Original research asset · version 1.0 · 11 September 2026
Short answer: an overnight oxygen report is most useful as a structured record of signal quality, oxygen estimates, pulse behavior, recording time, and repeated patterns. It is not the same thing as a sleep study, and it should not be treated as a diagnosis on its own.
This atlas gives you a neutral vocabulary for discussing home SpO2 data. It combines a practical data dictionary, a clearly labeled synthetic example, a source-linked evidence table, and a repeatable quality checklist.
The CSV contains synthetic five-minute readings with signal-quality labels and teaching segments. No patient data is included.
How to use this home oxygen data atlas
Use the page in three layers. First, use the data dictionary when a report uses unfamiliar labels. Second, use the pattern library to describe what a graph looks like without jumping to a diagnosis. Third, use the evidence table to keep claims proportional to the research behind them.
If you already have a report, begin with the plain-language guide to reading an overnight SpO2 report. This atlas adds the research context and a shareable example dataset rather than repeating that article.
The data dictionary: what the common fields actually mean
Different devices use different labels, averaging windows, sampling rates, event rules, and report layouts. Treat the names below as a translation layer, not as proof that two apps calculate the same value in the same way.
| Report field | Plain-language meaning | Important limitation |
|---|---|---|
| SpO2 | An estimate of peripheral oxygen saturation from an optical sensor. | It is an estimate, not a direct arterial blood-gas measurement. Motion, low perfusion, temperature, fit, skin pigmentation, and other factors can affect the signal. |
| Pulse rate | The pulse rate detected by the same sensor at a given moment. | A pulse value can look plausible even when the oxygen signal is noisy. Check the quality indicator and the shape of the trace. |
| Recording time | How long the device was switched on and collecting data. | Recording time is not automatically the same as sleep time. That distinction matters when a metric is expressed per hour of sleep. |
| Minimum SpO2 | The lowest reported value in the selected period. | A single low point can be artifact, movement, poor contact, or a real event. It needs context from duration, recovery, and signal quality. |
| Average or median SpO2 | A summary of the values in a recording period. | An average can hide short repeated dips. A median and a distribution are often more informative than one headline number. |
| ODI 3% or ODI 4% | The number of oxygen desaturations meeting a device's chosen percentage-drop rule per hour of analyzed time. | ODI is not AHI. Cutoffs, baseline rules, duration rules, and the denominator can vary, so the number is not portable between all devices. |
| Time below a threshold | The duration spent below a displayed saturation threshold, often shown as a percentage or minutes. | The threshold is not a universal diagnosis line. Review the report definition, recording quality, and clinical context together. |
| Signal quality or invalid time | The portion of the recording the device considers usable, weak, missing, or affected by artifact. | A long recording with a low usable-signal percentage can be less informative than a shorter clean recording. |
| Event markers | Flags created by the device or its software when a rule is met. | An event marker is a software classification, not a confirmed cause. It should prompt review, not a self-diagnosis. |
ODI is useful, but it is not a replacement for AHI
The oxygen desaturation index, or ODI, counts oxygen drops that meet a rule such as a 3% or 4% fall from a local baseline. The apnea-hypopnea index, or AHI, counts breathing events per hour of sleep using a sleep-study method. They can move in the same direction, but they are not interchangeable.
That difference matters because an oxygen drop can have several explanations, and not every breathing event creates the same oxygen response. A device may also use recording time rather than confirmed sleep time. The 2020 systematic review of ODI for adult OSA found eight eligible studies covering 1,924 patients, with wide ranges for sensitivity and specificity and important heterogeneity in how events were defined. The practical conclusion is not that ODI is useless; it is that a number needs its method attached to it.
When you share an ODI, include at least the device or software version, the rule shown in the report, the analysis period, the usable-signal percentage, and whether the number is based on recording time or estimated sleep time.
A synthetic overnight atlas: four patterns worth learning
The visualization below is an intentionally synthetic teaching example. It uses four segments that a report reviewer may encounter: a stable high-quality trace, an isolated questionable dip, a repeated desaturation-like pattern, and a sensor-contact interruption. The shapes are not taken from a patient and should not be used to set personal thresholds.
1. Stable and usable
A relatively narrow band with a good signal-quality indicator is the easiest segment to use for trend review. It still does not prove that sleep was normal, but it is a better foundation than a noisy trace.
