Model Details

Basic info:

Presage vitals by video analysis generates pulse respiratory quotient (PRQ), or the ratio of the number of heartbeats per breathing cycle. An API has also been developed to allow users easy access to compute this metric for commercial and scientific applications.

Organization developing model: Presage Technologies

Model date: 20250911t173426

Model version: 1.6.0

Model type:A deterministic computer vision model with two primary stages. The first identifies and tracks key feature points on the subject’s face and chest to compute pulse rate and breath rate, respectively. The second stage uses signal processing to analyze the temporal fluctuations of these features to compute pulse rate and breath rate, from which PRQ is computed. PRQ is computed as the ratio of heart rate to breath rate evaluated over the same temporal window.

License: The algorithm is currently proprietary, and licenses are granted with predefined agreement.

Where to send questions: Questions can be sent to: support@presagetech.com

Intended Use

Model uses:

This PRQ model was intended for use by qualified clinicians, polygraphers and researchers for the analysis and non-diagnostic utility of PRQ quantification and derivative metrics. It was intended to be used with a video from a stationary device (such as a handheld, mobile or laptop camera), that contains the subject’s face in view, and be of 30 consecutive seconds in length and acquired at a minimum of 25 frames per second. The user’s face must be visible and unobstructed within the video, and the user must not make sudden large motions or move their face beyond 90 degrees of optical axis during this time. It is only intended to measure on participants with pulse rate values in the range of 40-180 bpm and breath rate values in the range of 4-31bpm.

Out-of-scope uses:

The Presage PRQ model is not intended for diagnostic purposes. Do not self-diagnose or self-medicate on the basis of the measurements. No alarms are provided. It is currently not intended for use in highly dynamic environments, or with a highly moving camera. We ensure all users have acknowledged and agreed to our license agreement and terms of service for usage prior to use.

Factors

The PRQ model first requires Mediapipe’s face detection algorithm to identify the face and 468 key feature points on the face (mesh model). Thus, if these features are not identifiable by Mediapipe’s algorithm, then PRQ will not be calculable. Reference the Mediapipe model cards can be found here: Full Range Face detection model card.

User: User’s face orientation must be within angular parameters for detection by Mediapipe, off axis view decreases model’s detection abilities. User’s face and chest must be sufficiently visible, illuminated and positioned within the imaging window. User’s face and body must be clear, and not blurry due to excessive motion or camera system. Users are instructed to maintain a fixed position.
Environment: Model is developed on videos with various lighting and motion conditions. However, its quality can degrade in extreme conditions.

These factors can affect model performance:

Lighting: Sufficient illumination as well as temporal changes in lighting can affect the ability to correctly measure the plethysmogram and thus PRQ.
User: User demographics, including skin tone and facial hair, may impact return rate. Darker skin tone will require a higher threshold on illumination intensity to produce accurate results, which if not met may reduce return rate. Facial hair or other facial obstructions will reduce the area of analysis, which may affect return rate.
Motion: High degrees of user motion are not tolerated by the PRQ algorithm. It is recommended to users that both the camera be mounted and the user maintains a fixed position before acquiring measurement.

Other factors:

Instrumentation: Several cameras, including mobile (Android), USB (Econ) and webcam (Logi) based devices with varying sensitivities and image resolutions have been tested for efficacy. The biggest factor with respect to instrumentation is the importance of locking camera parameters during acquisition, including exposure, gain and white balance, while also operating on the uncompressed video stream. The SmartSpectra SDK accounts for these factors, as well as variance in image size, resolution, and aspect ratio. When evaluating on pre-recorded videos, the biggest factor to consider is the codec, which can incorporate intraframe or interframe compression, both of which may impact accuracy.

Metrics

RMSD of point estimates of PRQ: root mean square deviation between all PRQ values measured within a 30 second period and ground truth measurements evaluated over the same window.
MAE of point estimate and 95% CI of PRQ: median absolute error or deviation of all measures of PRQ as compared to ground truth.
Mean proportion of returned values of PRQ: for every 30 seconds of video, a single weighted measurement can be measured. Of all possible sets of 30 second clips within a video, this is a measure of the proportion of them that returned a valid measurement of PRQ.

Evaluation Data

The evaluation data consists of a set of 225 videos. Corresponding quantities of PRQ were measured from a Biopac research system which quantified heart rate from 3 lead ECG and breath rate from a strain gauge sensor centered on the chest. A clip of 30s from each video was run through the Presage PRQ model for evaluation, leading to a total number of 749 samples. Videos were acquired on users covering a range of demographic variability, including age, gender and Fitzpatrick scale.

Quantitative Data

RMSD of point estimate of PRQ: 1.11
MAE of point estimate and 95% CI of PRQ: 0.61 with 95% CI of [0.53, 0.69]
Mean proportion of returned values of PRQ and 95% CI: 0.26

Distribution of error figures:

Skin Tone (Fitzpatrick)	% of Dataset (num samples)	RMSD	MAE [95% CI]	Mean Return Rate
1	0.18 (135)	1.09	0.78 [0.59, 0.96]	0.37
2	0.14 (104)	1.13	0.65 [0.44, 0.86]	0.43
3	0.09 (68)	0.56	0.34 [0.21, 0.48]	0.19
4	0.16 (119)	0.53	0.38 [0.28, 0.47]	0.29
5	0.11 (84)	0.77	0.52 [0.38, 0.66]	0.29
6	0.14 (109)	0.50	0.30 [0.22, 0.38]	0.06

Sex	% of Dataset (num samples)	RMSD	MAE [95% CI]	Mean Return Rate
M	0.36 (270)	1.04	0.64 [0.52, 0.76]	0.30
F	0.46 (349)	0.80	0.53 [0.45, 0.61]	0.27

Camera Type	% of Dataset (num samples)	RMSD	MAE [95% CI]	Mean Return Rate
Android	0.49 (366)	1.00	0.62 [0.52, 0.72]	0.25
Econ	0.51 (387)	1.19	0.61 [0.49, 0.72]	0.26

Ethical Considerations

As a remote sensing device, the risks posed to the subjects in the trial are minimal, including the association of each subject with corresponding biometric data. Mitigation of these risks include de-identifying all subject data, including videos, prior to saving it. Additionally, all data is securely stored in a HIPPA compliant database with access to a select number of trained researchers.

The model is not intended for human life-critical decisions, diagnostics or prognostication.

Presage PRQ Metrics for Spot Mode Model Card