Community
Explore the clinical research behind SmartSpectra and connect with developers building contactless applications.
Connect
Developer Updates
SDK releases, API changes, and best practices delivered to your inbox.
Discord
Ask questions, share projects, and connect with other developers building contactless vital sign applications with SmartSpectra.
Join the DiscordGitHub
Explore open-source SDK samples, integration guides, and developer tools for building rPPG and contactless health monitoring applications.
Follow on GitHubResources
Validation and information on the continuous pulse rate algorithm. Extracts pulse rate from video using Presage's proprietary computer vision pipeline analyzing facial blood flow patterns through consumer-grade cameras.
119 repositories
Your AI-powered safety mirror: Combining Gemini’s medical intelligence with live biometric sensing to keep you safe from high-risk substance interactions.
View on GitHub135 items
Your AI-powered safety mirror: Combining Gemini’s medical intelligence with live biometric sensing to keep you safe from high-risk substance interactions.
View ProjectClinical Evidence
Amit S. Padaki1*; Alexa L. Zarzour2; Kelly R. Keene1; Carlo A. Canepa1; Dana R. Levin3; Erik L. Antonsen3
Peer-reviewed study published in Frontiers in Medical Technology demonstrating the clinical accuracy and reliability of SmartSpectra's contactless vital signs measurement technology against hospital-grade reference equipment.
Paul Lagasse, Medical Research and Development Command
The U.S. Army Institute of Surgical Research is partnering with Presage Technologies to advance imaging-based vital sign monitoring for hemorrhage triage, aiming to improve rapid assessment of casualties in austere and combat environments.
Carlo A. Canepa, MD; Dana R. Levin, MD, MPH; Amit S. Padaki, MD, MS
Study conducted at the HI-SEAS space-analog site in Mauna Loa, Hawai'i using Presage Technologies software to acquire heart rate and respiratory rate via camera. Camera-acquired heart rate showed high correlation (R ~0.95) to conventional measurements, demonstrating feasibility of contactless vital signs in wilderness and space-analog environments.
Amit S. Padaki1*; Alexa L. Zarzour2; Kelly R. Keene1; Carlo A. Canepa1; Dana R. Levin3; Erik L. Antonsen3
1. Baylor College of Medicine, Houston, TX · 2. Sam Houston State University College of Osteopathic Medicine, Conroe, TX · 3. Massachusetts General Hospital, Boston, MA · Frontiers in Medical Technology, Volume 7 – 2025, Section: Diagnostic and Therapeutic Devices · DOI: 10.3389/fmedt.2025.1728913 · Open Access (CC BY 4.0) · Published January 20, 2026
ABSTRACT
Objectives: To evaluate the accuracy of heart rate and respiratory rate acquired by commercially-available device cameras and software. Methods: One-hundred and eleven subjects were enrolled at an urban academic teaching hospital in Texas. Heart rate (HR) and respiratory rate (RR) measurements were obtained using three commercially-available cameras and processed using software from Presage Technologies. These values were compared to manual counts of a three-lead ECG and end-tidal CO2 from a patient monitor. Results: The cameras were able to capture HR in 83% of the measurements and RR in 94% of the RR measurements. Camera-acquired HR showed an extremely high correlation with R∼0.99 and a root-mean-square error (RMSE) of 1.62. Respiratory rate showed a high correlation with R∼0.91 and an RMSE of 1.71. Conclusions: Heart rate and respiratory rate can be accurately acquired using commercially-available camera devices and software for signal processing.
INTRODUCTION
Vital signs are critically important to modern emergency medical care. These metrics [heart rate (HR), respiratory rate (RR), blood pressure, temperature, and pulse oximetry] have been shown to predict a patient's morbidity (1), mortality (2), need for ICU-level admission (3), and suitability for discharge (4). As such, vital signs are used at multiple points of emergency department (ED) care, from triage through treatment through disposition.
Unfortunately, the acquisition of vital signs can be a time-intensive process. Time-and-motion studies have shown the measurement and recording of vital signs may take over five minutes per patient and longer with interruptions (5). Increased crowding, lack of available staff, and length-of-stay have all been shown to delay or decrease the frequency of vital sign measurements (6, 7). Some studies have shown a majority of both children and adults may be missing vital signs during their ED stays (8, 9). Other studies have shown repeat vital signs to be taken less frequently than every two hours (10). With increasing boarding and crowding in ED settings, the frequency of vital sign acquisition may be significantly worse than this reported data.
