Clinical performance evaluation of a cuffless continuous non-invasive blood pressure monitoring device in ICU patients

Abstract

Purpose

To evaluate the accuracy, stability, and dynamic BP tracking capability of a cuffless continuous non-invasive blood pressure (CNIBP) wearable system (Shenzhen Mindray Bio-Medical Electronics Co., Ltd.) to track dynamic blood pressure (BP) changes in intensive care unit (ICU) patients.

Methods

A multicenter, prospective clinical trial enrolled ICU patients aged ≥ 12 years who required continuous invasive blood pressure (IBP) monitoring for ≥ 24 h. The mWear wearable monitoring system (Mindray) served as the test device, with IBP as the reference standard. The study followed the International Organization for Standardization (ISO) 81060-3:2022 guidelines to evaluate the accuracy, stability, and dynamic BP tracking.

Results

Forty-nine participants (mean age, 57.3 ± 18.7 years) were included in the study. The mean differences between CNIBP and IBP measurements for systolic blood pressure (SBP), diastolic blood pressure (DBP), and mean arterial pressure (MAP) were -0.31 mmHg, -0.81 mmHg, and -0.37 mmHg, respectively, all within the predefined 95% confidence interval (CI). The corrected experimental standard deviations (SD) were 5.3, 3.6, and 3.6 mmHg, respectively, consistent with the ISO 81060-3:2022 criteria (mean difference: -6 to 6 mmHg; corrected experimental SD: ≤10 mmHg). The Pearson correlation coefficient (PCC) exceeded 0.9, indicating a strong linear correlation. Stability assessments demonstrated that CNIBP remained within ISO error ranges over 24 h. For BP change-tracking ability, the median absolute error (E_percent, 50th percentile) averaged ≤25%, and the 85th percentile averaged ≤50%, both of which met ISO requirements.

Conclusion

The findings indicate that CNIBP device measurements are consistent with IBP readings in terms of accuracy, stability, and dynamic BP tracking capability within the ICU. This consistency effectively supports real-time hemodynamic assessment in critically ill patients.

Keywords

continuous non-invasive blood pressure accuracy stability dynamic blood pressure tracking capability pulse wave transit time intensive care unit

Introduction

Blood pressure (BP) is a fundamental vital sign routinely monitored in intensive care unit (ICU) patients. Dynamic fluctuations in BP reflect the underlying physiological state and may indicate pathological processes associated with adverse outcomes.^1,2 Therefore, real-time, accurate, and continuous BP monitoring is essential for timely clinical decision-making and effective therapeutic interventions in critically ill patients.

Invasive blood pressure (IBP) monitoring provides a direct measurement of intravascular pressure via an arterial catheter and is widely regarded as the gold standard for continuous BP assessment.³ Despite its accuracy, IBP is associated with risks, such as infection,⁴ catheter insertion failure, and artifacts resulting from resonance and damping effects.⁵ Additional technical challenges may also arise in patients with peripheral vascular abnormalities.⁶ In contrast, traditional cuff-based non-invasive blood pressure (NIBP) monitoring—including oscillometric⁷ and auscultatory methods⁸—avoids vascular puncture; however, it remains intermittent, time-consuming, and sensitive to variations in cuff size,⁹ placement, patient position,¹⁰ age,¹¹ and arterial stiffness, which can compromise measurement accuracy.¹² Moreover, its inability to provide continuous BP monitoring may lead to missed or delayed detection of hypotensive events during anaesthetic induction.¹³

The introduction of cuffless continuous BP monitoring systems have emerged to address the limitations of traditional cuff-based methods. Early investigations began in 1963 with arterial tension measurement,¹⁴ followed by the introduction of continuous finger-cuff BP monitoring technology using the volume-clamp method.¹⁵ More recently, advancements in sensors and signal processing have enabled the development of estimation methods based on pulse wave velocity (PWV) and pulse wave analysis (PWA), which have become prominent areas of research.¹⁶ Despite the diversification of these technological approaches, cuffless devices still face challenges related to measurement accuracy, individual calibration, and clinical validation.

