Personalized-baseline pulse oximetry for early detection of altitude illness on expedition
At sea level, a pulse oximetry reading below 95% is a red flag. At 5,000 m, a reading of 85% might be entirely normal for one trekker and a sign of impending trouble for their tent-mate. Altitude physiology destroys the usefulness of a single universal cutoff — which is exactly why expedition medicine has moved toward tracking each climber against their own, individually-established baseline.
As barometric pressure falls with altitude, the partial pressure of inspired oxygen (PiO2) falls in parallel, even though the fractional concentration of oxygen in air stays at 21%. At the summit of Everest (8,849 m), barometric pressure is roughly a third of sea level, and PiO2 drops correspondingly — arterial oxygen saturation in acclimatized climbers at extreme altitude can sit in the 50s to 60s, a value that would represent a medical emergency at sea level.
This is the core reason a single SpO2 number cannot be interpreted the same way at every elevation: the entire oxygen cascade — from inspired air, to alveoli, to arterial blood, to tissue — is compressed downward as altitude increases. A "low" reading is expected and adaptive at high camp; the same number lower down would signal serious illness.
Even at a fixed altitude, healthy, well-acclimatized individuals show remarkably wide SpO2 variation — commonly cited ranges at ~5,000 m span roughly 80–92%, driven by differences in ventilatory drive, hemoglobin concentration, pulmonary vascular response, and prior acclimatization history. Someone with a naturally brisk hypoxic ventilatory response may sit comfortably at 90%, while a companion with an equally healthy but blunted response may run at 82% with no symptoms at all.
This spread means that any single "safe" threshold applied to a group will misclassify people in both directions: it will falsely alarm on climbers whose personal normal is simply lower, and it will miss a dangerous drop in a climber whose personal normal happens to be higher than the group cutoff.
A drop from a personal baseline of 90% down to 84% (a 6-point fall) is a meaningful deviation worth acting on — even though 84% might look "unremarkable" against a generic 80–90% altitude reference range that ignores who that particular number belongs to.
A useful baseline is not a single opportunistic reading — it is taken under standardized, reproducible conditions after the climber has had adequate rest at a new altitude (typically at least one overnight, once acute arrival effects have settled). Best practice measures resting SpO2 seated or supine, after several quiet minutes, on a warmed finger, away from exertion or recent caffeine.
That value is then logged as the reference point for that specific camp altitude. Because SpO2 at altitude keeps drifting downward for the first 24–72 hours of acclimatization before stabilizing, many expedition protocols re-baseline at each new camp rather than assuming yesterday's number still applies today.
A single SpO2 reading is a snapshot with limited diagnostic value; a trend is a story. Serial measurement — the same climber, checked in the same way, at similar times each day — turns pulse oximetry from a noisy point estimate into a signal that can reveal a genuine physiological change against the climber's own established pattern.
For a series of SpO2 readings to be comparable, the measurement conditions have to be held as constant as reasonably possible. That means: resting (not immediately after exertion, eating, or climbing into the tent), the same position (seated or lying, not standing and swaying), the same finger where feasible, and roughly the same time of day — since SpO2 can dip further during sleep due to periodic breathing at altitude.
Expedition doctors typically standardize on a morning reading (before the day's exertion) and sometimes an evening reading, logged in a simple table alongside heart rate and a brief symptom note. The discipline of the routine is what makes the resulting trend trustworthy.
Consumer and even clinical-grade pulse oximeters rely on differential light absorption through pulsatile arterial blood, and several factors specific to expedition conditions degrade that signal:
• Peripheral vasoconstriction from cold reduces pulsatile flow at the fingertip, weakening the signal and sometimes causing the device to under-read or fail to register at all • Motion artifact from shivering, walking, or an unsteady hand introduces noise that can swing readings by several points • Cold hands slow peripheral circulation, and rewarming the finger for 1–2 minutes before reading materially improves reliability • Most pulse oximeters are calibrated and validated against arterial blood gas data mostly in the 70–100% saturation range under normothermic, normoxic-adjacent lab conditions — accuracy claims (typically ±2%) become less certain as saturation falls well below that validated range at extreme altitude
None of these issues make oximetry useless — they make single readings unreliable and repeated, careful measurement essential.