Many emergency departments have sought to improve vital sign acquisition through various quality improvement initiatives. Some researchers have also evaluated alternative means of vital sign acquisition. Telemedical providers, for instance, have evaluated the accuracy of patients' self-reported vital signs (11). Device studies have also looked at non-contact vital signs via radar (12) and thermal imaging (13). More recently, studies have evaluated the feasibility of acquiring vital signs using cameras (14).
These studies of camera-acquired vital signs have shown promise, though they are limited by small sample sizes (15, 16). Here, we present the clinical validation of camera-acquired vital signs obtained from commercially available products over a larger sample size.
METHODS
Patients were recruited on a volunteer basis at the Ben Taub Hospital emergency department in Houston, TX. Subjects were required to be between 18 and 75 years of age. Subjects were excluded if they required immediate medical care, had facial scarring, or if they self-reported pregnancy.
One hundred and eleven subjects participated in the study. These included hospital employees and friends and family members of patients in the ED. Thirty-eight were male, seventy-three were female, with ages ranging from 18 to 70 years old. Subjects were predominantly Black and Hispanic.
Table 1 — Patient demographics (Total n = 111): Gender: Male 38, Female 73 | Age (years): Mean 38.8, Range 18–70 | Height (inches): Mean 66.5, Range 61–75 | Weight (lbs): Mean 165.3, Range 108–350 | Race: White/Caucasian 3, Black/African American 70, Hispanic 24, Asian/Pacific Islander 9, American Indian/Alaskan 5 | Skin Tone (Fitzpatrick Scale): 1→10, 2→14, 3→18, 4→14, 5→29, 6→25 | Heart rate range: 54–106 | Respiratory rate range: 6–26
Subjects were taken to a dedicated room used for the research project. They were seated in a chair and placed on continuous monitoring with a 3-lead ECG and continuous end-tidal capnometry. Subjects were instructed to face the study cameras while keeping their face still within the camera frame.
Camera data was captured simultaneously with three devices: a Samsung Galaxy S24 Android phone, an iPhone 16 Pro, and a Logitech c920e webcam. The smartphone data was captured over thirty seconds. The webcam data was captured over sixty seconds (thirty seconds overlapping with the smartphones followed by an additional thirty seconds).
The camera data was processed using software from Presage Technologies (Herndon, VA). Signal analysis of pulse plethysmography and chest wall movements were used to calculate heart rate and respiratory rate, respectively. Data processing occurred on the smartphones and a laptop connected to the webcam. Readouts were provided within about thirty seconds. These values are subsequently referred to as Android HR/RR, iPhone HR/RR, and Webcam HR/RR respectively. These measurements were also aggregated into a Pooled HR/RR.
The control data was acquired by a manual count of the waveforms from a Phillips Intellivue MX450 patient monitor over the same time intervals. The presence of a QRS upstroke on the ECG monitor was considered a heartbeat. The downslope of an inhalation on end-tidal capnometry was considered a breath. These values are subsequently referred as control HR/RR. Ten patients were selected at random for review by a second researcher to confirm accuracy of the control measurements.
Statistical analysis was performed using Microsoft Excel and the JASP statistical software. Linear regressions were performed to calculate Pearson correlation coefficients and root-mean-square errors (RMSEs) for the camera-acquired vital signs.
RESULTS
Capture rates: 278 of 333 possible HR measurements were captured by the software, roughly 83%. 314 of 333 possible RR measurements were captured by the software, roughly 94%. These values were similar across the three devices.
Table 2 — Camera HR/RR capture rates: Webcam: HR 90/111 (81.1%), RR 106/111 (95.5%) | Android: HR 96/111 (86.5%), RR 103/111 (92.8%) | iPhone: HR 92/111 (82.8%), RR 105/111 (94.6%) | Pooled: HR 278/333 (83.5%), RR 314/333 (94.3%)
Correlation and error: Linear regression of the pooled captured HR measurements yielded a correlation coefficient of 0.990 and an RMSE of 1.62. The mean difference was 0.52 beats/minute with an interquartile range (IQR) of 0–1.