To address these limitations, Shenzhen Mindray Bio-Medical Electronics Co., Ltd. developed a new type of cuffless BP monitor based on the pulse wave transit time (PWTT) principle. PWTT refers to the time interval between the R wave of the electrocardiogram (ECG) and the arrival of the peripheral photoplethysmography (PPG) signal. This technology is integrated into a wearable terminal, enabling continuous non-invasive BP monitoring. Compared with traditional cuff-based devices, the wrist-worn design enhances patient comfort. Moreover, unlike devices that only support intermittent measurements¹⁷ or exhibit 24h drift,^18,19 this device provides true continuous output and can accurately track rapid BP fluctuations.

This study is the first to systematically validate the accuracy, long-term stability, and dynamic tracking capability of a wrist-worn, continuous non-invasive blood pressure (CNIBP) monitoring device based on PWTT in an ICU setting, in accordance with the International Organization for Standardization (ISO) 81060-3:2022.²⁰ Unlike previous studies that focused only on short-term accuracy, this standard provides a more comprehensive statistical framework, enhancing the reliability of the evaluation. The present validation establishes a solid foundation for the clinical application of CNIBP monitoring in critically ill patients. This approach offers a safe and effective alternative, particularly when arterial cannulation is contraindicated or prolonged monitoring is required, thereby supporting real-time hemodynamic management.

Methods

Written informed consent was obtained from all participants or their legal guardians prior to enrollment in the study.

Device details

Test Device: The mWear wearable monitoring system (Shenzhen Mindray Bio-Medical Electronics Co., Ltd.) was utilized, comprising an EP30 main unit, a pulse oximetry sensor, a 3-lead ECG, and a multi-parameter mobile module (ES30) for CNIBP monitoring based on PWTT (Figure S1). The device provides CNIBP readings at a minimum interval of 3 s and requires recalibration every 24 h. Additional physiological parameters measured included body temperature, oxygen saturation (SpO₂), and respiratory rate (RR), with all data displayed on a monitor for clinical analysis.

Reference Standard: In accordance with ISO 81060-3:2022, CNIBP measurements were compared with IBP values obtained from arterial catheters placed in the radial, brachial, or femoral arteries. CNIBP data were collected using an ePM 12M patient monitor (Shenzhen Mindray Bio-Medical Electronics Co., Ltd.).

Device familiarization

The research team comprised ICU nurses, each with at least five years of clinical experience and having performed over 400 IBP measurements. All procedures were conducted in strict accordance with ISO 81060-3:2022, a joint standard issued by the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC), ensuring methodological rigor. To ensure standardized data collection, the team underwent a familiarization phase with the device and calibration measurement process prior to study initiation and protocol implementation.

Sample size

The sample size was determined in accordance with ISO 81060-3:2022 guidelines.²⁰ An estimated intraclass correlation coefficient (I_{CC_est} = 0.87) for systolic, diastolic, and mean arterial pressures was derived from preliminary pilot data. A minimum sample size of 45 participants was required to ensure sufficient statistical power and meet the relevant regulatory requirements (see Equations S1-S5 and S8 in the Supplementary Material).

Recruitment

This multicenter, prospective trial recruited participants between January 8 and August 14, 2024, from the ICUs of three tertiary hospitals in Shenzhen, China: Shenzhen People’s Hospital, Shenzhen Bao’an District People’s Hospital, and Huazhong University of Science and Technology Union Shenzhen Hospital. Following screening and written informed consent, 54 participants were enrolled. Of these 49 participants who met the inclusion criteria—being over 12 years of age and requiring continuous IBP monitoring for ≥ 24 h via arterial catheters in the radial, brachial, or ulnar arteries—were ultimately included in the study.

The exclusion criteria were as follows: a perfusion index (PI) below 0.3; persistent agitation or tremors; pregnancy, frequent arrhythmias such as atrial fibrillation (AF); concurrent participation in other clinical trials; and investigator-determined unsuitability. Participants who withdrew voluntarily, did not complete 24 h monitoring, or experienced technical disruptions were excluded from the final analysis (Figure S2).

Blood pressure measurement procedure

Patients in the ICU who required continuous IBP monitoring for ≥ 24 h due to clinical necessity were enrolled in this study. All participants were bedridden and simultaneously connected to both the CNIBP device and an IBP system during data acquisition. Prior to trial initiation, arterial catheter placement was verified to ensure that the IBP transducer was positioned at the level of the right atrium. The CNIBP device was applied according to the manufacturer’s guidelines, with both CNIBP and IBP monitoring performed on the same upper limb and the SpO₂ probe aligned at heart level. Initial IBP values were used to calibrate the CNIBP device, and the calibration time was recorded. Continuous BP data, including systolic blood pressure (SBP), diastolic blood pressure (DBP), and mean arterial pressure (MAP), were collected ≥ 24 h post-calibration. If the vasoactive drug regimen was modified, including initiation, discontinuation, or dose adjustment, the device was recalibrated within 5 min, and a new 24 h data collection cycle was initiated. No adverse events related to device wearing or arterial catheterization were observed during the study period.

