Why Patch has two temperature sensors

Patch carries two temperature sensors. One measures the temperature of the skin it sits on. The other measures the temperature of the air around the patch. The first is the one people expect. The second is the one that makes the first mean something.

The trouble with skin on its own

Skin temperature is not the temperature of the body. It is the temperature of a surface that sits between a warm body and whatever is outside it, and it is pulled in both directions. Step from a warm room into a cold one and the skin on your arm cools within minutes, with nothing having changed inside you. Pull a sleeve over the patch and the reading climbs. Sit in the sun, lean on a cold wall, sleep under a blanket, go for a walk: every one of these moves the skin reading, and none of them says anything about you.

A skin sensor on its own cannot tell those apart from a change that comes from the body. It reports a number, and the number is honest about the skin and silent about why. Taken alone, a skin reading that drifts over an afternoon is a question with no answer.

What the second sensor adds

The air sensor is the missing half of the story. It tells us what the skin was exposed to. With both readings we can look at the relationship between them rather than at either one alone. If the skin cools and the air cooled first, the change most likely came from outside. If the skin warms while the air stays where it was, that is a different kind of change. The gap between the two readings, and the way that gap moves over time, carries information that neither reading has by itself.

The same logic matters over longer stretches. Patch is designed to learn your baseline over days and weeks, and we need that baseline to describe you rather than your surroundings. Without the air reading, a cold week would look like a shift in you. With it, the two can be kept apart, and the baseline stays yours.

Two sensors, opposite jobs

The two sensors want opposite things. The skin sensor wants the best contact it can get, with as little as possible between it and the body. The air sensor wants the reverse: as little influence from the body as possible, so that it sees the air and not the warmth coming through the patch. On something as small and as thin as Patch, those two goals pull against each other. The whole device is warmed by the arm it sits on, and there is not much of it that is far from skin. Where the air sensor goes, and what sits between it and the skin, is a design problem in its own right, and one of the reasons the shape of the patch matters as much as the circuit on it.

Measure it, don’t model it

We could have tried to correct a single skin reading with a model of what the environment was probably doing. We chose to measure the environment instead. A measured value beats a guessed one, and a second sensor is a small addition for what it gives back: a skin reading that can be read in context, and a baseline that belongs to the person wearing the patch rather than to the weather.