Linear regression of the pooled captured RR measurements yielded a correlation coefficient of 0.906 and an RMSE of 1.71. The mean difference was −0.18 breaths/minute with an IQR of −1 to 1.
Table 3 — Correlation and RMSE of camera-acquired vital signs vs. Control: Webcam: HR Corr .992, HR RMSE 1.32, RR Corr .924, RR RMSE 1.75 | Android: HR Corr .991, HR RMSE 1.73, RR Corr .915, RR RMSE 1.62 | iPhone: HR Corr .989, HR RMSE 1.77, RR Corr .903, RR RMSE 1.74 | Pooled: HR Corr .990, HR RMSE 1.62, RR Corr .906, RR RMSE 1.71
The random scatter on the Bland-Altman plots for HR and RR suggest that there is good agreement between the methods with no systematic bias or trends.
Proportion meeting FDA recommendations: The FDA standard for heart rate measurement in medical devices is accuracy to within 3 beats per minute. 266 of 278 or about 96% of captured measurements would meet this standard. The RMSE of 1.62 would also meet this standard. The FDA standard for respiratory rate measurement in medical devices is accuracy to within 2 breaths per minute. 280 of 314 or about 89% of captured RR measurements would meet this standard. The RMSE of 1.71 would also meet this standard.
Interrater reliability: The ten randomly sampled control group measurements were reliably reproduced between the two staff counts. All HR controls were within 1. Nine of ten RR controls were within 1, with the last value being different by 2.
DISCUSSION
These data show that commercially available cameras can be used to accurately acquire both HR and RR. Camera-acquired HR showed a near-perfect correlation of 0.99. Camera-acquired RR showed a high correlation of 0.91.
These high correlations could improve provider confidence in vital sign accuracy. Respiratory rate, in particular, is known to be measured inconsistently and inaccurately (17). Studies of manual measurement of RR have shown variability in measurement, value bias indicative of estimation, and even outright recording omissions (18). Studies using RR measurements taken from video recordings show only a moderate intraclass correlation coefficients (ICCs) of 0.64–0.662 (19, 20). Latten et al. estimated these inaccurate measurements of RR could then lead to inaccurate scoring in 8%–37% of clinical decision rules (20). The ICC of RR taken from actual patients is likely lower than the ICC of RR taken from video recordings. This would be further complicated by delayed or absent data. The downstream effect on clinical decision making would likely be even more pronounced. Having more consistent and reliable measurements could improve provider confidence in and utilization of RR in emergency department patient management.
This study is subject to several limitations. We first note that this study was performed on generally healthy volunteers. Participants were specifically excluded if they required medical attention. Heart rates ranged from 54 to 106 and respiratory rates ranged from 6 to 26. Additional testing will be needed to evaluate subjects with more abnormal rates. Further, all patients appeared to be in sinus rhythm on the monitor. Additional testing will be needed to evaluate subjects with irregular heart rhythms.
Additionally, this study was performed with fixed cameras and lighting. A prior study of this technology showed lower correlation for both heart rate (R ∼ 0.95) and respiratory rate (R ∼ 0.65) when the technology was used in a less controlled environment (21). (Of note, the prior study used an earlier software version and also used pulse oximetry as a HR control and visual count as an RR control.)
Further, this study provided only HR and RR. Complete vital signs typically include temperature, blood pressure, and pulse oximetry in addition to those provided here.
Future studies could also be expanded in multiple other directions. Subjects could potentially activate the device themselves, allowing for vital sign acquisition without a staff member present. With advances in target recognition and selection, a single mounted camera could be eventually be used to obtain vital sign information regarding multiple subjects in a single room. The available data could be used to trend blood pressure or other physiologic parameters such as compensatory reserve.
Overall, this technology could show benefit in multiple settings. First, it could improve frequency of vital sign acquisition in settings with crowding or low staffing. This could not only improve single-patient care, but also throughput. Second, this technology could improve care in austere and telemedical settings in which no provider would be available to take vital signs at all. Third, this technology could allow for safe vital sign acquisition from a distance in situations where patients or their immediate environments could be unsafe. Improving the software to return parameters beyond heart rate and respiratory rate would increase the potential benefits and use cases further.