A device with a quoted accuracy of ±2% at 90% saturation may drift further off at 75–80% saturation, the range many climbers reach above 7,000 m — one more reason a trend across many readings is trusted over any single number.
Practically, expedition medical logs record date, time, altitude/camp, resting SpO2, heart rate, and a short symptom note for each climber. Over days, this builds a personal time series that can be plotted against that climber's baseline band — visually separating ordinary day-to-day wobble from a sustained downward trend.
The goal at this stage is not yet judgment — it is simply building a clean, comparable dataset. Deviation detection, the next stage, depends entirely on the quality and consistency of this serial record.
Once a baseline exists and a comparable series of daily readings is flowing in, the meaningful question shifts from "is this SpO2 low?" to "is this SpO2 low for this person, at this altitude, compared to how they have been running?" That reframing is the practical core of individualized baseline monitoring.
The deviation-from-baseline approach reframes each new reading as a delta: today's SpO2 minus this climber's established personal reference at this altitude. A climber whose baseline is 91% dropping to 86% has fallen 5 points — a meaningful deviation. A different climber whose baseline is 84% reading 84% today has moved zero points, even though the absolute number is lower than the first climber's "abnormal" reading.
This is precisely why fixed-threshold alerting (e.g., "alert below 85%") both over- and under-triggers across a group: it ignores where each individual started.
Not every day-to-day fluctuation is meaningful. Sleep quality, hydration, recent exertion, cold fingers, and simple measurement noise can shift a reading by a couple of points without any underlying pathology. Because of this, most practical protocols do not act on a single reading that dips below threshold — they look for a sustained deviation across two or more consecutive checks, ideally corroborated by trend direction (worsening rather than an isolated blip).
The individual's own variability matters here too: a climber who has always shown a wide day-to-day spread needs a larger, more sustained deviation before it is trusted as real change, compared to a climber whose baseline has historically been rock-steady.
A meaningful SpO2 deviation is a useful early flag that something has changed physiologically — but it is not, by itself, a diagnosis. Falling saturation can reflect the earliest stages of acute mountain sickness (AMS), the more dangerous high-altitude pulmonary edema (HAPE), inadequate acclimatization pace, a concurrent respiratory infection, or simply poor technique on that particular reading.
Deviation detection's real job is to narrow attention: it tells the expedition team who to look at more closely and how urgently, not what is wrong. That is why the next stage pairs the oximetry trend with a structured clinical symptom score.
Pulse oximetry and clinical symptom scoring measure different things — one an objective physiological number, the other a structured self-report — and neither is reliable alone at altitude. Used together, an SpO2 deviation from personal baseline combined with a Lake Louise AMS score gives expedition medical staff a substantially clearer picture than either signal in isolation.
The Lake Louise Acute Mountain Sickness Score is a short self-report questionnaire covering headache, gastrointestinal symptoms, fatigue/weakness, and dizziness/lightheadedness, each rated 0–3 for severity, alongside a clinical assessment component in the full version. A total self-report score of 3 or more, including at least some headache, in the setting of a recent altitude gain, is generally used as the working definition of AMS in both field and research settings.
Unlike SpO2, the Lake Louise score captures the climber's subjective experience — and subjective experience is, in practice, what determines whether someone can safely continue climbing.
Published field studies at altitude consistently find only a weak-to-moderate correlation between SpO2 and AMS severity — some climbers develop significant symptoms with SpO2 close to their personal baseline, while others show a substantial deviation with minimal or no symptoms. This divergence has several plausible explanations: individual differences in cerebral sensitivity to hypoxia, the role of hypoxic ventilatory response variability, and the fact that AMS symptoms likely relate more directly to mild cerebral edema than to peripheral oxygen saturation per se.
Because of this imperfect overlap, using SpO2 deviation as a stand-alone diagnostic test for AMS would miss real cases and falsely flag others — the same problem, in a different form, as relying on a fixed population SpO2 threshold.
