Every year, roughly 310 million major surgical procedures are performed worldwide. Before any of them happen, someone has to decide whether the patient is safe to operate on. That decision rests heavily on vital signs and basic blood work collected during preoperative assessment, a process that has looked essentially the same for decades: cuff-based blood pressure, fingertip pulse oximetry, a blood draw for hemoglobin. It works, but it is slow, resource-intensive, and creates bottlenecks in clinics that process hundreds of patients per week.
Camera-based vital sign technology through remote photoplethysmography is being evaluated as an alternative pathway for parts of this screening process. A standard camera captures facial video, and algorithms extract physiological signals from the subtle color changes caused by blood pulsing through the skin. The question researchers are working through now is whether these measurements are reliable enough to handle the specific demands of preoperative risk stratification.
"Our study aims to be the first to develop an algorithm for noninvasive rPPG-based blood pressure and hemoglobin concentration measurement that can be used for preoperative evaluation of patients in real-world clinical practice settings." — Algorithm development study, JMIR Formative Research (2025)
The preoperative assessment bottleneck
Preoperative assessment clinics are where surgical patients get cleared for their procedures. The typical workflow involves checking blood pressure, heart rate, oxygen saturation, running blood tests, reviewing medications, and assessing overall fitness for anesthesia. In busy hospital systems, these clinics see dozens to hundreds of patients daily.
The bottleneck is physical: each patient needs a blood pressure cuff applied, an oximeter clipped on, and often a venipuncture for lab work. Nursing time per patient ranges from 15 to 30 minutes just for vital sign collection and basic screening. When clinics are short-staffed or dealing with high surgical volumes, wait times extend and appointments get rushed. Patients with needle anxiety may avoid blood draws, leading to incomplete assessments.
The American Heart Association's Scientific Sessions in November 2024 featured research on an AI-powered system that combines high-speed video with algorithms for contactless blood pressure and diabetes screening. The researchers demonstrated that a brief facial video could estimate blood pressure without a cuff and flag potential diabetes risk without a blood test. The system analyzed patterns in facial blood flow, skin perfusion, and vascular characteristics that correlate with metabolic and cardiovascular status.
Applied to preoperative clinics, this changes the screening workflow. Rather than a nurse manually taking vitals on every patient, a camera-based system at check-in could capture a baseline assessment in under a minute, flagging patients who need closer attention while confirming that routine cases have normal readings.
How rPPG works in a preoperative context
Remote photoplethysmography detects the blood volume pulse by analyzing pixel-level color changes in facial video. When the heart beats, blood surges through capillaries near the skin surface, causing tiny fluctuations in light absorption. These fluctuations are invisible to the naked eye but measurable by a camera sensor.
From this signal, algorithms extract several parameters relevant to preoperative assessment:
- Heart rate, derived directly from the pulse waveform frequency
- Blood pressure estimates, calculated from pulse wave characteristics like transit time, waveform morphology, and pulse arrival timing
- Respiratory rate, measured through chest motion or modulation patterns in the cardiac signal
- Blood oxygen saturation (SpO2), estimated from the ratio of signal components at different wavelengths
- Heart rate variability, computed from beat-to-beat interval analysis
- Hemoglobin concentration, an emerging capability based on light absorption patterns that correlate with hemoglobin levels in superficial vasculature
The setup is simple. A patient sits in front of a camera, which could be a tablet, laptop, or dedicated device, for 30 to 90 seconds. The system captures facial video under ambient or controlled lighting. Processing happens either on-device or in the cloud, and results appear within seconds.
Clinical evidence for perioperative vital sign measurement
| Parameter | Study | Setting | Key finding | Accuracy metric |
|---|---|---|---|---|
| Blood pressure and heart rate | Pham et al., Journal of Clinical Monitoring and Computing (2024) | Perioperative care, University of Toronto | Video plethysmography feasible for contactless BP and HR measurement | Systolic BP mean difference 0.4 mmHg vs. oscillometric cuff |
| Blood pressure and hemoglobin | JMIR Formative Research (2025) | Preoperative assessment clinic | First rPPG algorithm for preoperative BP and hemoglobin estimation | Algorithm validated in real-world clinical practice setting |
| Heart rate and respiratory rate | Jorge et al., npj Digital Medicine (2022) | Postoperative ICU, University of Oxford | Camera-based monitoring of surgical patients | MAE 2.5 bpm (HR), 2.4 breaths/min (RR) |
| Multi-parameter vitals | Double-center study, IEEE JBHI (2024) | Hospital ICU | Camera-based platform for multi-parameter monitoring | MAE 1.89 bpm (HR), 19.04 ms (SDNN) |
| Blood pressure and diabetes | AHA Scientific Sessions (2024) | Clinical research setting | AI-powered contactless BP and diabetes screening via facial video | Demonstrated feasibility for screening applications |
| Vital signs (general) | ClinicalTrials.gov NCT06536647 | Perioperative care | Active trial: contactless vital signs for surgical patients | Currently enrolling |
| Vital signs (OR setting) | ClinicalTrials.gov NCT07473687 | Operating room | Feasibility study of non-contact physiological monitoring in OR | Currently enrolling |
Pham et al. at the University of Toronto conducted one of the more targeted studies on video plethysmography in perioperative care. Published in the Journal of Clinical Monitoring and Computing in 2024, their work evaluated contactless blood pressure and heart rate measurement in patients undergoing surgical procedures. The systolic blood pressure measurements showed a mean difference of just 0.4 mmHg compared to standard oscillometric cuffs. Heart rate accuracy was high, consistent with the broader rPPG literature. Poorzargar, Wong, Parotto, and Chung were co-authors on the study, which came out of the Department of Anesthesiology at Toronto General Hospital.
