How much does skin temperature change around an illness?

Overnight skin temperature rose an average of 0.28 °C above each member's own baseline on the night leading into a logged sick day (likely between 0.24 and 0.33 °C). That result covers 764 members and 856 logged sick days, with each member counted equally. The night of the sick day was still 0.22 °C above baseline. By the second night after the sick day, the likely range of the average included no change.

How much does skin temperature rise when you get sick?

Overnight skin temperature rose by a fraction of a degree around a logged sick day. Most of the rise came on the night leading into that day and the night of the sick day. Each night was compared with the member's own baseline, defined below.

A logged sick day was a calendar day the member tagged as Sick, Sick Day, Cold/Flu or Illness by hand in the app. The tag had to be saved before the app first recorded the following night's sleep. Alerts the app raises from temperature or heart readings were not used as sick days. Only the first logged day of an illness counted: a day with another of those four tags in the 28 days before it was left out, and counted days from the same person were at least 43 days apart. Skin temperature was the ring's nightly temperature deviation, the difference in degrees Celsius between that night's skin temperature and the member's longer-term level as the app calculates it.

The personal baseline was the member's average of that nightly deviation over 14 nights that ended two weeks before the logged day, from at least 10 valid nights recorded on the same ring and firmware version. Ending it two weeks early keeps the nights just before the logged day out of the baseline, so a change in those nights shows up. Each night from 14 before the logged day to 14 after it was compared with that baseline. Night zero was the sleep that ended on the morning of the logged day, the night leading into the sick day. Night one was the night of the sick day.

Each person counted equally, however many illnesses they logged; the uncertainty ranges resampled whole people.

The chart shows skin temperature near baseline through the early nights, then climbing to a peak on the night leading into the sick day. When illnesses the app had already flagged with a Fever or Immune Load alert were left out, the peak rise was smaller. Part of the average rise therefore comes from illnesses the app had already flagged.

Other wearables have been tested as fever detectors. A model using a smart ring's skin temperature, heart rate and sleep data from the nights around a self-reported fever picked out some of those fever days [1]. In hospital patients with blood cancers, a wearable temperature device agreed well with a standard thermometer and often caught fever earlier [2]. This analysis did not test whether the Ultrahuman Ring detects fever.

How long does skin temperature stay raised after a sick day?

The rise in skin temperature lasted only a few nights after a logged sick day. The night after the sick day was the last night with a clear rise, at 0.05 °C above baseline (likely between 0.01 and 0.08 °C). After that, the averages sat near baseline before dipping slightly below it.

Why does skin temperature dip below normal after an illness?

Skin temperature read slightly below baseline on later nights after a logged sick day, which fits how the ring's nightly reading is built. The lowest average came on night 10 after the sick day. The ring measures each night against the member's rolling level, an average of their recent ordinary nights. Smaller rises during an illness enter that level. After recovery, the reading can sit below zero until those nights leave the average.

The dip appeared in every stricter version of the analysis. Among members with a reading on every night of the week after the logged day, skin temperature had already dipped below baseline by the end of that week. Missing readings therefore do not explain the dip over that week.

What this analysis can't tell you

Members chose when to log a sick day, and that choice shapes these numbers. The app raises an automatic Fever alert from this same temperature reading, and some members may have logged a sick day after seeing it. That would favor illnesses with a visible rise. Members also do not tag every illness, so unlogged sick days may look different. A tag records the day and not always the hour, so the peak night may sit slightly off the start of the illness. Because the app's baseline slowly takes in warmer nights, later nights of a longer illness may look less warm than they were. Travel across time zones was not accounted for. The analysis cannot say what caused an illness or when it ended.

Frequently asked questions

Does a warm night on my ring mean I have a fever?

A warm night on the Ultrahuman Ring does not show a fever on its own. This analysis compared skin temperature with each member's own baseline and did not check it against a thermometer. To check for a fever, use a thermometer.

Does skin temperature rise before a logged sick day?

Yes: in this group, skin temperature peaked on the night leading into the logged sick day and was slightly above baseline on the nights before that. Most of that earlier, smaller rise came from illnesses the app had already flagged with an automatic alert. The analysis lines nights up against the day members logged, so it cannot say when they first felt ill.

Who does this finding not apply to?

This finding describes adults who wear the Ultrahuman Ring and logged a sick day by hand in the app. It says nothing about children or about illnesses people did not log. Each member counted equally, so the averages describe a typical member's sick day, and one person's nights can differ from it.

References