If a sensor gives nearly the same reading every time, does that make it right? Not necessarily. A cluster of repeated values tells us about precision, while closeness to the quantity we want to know concerns accuracy. The distinction is easy to see with a bathroom scale: if it is consistently offset, its readings can be repeatable and still misleading. For health measurements, both properties matter, but a stream of numbers alone does not establish either one.
The target and the cluster
Picture a target marked at ten units. A sensor that reports 9.8, 9.9, and 9.8 is precise in the everyday sense that its readings cluster closely. If the best-supported value is ten, those readings are also close to the target. Now imagine the sensor repeatedly reports 8.4, 8.5, and 8.4. That cluster remains tight, but it is displaced. Precision describes repeatability; accuracy concerns closeness to the quantity being measured.
This distinction helps prevent a common mistake: treating consistency as proof of correctness. A device can reproduce a bias, meaning a systematic offset, very reliably. Conversely, readings can vary around a reasonable central value because conditions or random effects differ. The two patterns call for different questions about the measurement.
What a sensor actually detects
A sensor is a tool that responds to an input from the physical environment. Biomedical sensors can detect biological, chemical, or physical processes and report or transmit data [2]. A thermometer translates a physical response to heat into a temperature value; a pulse oximeter uses changes in light absorption to estimate blood oxygen and heart rate [2].
The reported number is therefore not the quantity arriving untouched from the body. It passes through a chain: an input interacts with a sensing element, the instrument converts that response into a signal, and the system turns the signal into a displayed value. Understanding that chain helps explain why calibration, design, placement, and conditions of use can affect how a reading should be interpreted.
Repeatability is not the whole uncertainty
Measurement uncertainty describes the range of values that could reasonably be attributed to a measured quantity, given the available information. NIST’s reference on uncertainty explains approaches to expressing, evaluating, and combining uncertainty components [1]. A result with a small spread across repeated readings may still have uncertainty from calibration, environmental conditions, or how well the measurement method represents the intended quantity.
For instance, a wearable may track a changing signal consistently while the relationship between that signal and a health quantity remains imperfect. NIBIB describes wearables that track heart rate and sleep patterns, while noting that many advanced sensors are not practical for routine medical care [2]. A sensor’s ability to collect data is not by itself evidence that it diagnoses disease.
Ask what claim the number supports
When evaluating a measurement, ask what was measured, how the instrument converts the signal, and what reference supports the reported value. A reading might be useful for following a trend even if its absolute value is less certain. But that use depends on whether the device’s behavior and the decision context make the trend meaningful.
Precision and accuracy are connected, but neither is a synonym for confidence. Tight clustering says something about repeatability under the conditions observed. Accuracy requires a meaningful comparison with the quantity of interest. In health settings, the responsible interpretation also considers intended use and clinical evidence; a measurement can be technically impressive without independently settling a medical question.
Sources & further reading
- Uncertainty of Measurement Results NIST
- Sensors NIBIB
Prepared for the independent PSI Center reading library. About edition dates and source checking.
This explainer describes scientific principles. It does not assess individual treatment choices.