All calibrations in this study were conducted using invasive BP readings to ensure the accuracy of the reference standard and rigorously validate the sensor tracking algorithm, thereby minimizing confounding errors associated with oscillometric cuffs. However, in routine clinical practice, the device is calibrated using standard oscillometric non-invasive BP measurements. During data collection, significant signal artifacts or technical interruptions—such as changes in patient position, suctioning procedures, or temporary sensor detachment—led to the immediate marking and exclusion of the affected data segments to maintain the integrity of paired continuous non-invasive and invasive BP measurements. Research nurses who received standardized training recorded all exclusion decisions in real time according to predefined criteria, documenting both the reason for exclusion and its duration to ensure consistency and objectivity. Because exclusion events primarily resulted from transient, random clinical procedures and were not systematically associated with patient severity of condition or BP levels, the risk of selection bias was considered negligible. In clinical emergencies, data were retained whenever signal quality was sufficient, reflecting real-world monitoring conditions. The study protocol did not interfere with routine clinical management, thereby allowing for the natural observation of hemodynamic parameters.

Data preparation

The performance of the CNIBP device was evaluated across three dimensions: accuracy, stability, and BP change-tracking capability, in accordance with ISO 81060-3:2022. The methodology is described in detail below.

For the accuracy assessment, the CNIBP device had a minimum output period of 3 s. In accordance with standard requirements, the IBP data were divided into non-overlapping segments at 1-min intervals. The average value of each segment (P_ref _j, is the j^th reference reading from the first analysis period of the first test period of the i^th subject) was paired with the CNIBP output value at the end of the segment (P_sut _j, is the j^th BP determination from the first analysis period of the first test period of the i^th subject) to form a paired dataset for accuracy evaluation (Figure S3). The difference x _j was then calculated using Equation S1.

Stability assessment: Given the device maximum re-initialization time of 24 h, six accuracy tests were performed (five within the first 5 h and one at the end of the initialization cycle), each lasting approximately 15 min. The test period was divided into equal analysis and interval segments. With 44 repeated measurements per participant and a CNIBP minimum output cycle of 3 s, each analysis segment lasted 2.5 min. Synchronized CNIBP and IBP data were collected during these segments to form a paired dataset for the stability assessment (Figure S4).

To assess the ability to track BP changes, continuous BP data were collected for 30 min following calibration. Given the device interpretation change evaluation interval (I_CE) is 120s, BP changes during this interval were calculated using (Equation S9), with t_start and t_end representing the start and end times of BP fluctuations, respectively. For the CNIBP to be considered capable of accurately tracking BP changes within I_CE, the analysis segment was required to contain more than 50 BP fluctuations. A BP fluctuation was defined as a change in IBP or CNIBP, where the absolute value of ΔP meets SBP ΔP ≥ 15 mmHg, DBP ΔP ≥ 10 mmHg, or MAP ΔP ≥ 12 mmHg (Equation S10). The absolute error was calculated using (Equation S11), and the specific values are shown in Figure S5.

Statistical analyses

All statistical analyses were performed using SPSS (Version 29.0.1.0). BP values are reported as the mean ± standard deviation (SD), age as the median ± interquartile range (IQR), and categorical variables as percentages.

The accuracy and stability of CNIBP were evaluated using mean error (x̄, Equation S6) and corrected standard deviation ( $s_{c o r r}$ , Equation S7), in accordance with ISO 81060-3:2022. The 95% confidence interval (CI: L95%, U95%) for SBP, DBP, and MAP x̄ were calculated. Acceptability was defined as the $\bar{x}$ not exceeding 6 mmHg and the SD not exceeding 10 mmHg, based on established ISO validation practices for BP measurement devices. Additionally, the number of independent measurements (N_ind) was required to be ≥ 278. The tracking ability for BP changes was assessed using absolute error ( $E_{p e r c e n t}$ , Equation S11). The Pearson correlation coefficient (PCC) was calculated to assess the correlation between CNIBP and IBP, and Bland-Altman plots were used to evaluate the distribution of mean errors between the test and reference devices. For variability assessment, the acceptance criteria were as follows: the average of the 50th percentile (median) of E_percent values of all subjects was required to be no greater than 25%, and the average of the 85th percentile of E_percent values of all subjects was required to be no greater than 50%.