CONCLUSIONS
This study shows that heart rate and respiratory can be accurately acquired using commercially available camera devices. Use of this type of non-contact measurement could improve the frequency of vital sign measurements in the ED and clinician trust in vital signs measured in the ED, potentially leading to improved patient care and throughput. This type of study helps lay the foundation for inclusion of camera-based vital signs into clinical workflows in an ER. Additional work is required to document effectiveness in busy clinical settings and the effects on workflow and timing.
STATEMENTS
Data availability: The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. Ethics: The studies involving humans were approved by Baylor Institutional Review Board, protocol H-54364. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.
Author contributions: AP: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. AZ: Data curation, Investigation, Methodology, Writing – review & editing. KK: Data curation, Investigation, Methodology, Supervision, Writing – review & editing. CC: Investigation, Methodology, Writing – review & editing. DL: Conceptualization, Formal analysis, Funding acquisition, Visualization, Writing – review & editing. EA: Conceptualization, Formal analysis, Funding acquisition, Visualization, Writing – review & editing.
FUNDING
The author(s) declared that financial support was received for this work and/or its publication. Grant funding from Department of Defense Award HQ08452390071 via Presage Technologies. Presage Technologies was not involved in the design of this study, data collection, or preparation of this manuscript.
CONFLICT OF INTEREST
AP, AZ, KK report grant money to the Baylor College of Medicine to conduct research conceived and sponsored by Presage Technologies. The remaining author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement: The author(s) declared that generative AI was not used in the creation of this manuscript.
Publisher's note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
REFERENCES
1. Quinten VM, et al. Repeated vital sign measurements in the emergency department predict patient deterioration within 72 h. Scand J Trauma Resusc Emerg Med. (2018) 26:1–12. doi: 10.1186/s13049-018-0525-y
2. Candel BG, et al. The association between vital signs and clinical outcomes in emergency department patients of different age categories. Emerg Med J. (2022) 39(12):903–11. doi: 10.1136/emermed-2020-210628
3. Barfod C, et al. Abnormal vital signs are strong predictors for intensive care unit admission and in-hospital mortality in adults triaged in the emergency department. Scand J Trauma Resusc Emerg Med. (2012) 20:1–10. doi: 10.1186/1757-7241-20-28
4. Nguyen OK, et al. Vital signs are still vital: instability on discharge and the risk of post-discharge adverse outcomes. J Gen Intern Med. (2017) 32:42–8. doi: 10.1007/s11606-016-3826-8
5. Dall'Ora C, et al. How long do nursing staff take to measure and record patients' vital signs observations in hospital? A time-and-motion study. Int J Nurs Stud. (2021) 118:103921. doi: 10.1016/j.ijnurstu.2021.103921
6. Johnson KD, et al. The factors that affect the frequency of vital sign monitoring in the emergency department. J Emerg Nurs. (2014) 40(1):27–35. doi: 10.1016/j.jen.2012.07.023
7. Hope J, et al. Why vital signs observations are delayed and interrupted on acute hospital wards: a multisite observational study. Int J Nurs Stud. (2025) 164:105018. doi: 10.1016/j.ijnurstu.2025.105018
8. Gravel J, et al. High rate of missing vital signs data at triage in a paediatric emergency department. Paediatr Child Health. (2006) 11(4):211–5. doi: 10.1093/pch/11.4.211
9. Gerdtz MF, et al. Evaluation of a multifaceted intervention on documentation of vital signs at triage. Emerg Med Australas. (2013) 25(6):580–7. doi: 10.1111/1742-6723.12153
10. Miltner RS, et al. Exploring the frequency of blood pressure documentation in emergency departments. J Nurs Scholarsh. (2014) 46(2):98–105. doi: 10.1111/jnu.12060
11. Kagiyama N, et al. Validation of telemedicine-based self-assessment of vital signs for patients with COVID-19. J Telemed Telecare. (2023) 29(8):600–6. doi: 10.1177/1357633X211011825
12. Wu Q, et al. A non-contact vital signs detection in a multi-channel 77 GHz LFMCW radar system. IEEE Access. (2021) 9:49614–28. doi: 10.1109/ACCESS.2021.3068480
