What does a recorded heartbeat look like between the dots on a monitor? A digital sensor does not continuously store every instant. It takes measurements at selected times, then software draws a trace through them. The spacing of those measurements is the sampling rate. If changes happen faster than the snapshots can track, the displayed waveform may flatten, distort, or even suggest a false rhythm.
A signal becomes a sequence
A sensor detects an input from the physical environment and turns it into data. Depending on the device, that input might be pressure, temperature, light, electric current, or a chemical signal. A biological sensor therefore does not record “the body” all at once; it reports a particular measurable quantity, often after converting it into a voltage or another signal that can be digitized. [1]
Sampling means measuring the signal repeatedly at discrete intervals. If a monitor takes 100 samples each second, its sampling rate is 100 hertz. The stored record is a list of values with timestamps, not a perfectly continuous curve. A line drawn between values is a visual reconstruction, and its smoothness should not be confused with new information about what happened between samples.
Fast motion can masquerade as slow motion
Consider a wheel turning under a camera. If each frame catches the spoke at a similar position, the wheel may appear to turn slowly or backward even while it spins quickly. A waveform can suffer the same ambiguity: when samples arrive too slowly relative to the signal’s oscillations, a high-frequency pattern can be misread as a lower-frequency one. This distortion is called aliasing.
For a clean sinusoidal signal, the Nyquist principle says the sampling rate must be greater than twice the highest frequency one wants to distinguish. This is a minimum condition, not a universal guarantee of faithful recording. Real biomedical signals contain mixtures of frequencies and noise, and practical systems often filter unwanted high-frequency components before sampling. If important content is removed or already aliased, later plotting cannot restore it.
The display is not the measurement
Imagine a sharp electrical spike occurring between two sample times. A low-rate device may miss it entirely; a slightly different sampling schedule may catch one point and draw a misleading triangular bump. Either display can look plausible. That is why waveform shape depends on acquisition settings as well as physiology, and why comparing traces requires knowing their sampling and filtering conditions.
Sensors can measure quite different things even when their output is shown as a line. A pulse oximeter uses changes in absorption of selected light wavelengths to estimate oxygen in blood and heart rate; a continuous glucose monitor detects glucose in fluid between cells. These examples show that a readout is the product of a sensing mechanism and a signal-processing chain, not a direct transparent window into a biological variable. [1]
More samples are not automatically better
Raising the sampling rate can preserve faster changes, but it also creates more data and can capture more noise. The useful rate depends on the fastest feature relevant to the question and the sensor’s own bandwidth, meaning the range of signal frequencies it can respond to. If a sensor cannot physically detect a rapid fluctuation, storing many samples of its sluggish output does not recover that fluctuation.
Computational models can help researchers explore how measurement choices affect a simulated system, but such simulations remain dependent on assumptions and inputs. Models are used to study complex systems under different conditions; they can help select experiments or interpret patterns, yet they do not turn an undersampled real-world signal into a fully observed one. [2] A waveform is therefore best read with two questions in mind: what physical quantity was sensed, and how often was it actually measured?
Sources & further reading
- Sensors NIBIB
- Computational Modeling 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.
