PulsePrint

Inspiration

Most smartphone heart-monitoring tools focus on a single measurement: "What is your heart rate right now?" But the same reading can mean different things for different people.

PulsePrint explores a different question: what if the phone learned what is normal for you, then detected meaningful changes from your own baseline?

What it does

PulsePrint uses a smartphone's rear camera and flash to capture a fingertip photoplethysmography (PPG) signal.

The working prototype captures the optical signal, displays the resulting PPG waveform, evaluates signal quality, and rejects measurements it does not consider reliable enough to interpret.

Repeated valid measurements are designed to establish a personal physiological baseline using features such as heart rate, beat-to-beat intervals, variability and waveform characteristics.

New measurements can then be compared with that individual's baseline. PulsePrint does not diagnose disease. Its goal is to identify persistent deviations that may justify repeating a measurement or seeking clinical assessment.

How it works

Finger on camera + flash → PPG waveform → signal-quality check → feature extraction → personal baseline → deviation score → safe response.

The current prototype performs camera capture and signal processing directly in the browser. Poor-quality recordings caused by motion, pressure, saturation, weak pulse variation or other interference are rejected rather than interpreted.

No video is uploaded or stored.

Scientific approach

The core research question is whether repeated real-world smartphone PPG can reliably support within-person deviation detection while abstaining when signal quality is insufficient.

Initial real-device testing has reinforced an important part of that question: good camera coverage does not necessarily mean the physiological signal is reliable enough to interpret. PulsePrint therefore evaluates signal quality before accepting a measurement rather than forcing a result.

Validation will compare measurements with reference devices and evaluate repeatability, false-alert rates, anomaly detection and signal acceptance/rejection rates across different skin-tone groups, smartphone cameras and measurement conditions.

Challenges

Smartphone cameras vary considerably, and optical physiological measurements can be affected by movement, finger pressure, temperature, device characteristics, camera processing and skin pigmentation.

Our initial prototype testing has shown why signal-quality control matters: a recording can appear usable while still being too noisy or irregular for reliable pulse estimation.

PulsePrint therefore treats abstention as a feature. When signal confidence is insufficient, the safer response is to ask the user to measure again rather than provide an unreliable result.

What's next

With the initial camera-based PPG prototype now built and deployed, the next stage is calibration and validation.

We plan to collect repeated real-world measurements, improve signal processing across different smartphone cameras and conditions, establish reliable personal baselines, and compare PulsePrint measurements with reference devices.

These experiments will determine whether within-person smartphone PPG deviation monitoring is sufficiently reliable before any clinical use is considered.