Results

Participants demographics

Between January 8 and August 14, 2024, a total of 54 eligible participants were recruited. After excluding patients who withdrew or provided invalid data, BP data from 49 participants (90.7%) were included in the accuracy analysis, while data from 47 participants (87.0%) were used for stability and variability analyses. As shown in Table 1, 49 participants had a mean age of 57.3 ± 18.7 years, with 31 (63.3%) being over 50, 23 (46.9%) over 60, and 14 (28.6%) over 70. 29 (59.2%) patients were male. The average calibration values of SBP, DBP, and MAP were 133.0 ± 24.8 mmHg, 63.7 ± 16.2 mmHg, and 86.8 ± 17.5 mmHg, respectively.

Table 1.

Clinical characteristics of the study participants.

Characteristics	Results
Age median (IQR), years	59 (28)
Subjects ≥50 years old, n(%)	31 (63.3%)
Subjects ≥60 years old, n(%)	23 (46.9%)
Subjects ≥70 years old, n(%)	14 (28.6%)
Male, n (%)	29 (59.2%)
Calibration SBP, mean (±SD), mmHg	133.0±24.8
Calibration DBP, mean (±SD), mmHg	63.7±16.2
Calibration MAP, mean (±SD), mmHg	86.8±17.5

Abbreviations: DBP: Diastolic blood pressure; MAP: Mean arterial pressure; IQR: Interquartile range; SBP: Systolic blood pressure; SD: Standard deviation.

Assessment of BP accuracy

To assess the accuracy of the CNIBP device, the N_ind of CNIBP measurements and the

\bar{x}

and s _corr values for SBP, DBP, and MAP are provided, comparing CNIBP with IBP measurements. The

\bar{x}

(95% CI) for SBP, DBP, and MAP were -0.3 mmHg (-0.54, -0.09), -0.8 mmHg (-0.96, -0.66), and -0.4 mmHg (-0.52, -0.21), respectively. The s _corr values were 5.3 (SBP), 3.6 (DBP), and 3.6 (MAP). The PCC values were 0.981 (SBP), 0.982 (DBP), and 0.976 (MAP) (Table 2). These results demonstrate that this CNIBP device monitoring fully complies with ISO standards, with the

\bar{x}

in subjects measurements maintained within ±6.0 mmHg, a corrected experimental SD ≤ 10.0 mmHg, and N_ind ≥ 278 cases. The Bland-Altman analysis (Figure 1 and Table S1) presented the agreement between the device under test and IBP measurements for SBP, DBP, and MAP. The analysis indicated a bias of -0.312 mmHg for SBP, with 95% limits of agreement ranging from -10.68 to 10.06 mmHg; a bias of -0.811 mmHg for DBP, with 95% limits of agreement ranging from -7.32 to 6.59 mmHg; and a bias of -0.367 mmHg for MAP, with 95% limits of agreement ranging from -7.95 to 6.32 mmHg. According to ISO 81060-3:2022 standard, the accuracy acceptance criteria for CNIBP monitoring devices are a mean difference of ≤ ±6 mmHg and an SD of the differences of ≤ 10 mmHg. The bias and SD of the differences in all three indicators in this study met the requirements of this standard. These findings support clinically acceptable agreement between CNIBP and IBP for patient BP monitoring and demonstrate the device high accuracy.

Table 2.

Agreement between continuous non-invasive blood pressure and invasive blood pressure for accuracy assessment.

BP type	N_ind of CNIBP(n)	$\bar{x}$ (mmHg) (95% CI)	s_corr (mmHg)	PCC
SBP	1892	-0.31 (-0.54, -0.09)	5.3	0.981
DBP	1791	-0.81 (-0.96, -0.66)	3.6	0.982
MAP	1630	-0.37 (-0.52, -0.21)	3.6	0.976

Abbreviations: BP: Blood pressure; CNIBP: Continuous non-invasive blood pressure; CI: Confidence interval; DBP: Diastolic blood pressure; MAP: Mean arterial pressure; N_ind: The number of independent measurements; PCC: Pearson correlation coefficient; SBP: Systolic blood pressure; $s_{c o r r} :$ Corrected standard deviation; SD: Standard deviation; $\bar{x}$ : Mean error.