2. Isolated dip
One sharp fall followed by quick recovery deserves a quality check before it is counted as a meaningful event. Look for movement, cold skin, loose fit, or a matching pulse artifact.
3. Repeated pattern
A repeating shape can be worth sharing with a provider, especially when it appears across more than one night. The pattern still cannot identify its cause from oxygen data alone.
4. Contact gap
A flatline, blank section, or sudden impossible transition is a data-integrity problem first. Do not turn missing sensor data into a low-oxygen conclusion.
5. Pulse co-movement
When pulse and oxygen change together, it may add context, but it does not explain why the change happened. Treat synchronized movement as a question for follow-up.
6. Night-to-night check
Repeating the same setup over several nights can make a trend more useful. It does not remove the need for clinical interpretation when symptoms or risk factors are present.
What the evidence supports, and where it stops
The table below is designed to be copied into a briefing, article, or clinical conversation without inflating what a home monitor can establish.
| Evidence point | What it supports | What it does not support |
|---|---|---|
| FDA pulse-oximeter guidance | Pulse oximeters estimate oxygen saturation and pulse, and several physical or device factors can influence the reading. | That any single home reading is automatically accurate or that a device can explain a symptom by itself. |
| ODI systematic review | ODI can be useful for screening or triage in selected contexts, and it can correlate with sleep-study measures in some populations. | That one ODI threshold is universal across devices, populations, scoring rules, or clinical questions. |
| Repeated-night study | Night-to-night variability is real, so repeated measurements can change the picture in people being evaluated for OSA. | That more nights automatically make a consumer device diagnostic, or that a person should change treatment without a provider. |
| AASM position statement | Home sleep apnea testing has a defined clinical role for selected adults and requires provider involvement and interpretation. | That an automatically scored oxygen graph is a general screening test for anyone without symptoms or risk assessment. |
| Optical measurement research | Skin pigmentation and other optical conditions belong in a serious discussion of device performance and validation. | That performance differences can be predicted from a single screenshot or generalized to every device and person. |
The six-step quality protocol for a useful overnight record
- Define the question before recording. Are you checking whether the sensor stays on, documenting a pattern to discuss with a provider, comparing nights after a change, or simply learning how the report works? A clear question prevents you from collecting numbers without a decision context.
- Use the same placement and setup. Follow the device instructions, keep the sensor comfortably secure, and note anything that changed. Fit, temperature, circulation, movement, nail products, and skin contact can all matter for an optical sensor.
- Keep a small context log. Record the approximate recording time, unusual sleep disruption, altitude or travel, illness, oxygen use if prescribed, and whether the device was removed. These notes are often more valuable than an isolated minimum value.
- Check signal quality before counting events. Mark sections with low quality, missing data, movement, or a loose sensor. A clean section and an artifact-heavy section should not carry the same weight.
- Compare patterns, not screenshots. If the same shape appears over several nights with consistent setup, it is more useful to discuss than one dramatic-looking dip. Compare the device's own definitions and keep the raw report available.
- Share the report with the right context. Include the full time range, signal-quality information, device model, app version if available, and your question. A provider can decide whether the data fits a larger evaluation.
How to describe patterns without over-reading them
Use neutral language when you write about a home report. Say “the trace shows repeated drops during the recorded period” rather than “this proves sleep apnea.” Say “the report flags low-quality segments” rather than “the device failed.” Say “the pattern is worth discussing” rather than “the result is dangerous.” That difference is not just tone; it keeps the claim aligned with the evidence.
The FDA notes that pulse-oximeter readings can be affected by factors such as poor circulation, skin temperature, skin pigmentation, skin thickness, tobacco use, and nail polish. A useful report review therefore starts with the possibility of measurement error before assigning a biological explanation. If a person has concerning symptoms or a provider has instructed them to seek care, the report should support that conversation, not delay it.
Also separate trend from threshold. A trend asks whether the same behavior appears repeatedly under similar conditions. A threshold asks whether a device-defined number was crossed. Both can be useful, but neither answers the clinical question alone.
How this asset can be cited or reused
This page is intentionally built as a linkable reference. A publisher can cite the definition table, embed the synthetic chart with attribution, or use the CSV to demonstrate a data-cleaning workflow. If you reuse the CSV, keep the label “synthetic teaching data” attached to it and link back to this page:
Suggested citation: NightlyVitals, “Home Oxygen Data Atlas: Reading Overnight SpO2 Evidence in 2026,” September 11, 2026. Synthetic example data only; evidence links and limitations are documented on the page.