13. Negishi T, et al. Contactless vital signs measurement system using RGB-thermal image sensors and its clinical screening test on patients with seasonal influenza. Sensors. (2020) 20(8):2171. doi: 10.3390/s20082171
14. McDuff D. Camera measurement of physiological vital signs. ACM Comput Surv. (2023) 55(9):1–40. doi: 10.1145/3558518
15. Khanam F-T-Z, et al. Non-contact automatic vital signs monitoring of neonates in NICU using video camera imaging. Comput Methods Biomech Biomed Eng Imaging Vis. (2023) 11(2):278–85. doi: 10.1080/21681163.2022.2069598
16. Caroppo A, et al. Vital signs estimation in elderly using camera-based photoplethysmography. Multimed Tools Appl. (2024) 83(24):65363–86. doi: 10.1007/s11042-023-18053-3
17. Loughlin PC, et al. Respiratory rate: the forgotten vital sign—make it count! Jt Comm J Qual Patient Saf. (2018) 44(8):494–9. doi: 10.1016/j.jcjq.2018.04.014
18. Kallioinenin N, et al. Quantitative systematic review: sources of inaccuracy in manually measured adult respiratory rate data. J Adv Nurs. (2021) 77(1):98–124. doi: 10.1111/jan.14584
19. Brabrand M, et al. Measurement of respiratory rate by multiple raters in a clinical setting is unreliable: a cross-sectional simulation study. J Crit Care. (2018) 44:404–6. doi: 10.1016/j.jcrc.2017.12.020
20. Latten GH, et al. Accuracy and interobserver-agreement of respiratory rate measurements by healthcare professionals, and its effect on the outcomes of clinical prediction/diagnostic rules. PLoS One. (2019) 14(10):e0223155. doi: 10.1371/journal.pone.0223155
21. Canepa CA, Levin DR, Padaki AS. Comparison of camera-acquired vital signs to conventional vital signs in a space-analog environment. Wilderness Environ Med. (2024) 36(1_suppl):10806032241291994. doi: 10.1177/10806032241291994
Paul Lagasse, Medical Research and Development Command
Fort Detrick, Md.
Uncontrolled blood loss from traumatic injury is the leading cause of preventable death on the battlefield, which makes prompt and accurate triage essential for maximizing warfighter survival. However, the body's own defense mechanisms can make it difficult to identify someone at greatest risk for the early onset of shock. A new technology that builds on pioneering research conducted at the U.S. Army Institute of Surgical Research may soon enable care givers to assess patient risk more quickly and accurately using data collected from ordinary digital cameras, thereby improving the survivability of combat casualties and increasing the combat power and effectiveness of field units.
Traditionally, medical professionals assess a patient's status by relying on the patient's vital signs – blood pressure, heart rate, and oxygen saturation. However, when subject to traumatic blood loss, the human body automatically engages a survival mechanism – called the compensatory reserve – that concentrates the remaining blood near the heart, brain, and other vital organs. This results in these standard vital signs to remain relatively normal for a time until the body can no longer compensate for the blood loss, triggering what doctors call a "crash" that is often fatal.
"It's analogous to driving your car and being unsure of how much gas you had left because the gas gage doesn't change until the very end, just as you are about to run out of gas," explains Dr. Victor Convertino, USAISR's senior scientist for combat casualty care and director of the Battlefield Health and Trauma Center for Human Integrative Physiology. "You may be able to drive for 300 miles, or you may only have three miles left. That's the challenge we're faced with."
Convertino's groundbreaking research into the body's inherent physiologic compensatory mechanisms led to the development of the compensatory reserve measurement, or CRM, a machine-learning algorithm that provides a more reliable prediction of hemorrhagic shock risk than traditional vital signs.
In addition to providing medical care providers with a more accurate assessment tool for hemorrhagic shock, the CRM is the first monitoring technology that can assess the blood volume status of an individual patient regardless of the patient's individual tolerance for blood loss, which determines how quickly they will crash from an ongoing loss of blood. This has long been a mystery for clinicians; Convertino calls it a "curveball" that can catch even the most attentive doctors and nurses off guard.