In one commonly cited altitude physiology teaching figure, roughly a third of climbers meeting Lake Louise criteria for AMS show SpO2 within a couple of points of their personal baseline — underscoring that oximetry is an adjunct signal, never the sole basis for a diagnosis.
Expedition medical protocols on major peaks typically use SpO2 deviation as a trigger to prompt a symptom check, and use the symptom score as the trigger to decide on action. A sustained deviation of several points from personal baseline, without any corresponding symptoms, usually warrants closer observation rather than immediate intervention. The same deviation accompanied by a rising Lake Louise score is treated with considerably more urgency.
This two-signal approach reduces both false alarms (an isolated SpO2 dip with a climber who otherwise feels completely well) and missed cases (a symptomatic climber whose oximetry happens to look unremarkable that day).
| Product | Indication | Trial Design | Key Result |
|---|---|---|---|
| Single Absolute SpO2 Threshold | Sensitivity: Low–moderate, biased by individual | Fixed cutoff (e.g. <85%) applied to every climber regardless of personal norm or altitude | False-alarm risk: high — flags naturally low-baseline climbers |
| Population Altitude-Adjusted Range | Sensitivity: Moderate, better than fixed cutoff | Expected SpO2 band derived from population data at each altitude band | False-alarm risk: moderate — still ignores individual variability |
| Individual Baseline Trending | Sensitivity: Moderate–high for that individual | Each climber compared only against their own established resting reference | False-alarm risk: low — but requires disciplined serial measurement |
| Symptom Score Alone (Lake Louise) | Sensitivity: Moderate, subjective and self-reported | Structured questionnaire without any physiological measurement | False-alarm risk: moderate — depends on honest self-report, can be masked |
Detecting a deviation is only useful if it connects to a clear, pre-agreed action. Expedition medicine on major peaks generally works from an explicit protocol: a defined SpO2 deviation combined with a defined symptom threshold triggers a specific, non-negotiable response — closer monitoring, halting further ascent, or descent — removing ambiguity at the exact moment judgment is most impaired by hypoxia.
The most effective protocols set the alert threshold before the expedition begins, not in the moment — because judgment about one's own condition deteriorates precisely as hypoxia and fatigue increase. A common structure: a deviation of 3–4 percentage points or more below personal baseline, sustained across at least two checks, is treated as a monitoring trigger; the same deviation combined with a Lake Louise score of 3 or more is treated as an ascent-stop or descent trigger.
Setting numbers in advance also protects against normalization of deviance — the gradual, well-documented tendency for a team under summit pressure to keep reinterpreting worsening data as "probably fine."
Rather than a single binary alarm, well-run expeditions typically use a tiered response:
• Normal: deviation under threshold, no symptoms — continue the planned itinerary with routine rechecks • Monitor: deviation at or above threshold, or mild symptoms — no further ascent that day, recheck SpO2 and symptoms every 4–6 hours, ensure hydration and rest • Alert / Descend: deviation at or above threshold combined with a Lake Louise score of 3 or higher, or any sign of ataxia, altered mental status, or breathlessness at rest — immediate descent, with supplemental oxygen and medication (e.g., acetazolamide, dexamethasone) as appropriate per the expedition's medical protocol
This tiering avoids two failure modes: under-reacting to a genuine early warning, and over-reacting to ordinary day-to-day noise in a way that erodes trust in the monitoring system.
The single most cited rule in expedition medicine remains: never ascend to sleep at a higher altitude while experiencing symptoms of AMS — an SpO2 deviation with a rising symptom score is exactly the pattern that rule exists to catch.
Individual-baseline trending is more work than reading a single number off a fixed chart — it requires establishing and logging a personal reference at every new camp, maintaining a disciplined serial measurement routine, and interpreting deviation in the context of that individual's known variability. On multi-week 8,000 m expeditions, medical teams generally judge that overhead worthwhile: it meaningfully reduces both missed early warnings and disruptive false alarms compared to a fixed threshold applied uniformly across a diverse group of climbers.
Used as one input among several — symptoms, mental status, exertional tolerance, and clinical judgment — a personalized SpO2 deviation trend becomes one of the more actionable early-warning tools available to a team without advanced diagnostic equipment at altitude.