Figure 1.

Bland-Altman plot comparing the continuous non-invasive blood pressure and invasive blood pressure device. Note that “ME” (Mean error) in the figure is equivalent to $\bar{x}$ (mean difference) used in the text and tables. Abbreviations: ME: Mean error; STD: Standard deviation.

Results for assessment of BP stability

The stability of the CNIBP device over a 24 h period was evaluated by analyzing 18 segments. For each segment, the N_ind of CNIBP measurements was reported, and the s _corr values between CNIBP and IBP measurements were compared across different BP types. The results demonstrated that for all participants, each BP type (SBP, DBP, and MAP) measured within the predefined time frame met the criteria of s _corr ≤ 10.0 mmHg and N_ind ≥ 278 (Table 3). These findings demonstrate that the CNIBP device exhibits excellent stability.

Table 3.

Agreement between continuous non-invasive blood pressure and invasive blood pressure for stability assessment.

CNIBP vs IBP		$\bar{x}$ (mmHg)			$s_{c o r r} (mmHg)$			N _ind
Time for BP measurement	BP type	1st Anal. Per.	2nd Anal. Per.	3rd Anal. Per.	1st Anal. Per.	2nd Anal. Per.	3rd Anal. Per.	1st Anal. Per.	2nd Anal. Per.	3rd Anal. Per.
1^st hour	SBP	0.7	-1.2	-0.8	5.0	5.6	5.2	1019	1287	1107
	MAP	0.5	-0.9	-0.6	3.5	3.6	3.5	818	1232	1112
	DBP	-0.01	-1.2	-0.9	3.5	3.5	3.4	677	997	915
2^nd hour	SBP	0.3	0.2	-0.1	5.1	4.5	5.4	1064	1207	1183
	MAP	0.2	-0.2	-0.1	3.4	3.2	4.0	1135	1035	1041
	DBP	-0.3	-0.8	-0.6	3.1	3.5	4.0	1054	724	809
3^rd hour	SBP	-1.0	-0.01	-0.3	5.7	5.3	5.2	1113	766	826
	MAP	-0.1	0.2	-0.1	3.7	3.5	3.6	921	915	712
	DBP	-0.1	-0.2	-0.5	3.5	3.0	3.4	782	985	630
4^th hour	SBP	-0.6	-0.1	-0.1	6.4	5.2	4.8	1229	1065	938
	MAP	-0.1	-0.1	0.1	3.9	3.2	3.5	1274	995	699
	DBP	-0.3	-0.5	-0.2	4.1	2.9	3.6	1427	842	696
5^th hour	SBP	0.2	0.1	-0.01	6.0	6.0	5.3	1018	1118	1115
	MAP	-0.1	-0.1	0.01	3.4	3.7	3.4	935	1320	1192
	DBP	-0.6	-0.6	-0.5	2.8	4.1	3.1	1015	1323	1067
24^th hour	SBP	-0.9	0.6	0.6	6.3	5.8	5.4	825	954	985
	MAP	-0.8	0.1	0.3	4.6	4.0	3.5	693	842	830
	DBP	-1.3	-0.7	-0.4	4.5	3.9	3.4	566	645	602

Abbreviations: Anal. Per.: Analysis Period; BP: Blood pressure; CNIBP: Continuous non-invasive blood pressure; DBP: Diastolic blood pressure; IBP: Invasive blood pressure; MAP: Mean arterial pressure; Nind: the number of independent measurements; SBP: Systolic blood pressure; $s_{c o r r} :$ Corrected standard deviation; $\bar{x}$ : Mean error.

Assessment of tracking dynamic BP changes

To assess the capability of the CNIBP device to track dynamic BP changes, it was first confirmed that all the data met the requirement of at least 50 BP fluctuations. Additionally, the distribution of BP changes must ensure that a minimum of 30% of the variations lead to an increase in BP, while at least 30% correspond to a decrease in BP (The required absolute changes in SPB, DBP, and MAP were set at 15, 10, and 12 mmHg, respectively). The

median E_{p e r c e n t}

and the 85th percentile of

E_{p e r c e n t}

were then calculated. The statistical analysis showed that the

E_{p e r c e n t}

for the 50th and 85th percentiles for SBP, DBP, and MAP were successfully constrained within the predefined thresholds of 25% and 50%, respectively. (Table 4). These findings indicate that the CNIBP device exhibits robust dynamic tracking capability.