To put this new diagnostic capability into the hands of medical professionals and warfighters, Convertino, Dr. Jose Salinas, and Dr. Eric Snider of USAISR's Automation and Engineering Group – one of five research groups within USAISR focusing on using state-of-the-art technologies such as advanced artificial intelligence to address critical medical gaps – have teamed up with health software firm Presage Technologies to incorporate the CRM algorithm into a new software application that could transform battlefield triage of trauma-induced hemorrhage.
The software builds on work that Presage has done with several DOD clients to demonstrate that commercial grade video cameras – such as those used in drones, webcams, and smart phones – are capable of capturing microscopic movement and color changes in a person's skin caused by their pulse. The software converts those changes into a waveform that can be compared against the CRM algorithm to predict the patient's risk of slipping into shock. Presage has developed the software to the point where it is ready for submission to the U.S. Food and Drug Administration seeking its clearance to legally market the software for use in novel military and civilian medical applications.
With such a capability readily available, medics operating in a combat environment could use drones to observe injured warfighters without risking injury to themselves, for example. Nurses responding to a mass casualty incident could use their phones to perform triage using just a few seconds of video footage for each patient. Physicians would be able to continuously monitor patients postoperatively for signs of internal bleeding, shock, and sepsis. Such cutting-edge, deployable trauma care solutions promise to materially improve warfighter survivability and lethality.
As part of this effort, the Metis Foundation, a San Antonio-based nonprofit that supports military medical research, is providing extramural assistance for the project using funding provided by the Medical Technology Enterprise Consortium, a nonprofit international group of over 600 academic institutions, businesses, nonprofits and other organizations in the biomedical technology sector. Convertino says that the foundation's participation has been crucial for ensuring the success of the project.
Convertino hopes that the collaboration with Presage to develop diagnostic capabilities for digital cameras will be the first of many innovative applications of the CRM, which he and his colleagues have spent years researching and refining.
"The beauty of the CRM algorithm is that it was purposely built to be hardware agnostic," says Convertino. "We can team with almost any industry partner who has a monitoring capability, integrate the algorithm into it, and then they can bring their device to the FDA as a software upgrade that doesn't require a lengthy approval process as a new medical device. That will help speed our ultimate goal, which is to get this capability into the hands of medics, whether they're military or civilian, so it can help them save lives with earlier lifesaving interventions."
Carlo A. Canepa, MD; Dana R. Levin, MD, MPH; Amit S. Padaki, MD, MS
Department of Emergency Medicine, Baylor College of Medicine, Houston, TX · Wilderness & Environmental Medicine, Vol. 36, Issue 1_suppl, pp. S77–S82 (Sept. 2025, published online Nov. 10, 2024) · DOI: 10.1177/10806032241291994 · License: CC BY-NC 4.0
ABSTRACT
Introduction: Vital sign acquisition is a key component of modern medical care. In wilderness and space medical settings, vital sign acquisition can be a difficult process because of limitations on available personnel or lack of access to the patient. Camera-acquired vital signs could address each of these difficulties.
Methods: Healthy volunteers used software designed by Presage Technologies to acquire heart rate and respiratory rate at the HI-SEAS space-analog site in Mauna Loa, Hawai'i. Camera-acquired vital signs were compared to more conventionally acquired vital signs.
Results: Camera-acquired heart rate showed high correlation to conventionally acquired heart rate (R ∼ 0.95). Camera-acquired respiratory rate showed moderate correlation (R ∼ 0.65).
Conclusions: These results show that camera acquisition of vital signs is theoretically feasible in wilderness and space-analog environments. HR may be highly accurate even using current technology. Additional studies will be needed to further validate other types of camera sensors and other potential environments such as partial gravity and microgravity.
