is usually not enough to clear the confidence floor for the day
• Metabolic health score: a recent bloodwork panel -- this score is sparse by design and typically refreshes every few months

How confidence degrades gracefully

When an input is missing, Omnio redistributes its weight to the inputs that are still available, capped at 2x each category's default weight so a single signal cannot dominate the result. Confidence is reported alongside every score as the ratio of available input weight to total input weight, so a day with only wearable data and no nutrition or bloodwork will show a lower confidence band than a day with every source present. If any category carrying more than 10% of the total weight is entirely missing, the score is flagged as partial. Below the minimum confidence threshold the score is withheld rather than guessed. The goal is that a number on screen always reflects enough data to be worth acting on.

Not a clinical assessment

Composite health scores are designed to help you spot trends and ask better questions, not to replace care from a qualified clinician. They are not a medical device, they are not FDA-cleared, and they should not be used to diagnose, treat, cure, or prevent any disease. If a score changes in a way that worries you -- or if you feel unwell regardless of what the number says -- talk to a licensed healthcare professional. Omnio's job is to make your data legible. Interpreting it in the context of your full medical history is theirs.