The CSV is deliberately small and transparent. It includes timestamps, a synthetic SpO2 estimate, synthetic pulse rate, signal quality, a teaching segment label, and an event flag. It is suitable for a chart demo, an editorial explainer, or a test of report-import software. It is not suitable for clinical research, model training claims, or personal medical decisions.
A compact annotation template for reports
If you are reviewing several nights, use the same short annotation on each file. Consistency makes the comparison more honest than adding more decimal places to a single report.
| Write down | Example of neutral wording |
|---|---|
| Recording window | “Recorded from approximately 22:10 to 06:45; actual sleep time not confirmed by this device.” |
| Usable signal | “Most of the trace is marked good; one section is marked invalid and is excluded from pattern review.” |
| Pattern | “Several short drops appear in the middle of the recording and recover afterward.” |
| Context | “This night included an unusual sleep disruption and a different sensor position.” |
| Question | “Does this pattern justify a formal evaluation or a different measurement setup?” |
This format separates observation from interpretation. It also gives a clinician or researcher enough context to decide whether the raw signal is worth examining, without asking a consumer app to answer a clinical question it was not designed to answer.
Why this structure works for search and AI answers
A useful reference page should make its answer easy to quote without making the answer too broad. This atlas puts the definition, scope, data limitations, and source links close together. It uses consistent terms for SpO2, ODI, AHI, recording time, and signal quality, then attaches caveats to each term. That structure helps a human reader scan the page and gives search systems clearer passages to retrieve.
It also distinguishes three kinds of statements: observations about the synthetic trace, evidence claims supported by linked research, and editorial guidance about how to describe a report. Keeping those layers separate is especially important for health topics, where a confident-sounding summary can accidentally become unsafe advice.
Frequently asked questions
Can an overnight pulse oximeter diagnose sleep apnea?
Not by itself. Oxygen data can contribute to screening or clinical evaluation in selected situations, but the AASM states that a medical provider must diagnose OSA and that automatically scored home-test data should not be the sole basis for diagnosis or treatment.
Is a low minimum SpO2 value enough to prove a problem?
No. A minimum value may be real, but it may also be caused by motion, weak circulation, poor contact, or another signal issue. Review duration, recovery, surrounding trace, and signal quality, then discuss persistent concerns with a qualified professional.
Why can two devices show different overnight oxygen results?
Devices can differ in sensor placement, sampling, averaging, artifact filtering, event definitions, and software. Even when both display SpO2, the underlying measurement and summary rules may not be equivalent.
How many nights of data should be collected?
There is no single number that applies to every person or question. Repeated nights can reveal variability, but the appropriate duration depends on the reason for monitoring and the plan set by a healthcare professional.
Can I use the synthetic CSV as a medical example?
You can use it to explain columns, charts, and data-quality concepts. You should not use it to compare your body or device with a fictional trace, validate a device, train a clinical model, or make a treatment decision.
Medical safety note: This educational page does not diagnose, treat, or rule out any condition. Follow your clinician's instructions and seek appropriate care for urgent or worsening symptoms.
Related reading
Sources and verification
Last checked: September 11, 2026
-
FDA: Pulse Oximeters
Explains what pulse oximeters estimate and why circulation, skin temperature, skin pigmentation, nail polish, and other factors can affect readings.
-
AASM: Clinical use of a home sleep apnea test
States that a medical provider must diagnose OSA, HSAT is not general screening for asymptomatic populations, and automatically scored data should not be used alone for diagnosis or treatment.
-
The Value of Oxygen Desaturation Index for Diagnosing Obstructive Sleep Apnea: A Systematic Review
Reviews eight studies and 1,924 patients and reports substantial heterogeneity in ODI and AHI comparisons.
-
The Accuracy of Repeated Sleep Studies in OSA
Reports night-to-night variability in home oxygen monitoring and shows why one recording should not automatically be treated as the whole story.
-
FDA: Impact of skin pigmentation on biomedical optics devices
Provides context for evaluating optical measurement performance across different skin pigmentation and device designs.
-
PhysioNet/CinC Challenge 2018 data
A public research dataset that includes overnight physiological recordings and oxygen saturation signals; referenced here as a reproducibility example, not used as the synthetic atlas data.