INTRODUCTION
Vital signs constitute a key component of medical care in most modern medical settings. These five metrics (heart rate, respiratory rate, blood pressure, temperature, and pulse oximetry) can be key indicators of a patient's degree of illness and response to various therapies. As such, vital signs play key roles in patient triage, treatment, and disposition. Vital signs have been shown to predict patient outcomes as wide-ranging as morbidity, mortality,1 transfer to a higher level of care,2 and need for readmission when discharged.3
Despite their importance, the acquisition of vital signs can be delayed in emergency departments and other conventional medical settings. Studies have shown that factors such as nurse staffing ratios4 and crowding5 can lead to decreased frequency of vital sign measurements. Some studies have shown vital sign updates occur as infrequently as only once every 2 hours.6 Many emergency departments are seeking to address this deficiency through a variety of quality-improvement initiatives. Of note, in modern telemedical care, patients may have no physical contact with a medical provider. Some studies have sought to assess the feasibility of patients self-reporting vital signs in those environments.7
Vital signs are even more important in wilderness, military, and other austere settings. There, a lack of access to laboratory testing, imaging, and other medical diagnostics often leaves vital signs as the sole objective marker of a patient's degree of illness. Spaceflight may be the most austere of these environments. All medical resources sent in spaceflight must be considered carefully for cost, mass, volume, and ease of use. Astronauts, often not medical experts, must also receive training in the use of medical resources. With this in mind, simpler, more reusable medical resources would be preferred. In addition, monitoring vital signs during astronaut exercise is often used for research purposes and to ensure each astronaut gets an adequate exercise prescription. However, wearable sensors are often described as uncomfortable, and movement can make them unreliable.8
In the past few years, multiple research groups have begun to validate non-contact vital sign acquisition in terrestrial environments.9 These studies, usually of camera-acquired vital signs, could potentially improve both single-patient care and clinical system function. Other groups have attempted to test wearable vital sign sensors for potential use in spaceflight.10 Here, we present our experience using a simple iPhone (™) application to acquire heart rate (HR) and respiratory rate (RR) using phone cameras at a space analog site.
METHODS
Six healthy astronaut analog volunteers were recruited on a convenience basis at the high-altitude Hawai'i Space Exploration Analog and Simulation (HI-SEAS) habitat in Mauna Loa, Big Island, Hawaii. Three participants were male, three were female, and their ages ranged from 18–52 years old. Measurements were taken multiple times over the course of a six-day simulated spaceflight. Measurements were taken at rest, after exercise, and after simulated extra-vehicular activity. Measurements were also taken in either a designated "medical room" or in other locations throughout the habitat.
Camera-acquired HR and RR were obtained using Presage Technologies11 software installed on an iPhone device. This process involved activating the software and placing the patient's face within the camera's field of view. HR and RR were calculated based on pulse plethysmography measurement and chest wall movement, respectively. No physical contact with the subject was required during this time. Data processing occurred on the phone itself, and readouts were provided within about 30 seconds. These values are subsequently referred to as App HR and App RR.
Conventionally acquired HR was acquired with a portable digital pulse oximeter. Conventionally acquired RR was acquired through visual measurement by an observer over 30 seconds, then doubled to calculate the minute rate. These values are subsequently referred to as Measured HR and Measured RR.
These values were then compared using a Student's t-test and standard linear regression.
RESULTS
Sixty measurements were made over the course of six days (averaging about ten measurements per participant.) Of note, 59 of these were in the seated position, and one was in the supine position.
App HR vs. Measured HR: App HR was highly correlated to Measured HR. Mean difference (App HR – Measured HR): −0.96 | Standard error: 0.80 | Interquartile range: (−3.65, 1.85) | Pearson correlation coefficient: 0.946
App RR vs. Measured RR: Moderate correlation was present. Mean difference (App RR – Measured RR): −0.227 | Standard error: 0.43 | Interquartile range: (−2.0, 1.45) | Pearson correlation coefficient: 0.652
Each study participant interacted with the application, and all participants anecdotally found it intuitive and easy to use. There were occasional issues with ambient lighting, subject movement, or facial positioning. These could be addressed through increasing the ambient lighting and repositioning of the subject. For one measurement, the app appeared to freeze on data collection and also required a reset.
DISCUSSION
Overall, these data suggest that HR can be reliably acquired using a non-contact smartphone application at a high-altitude space analog simulation site. RR can also be acquired in this manner, though it may have more errors.
Of note, study participants did note visual observation of the respiratory rate was itself difficult and may have led to error. Studies have shown that different manual measurements of RR have an intraclass correlation coefficient of 0.64, similar to the correlation between manual measurements and cameras seen in this study.12 The US Food and Drug Administration (FDA) itself has recommended the use of an absolute difference in RR (+/–2 BPM) rather than correlation to assess the accuracy of medical devices measuring RR.13 Using that standard, 36/60 measurements, or 60%, would meet the FDA standard for accuracy. Other studies have used an absolute difference of 4 BPM to assess accuracy.12 At that standard, 51/60, or 85%, would be considered accurate.