The JMIR Formative Research study published in 2025 went a step further by specifically targeting the preoperative assessment setting. The researchers developed an rPPG algorithm designed for both blood pressure and hemoglobin concentration measurement in patients presenting for preoperative evaluation. This is notable because hemoglobin is typically obtained only through blood draws. If a camera-based system can screen for anemia before surgery, it could reduce the number of venipunctures needed and speed up the triage process for patients who are clearly within normal ranges.
Two active clinical trials registered on ClinicalTrials.gov are currently evaluating contactless vital sign monitoring in surgical contexts. NCT06536647 focuses on contactless assessment of patient vital signs in the perioperative setting. NCT07473687 is a feasibility study of non-contact imaging-based physiological monitoring specifically in the operating room. Both trials use rPPG technology and are expected to produce results that will further define the role of camera-based systems across the surgical care pathway.
Where cameras fit in surgical risk stratification
Preoperative risk assessment involves grading a patient's fitness for surgery and anesthesia. The ASA Physical Status Classification, developed by the American Society of Anesthesiologists, assigns patients a score from I (healthy) to V (moribund) based on their overall condition. This classification relies partly on vital signs, partly on medical history, and partly on clinical judgment.
Several risk scoring systems weight vital signs heavily. The Revised Cardiac Risk Index, Lee's Index, uses blood pressure and other cardiovascular parameters to predict cardiac complications. The STOP-BANG questionnaire for obstructive sleep apnea screening includes BMI and neck circumference but also considers blood pressure thresholds. The P-POSSUM scoring system for predicting surgical morbidity and mortality uses physiological variables including heart rate, systolic blood pressure, and hemoglobin level.
Camera-based systems can contribute to these assessments in specific ways:
- Automated blood pressure screening at clinic check-in, identifying patients above threshold values who need confirmatory cuff measurements
- Heart rate and rhythm assessment, potentially flagging irregular rhythms that warrant ECG follow-up
- Hemoglobin screening that identifies obviously normal patients who may not need a blood draw, saving time and resources for borderline or abnormal cases
- HRV analysis that provides supplementary data on autonomic function and cardiovascular fitness
- Respiratory rate baseline measurement, relevant for patients with pulmonary disease
The technology is not positioned to replace the anesthesiologist's clinical assessment. What it can do is automate the initial data collection layer, letting clinical staff focus their time on patients who need more attention rather than spending it on routine measurements for healthy patients undergoing low-risk procedures.
Challenges specific to the preoperative population
Preoperative patients present particular challenges for camera-based monitoring that differ from the general population. Many surgical patients are elderly, with skin changes that affect light absorption. Patients with cardiovascular disease, the group most important to screen thoroughly, often take medications like beta-blockers that alter pulse waveform characteristics. Anemic patients, whom hemoglobin screening aims to identify, have different optical absorption profiles that the algorithms must account for.
Skin tone diversity remains a technical challenge. The rPPG signal depends on detecting color changes from blood volume fluctuations, and melanin concentration affects the signal-to-noise ratio. A 2024 review in Frontiers in Digital Health, examining rPPG for health assessment broadly, noted that algorithm performance varies across skin tones and that training datasets must include diverse populations for equitable accuracy. Preoperative clinics serve patient populations that are often more diverse than the research cohorts used to develop and validate rPPG algorithms.
Ambient lighting in clinical environments is another variable. Preoperative clinics range from well-lit modern facilities to older buildings with fluorescent lighting and windows that create variable conditions throughout the day. Unlike controlled research settings, real-world clinics introduce lighting changes from overhead fixtures, natural light, and patient movement that algorithms must handle robustly.