We acknowledge that this is a small pilot study only showing the feasibility of using this type of technology in austere and space-analog environments. No formal power calculations were performed, as the goal of the work was to demonstrate the engineering endpoint of adequate data capture processes. As this was not a systematic health study, no statistics were required to meet the engineering goal. Future studies are planned that will address systematic health parameter evaluations.
Further, our study calculates HR via pulse, a measurement of force transmission through the blood vessels. While HR and pulse are essentially interchangeable throughout normal heart rate and blood pressure ranges, this may not be the case at more extreme values. The current gold standard measurement for HR is electrocardiography (EKG). Future studies could compare camera-acquired HR to EKG-measured HR instead of pulse.
Future studies could potentially be expanded in multiple other new directions, including measurement of time required for vital sign measurement or measurement of vital signs during movement or active exercise. The camera used to acquire vital signs could also be varied to include fixed sensors more representative of spaceflight environments or integration into helmets as might be seen in military medicine. Additionally, while this study was performed at ∼8200 ft of altitude at the HI-SEAS habitat, this still occurred at normal terrestrial gravity of roughly 1G. Further studies will need to validate this technology in partial-gravity and zero-gravity environments.
CONCLUSION
Our study shows the feasibility of using a smartphone application to acquire HR and RR in wilderness and space-analog environments. Use of this type of non-contact measure could potentially improve the frequency of vital sign measurements in both conventional and austere medical settings. Specific to spaceflight, this technology could both expand and simplify medical resources available to future astronauts.
DECLARATION OF CONFLICTING INTERESTS
IRB: This study follows policies issued by the Baylor College of Medicine and was reviewed with experienced research staff. As a small engineering study that used volunteers to test data capture processes only, this study did not require IRB review.
FUNDING
The authors received no financial support for the research, authorship, and/or publication of this article.
ORCID IDs
Carlo A. Canepa https://orcid.org/0009-0001-5723-0128 | Amit S. Padaki https://orcid.org/0000-0002-2689-2233
REFERENCES
1. Candel BG, et al. The association between vital signs and clinical outcomes in emergency department patients of different age categories. Emerg Med J. 2019; 39(12): 903–11.
2. Hodgson NR, et al. Association of vital signs and process outcomes in emergency department patients. West J Emerg Med. 2019; 20(3): 433.
3. Gabayan GZ, et al. Emergency department vital signs and outcomes after discharge. Acad Emerg Med. 2017; 24(7): 846–54.
4. Redfern OC, et al. The association between nurse staffing levels and the timeliness of vital signs monitoring. BMJ Open. 2019;9(9):e032157.
5. Depinet HE, et al. The effect of emergency department crowding on reassessment of children with critically abnormal vital signs. Acad Emerg Med. 2014; 21(10): 1116–20.
6. Lambe K, et al. Frequency of vital sign assessment and clinical deterioration in an Australian emergency department. Australas Emerg Nurs J. 2016; 19(4): 217–22.
7. Kagiyama N, et al. Validation of telemedicine-based self-assessment of vital signs for patients with COVID-19. J Telemed Telecare. 2023; 29(8): 600–6.
8. Glazkova N, et al. Non-invasive wearable ECG-patch system for astronauts and patients on earth. Acta Astronaut. 2020; 166: 613–8.
9. Selvaraju V, et al. Continuous monitoring of vital signs using cameras: a systematic review. Sensors. 2022; 22(11): 4097.
10. Di Rienzo M, et al. Development of a smart garment for the assessment of cardiac mechanical performance and other vital signs during sleep in microgravity. Sens Actuators A Phys. 2018; 274: 19–27.
11. Presage Technologies. https://physiology.presagetech.com/. Accessed 06/19/2024.
12. Latten GH, et al. Accuracy and interobserver-agreement of respiratory rate measurements by healthcare professionals. PLoS One. 2019; 14(10): e0223155.
13. US Food and Drug Administration. "Re: K211926" https://www.accessdata.fda.gov/cdrh_docs/pdf21/K211906.pdf. Accessed 06/19/2024.