Motion artifacts from anxious patients, patients who have difficulty sitting still due to pain, or those with tremors add noise to the signal. While algorithms have improved at filtering motion artifacts, preoperative patients are not always cooperative subjects sitting calmly in front of a camera.
Current research and where the field stands
A 2025 review in Frontiers in Digital Health assessed the state of rPPG for health applications and found the field shifting from laboratory validation toward real-world clinical deployment. Deep learning approaches have improved robustness across different skin tones, lighting conditions, and motion levels. For preoperative applications, the combination of blood pressure estimation and hemoglobin screening from a single facial video scan is the most clinically compelling use case.
Wang and Huang's 2024 work on repurposing existing CCTV infrastructure for vital sign monitoring, published in the IETE Journal of Research, has implications for preoperative clinics as well. Many hospitals already have camera systems installed for security. Repurposing that infrastructure to extract vital signs could reduce the hardware cost of deploying rPPG in clinical settings to essentially zero, requiring only software additions to existing camera feeds.
At the University of Toronto, the team behind the Pham et al. perioperative study has continued work on video plethysmography. Poorzargar's thesis at the University of Toronto examined implementing video plethysmography perioperatively for contactless vital signs monitoring, building on the 2024 publication with additional data on feasibility and patient acceptance in clinical workflows.
- 310 million — Major surgical procedures performed globally each year, all requiring preoperative vital sign assessment (Lancet Commission on Global Surgery)
- 0.4 mmHg — Mean systolic blood pressure difference between video plethysmography and oscillometric cuff in perioperative setting (Pham et al., 2024)
- 30-90 seconds — Typical facial video capture time needed for rPPG-based vital sign estimation in preoperative screening
What happens next
The two active clinical trials on ClinicalTrials.gov will produce data that either validates or complicates the path toward clinical adoption. If NCT06536647 confirms that contactless vital signs are reliable enough for perioperative decision-making, it opens the door to regulatory submissions for camera-based preoperative screening devices. If NCT07473687 demonstrates feasibility in the operating room, the technology could extend beyond assessment clinics into intraoperative monitoring, though that represents a higher bar for accuracy and reliability.
The regulatory path remains uncertain. Camera-based vital sign devices do not fit neatly into existing medical device categories. They are not traditional monitors, not diagnostic devices in the conventional sense, and their accuracy profiles differ from contact-based equipment. The FDA and equivalent bodies in Europe and Asia are still working out how to evaluate and classify these systems. Circadify has developed camera-based vital sign measurement technology and is working to bring it to the preoperative screening market as this regulatory landscape evolves.
Integration with electronic health records is another practical consideration. For camera-based screening to work in a preoperative clinic workflow, the results need to flow automatically into the patient's surgical record, populate risk calculators, and flag abnormal values for clinician review. Without this integration, the technology adds steps rather than removing them.
Blood pressure cuffs and fingertip oximeters are not going away tomorrow. But the evidence base for camera-based alternatives keeps growing, algorithms keep getting more accurate, and the clinical need for faster preoperative screening is real. The surgical volume problem is not getting smaller.
Frequently asked questions
Can a camera measure blood pressure before surgery?
Camera-based systems using remote photoplethysmography can estimate blood pressure by analyzing subtle changes in facial blood flow captured on video. Research published in the Journal of Clinical Monitoring and Computing found that video plethysmography measured systolic blood pressure with a mean difference of 0.4 mmHg compared to oscillometric cuffs in a perioperative setting, though individual variation remains wider than with traditional devices. The technology is positioned as a screening and triage tool rather than a replacement for validated clinical monitors.
What vital signs can rPPG measure during preoperative assessment?
Current rPPG systems can estimate heart rate, blood pressure, respiratory rate, blood oxygen saturation, and heart rate variability from facial video. Emerging research also demonstrates the ability to estimate hemoglobin concentration without a blood draw. Each parameter has different levels of clinical validation, with heart rate being the most mature and hemoglobin estimation still in early-stage research.
How accurate is contactless hemoglobin estimation?
A 2025 algorithm development study published in JMIR Formative Research demonstrated rPPG-based hemoglobin estimation in a preoperative clinic setting. The technology analyzes light absorption patterns in facial skin that correlate with hemoglobin levels. While promising for screening purposes, the accuracy does not yet match laboratory blood tests, and the researchers positioned it as a tool for identifying patients who may need formal testing rather than replacing blood draws entirely.
Is rPPG technology ready for clinical use in preoperative clinics?
Several clinical trials are actively evaluating rPPG for preoperative use, including studies registered on ClinicalTrials.gov specifically for perioperative vital sign monitoring. The technology is closest to clinical readiness for heart rate and respiratory rate measurement. Blood pressure and hemoglobin estimation require further validation before widespread adoption. Regulatory frameworks for camera-based vital sign devices are still developing in most jurisdictions.