Table 4.

Agreement between continuous non-invasive blood pressure and invasive blood pressure for blood pressure changes assessment.

BP type	50th percentiles mean (%)	85th percentiles mean (%)	BP increase (%)	BP decrease (%)
SBP	18.7%	33.9%	56.4%	43.6%
MAP	12.8%	24.0%	55.6%	44.4%
DBP	18.0%	30.3%	52.1%	47.9%

Abbreviations: BP: Blood pressure; DBP: Diastolic blood pressure; MAP: Mean arterial pressure; SBP: Systolic blood pressure.

Discussion

This study is the first to investigate the PWTT-based BP device in routine clinical use across three critical dimensions: accuracy, stability, and BP change tracking capability, in strict adherence to ISO 81060-3:2022 guidelines. The results demonstrate that the CNIBP device achieves clinically acceptable agreement with IBP measurements (the gold standard). Both $\bar{x}$ and s _corr for SBP, DBP, and MAP remained within the ISO-defined tolerance thresholds (-6 mmHg ≤ $\bar{x}$ ≤ 6 mmHg, -10 mmHg ≤ s_corr ≤ 10 mmHg), supported by PCC exceeding 0.9. Bland-Altman plots further demonstrated clinically acceptable agreement between the methods. Stability assessments over 24 h revealed consistent performance, with $\bar{x}$ and s _corr for all BP types remaining within the ISO 81060-3:2022 limits across repeated measurements, confirming the device reliability for prolonged monitoring. Notably, the CNIBP device also demonstrated robust dynamic tracking capability, with a $median E_{p e r c e n t}$ ≤25% and 85th percentile of $E_{p e r c e n t}$ ≤50%, validating its ability to detect clinically significant changes within 120 s. The device effectively captured true BP fluctuations in real-world clinical settings and reduced the frequency of false alarms from minor fluctuations. These findings confirm that the accuracy, stability, and BP change-tracking capabilities of the CNIBP device meet international standards and stringent requirements for data reliability in clinical use.

The wrist-worn CNIBP device evaluated in this study is user-friendly, transmitting data in real time via Bluetooth to the bedside monitor without interfering with nursing procedures, such as patient turning or suctioning. Nurses reported that the device was convenient to use. Calibration is required every 24 h, and recalibration must occur within 5 min following adjustments to vasoactive medications. On average, one to two planned calibrations were performed per case, each taking approximately 30 sec, resulting in minimal disruption to the clinical workflow. However, frequent calibration during periods of rapid hemodynamic change may increase the nursing workload and may require further optimization. The device is reuseable, offering long-term cost advantages compared to single-use arterial catheters. Proficiency in device operation can be achieved after 30 min of standardized training, with no additional personnel required.

Compared to conventional methods, PWTT-based CNIBP offers several advantages over traditional oscillometric cuff-based NIBP techniques. These advantages include continuous 24 h monitoring for dynamic BP tracking, which enhances clinical sensitivity and early warning capabilities. The real-time responsiveness of PWTT-based CNIBP enables detection of rapid BP changes with minimal latency. Compared to IBP methods, PWTT-based CNIBP reduces safety risks by eliminating those associated with arterial puncture, making it suitable for patients with vascular abnormalities or coagulation disorders. Additionally, its wearable design enhances patient comfort, reduces procedural complexity, and improves patient compliance.

Current NIBP measurement technologies offer distinct advantages and limitations. Sleeveless watch-style devices that utilize PPG and PWA²¹ provide convenient wearability, making them suitable for outpatient settings and long-term hypertension management. However, their accuracy is compromised by limb movement and hypoperfusion, and these devices often fail to capture rapid BP fluctuations in real time, limiting their applicability in hemodynamically unstable intensive care environments. Finger-ring devices employing pulse decomposition analysis (PDA)²² deliver beat-by-beat BP waveforms with high temporal resolution and have been validated in ICU patients. However, the finger-ring design can cause discomfort and venous congestion, limiting prolonged use. Additionally, the validation of these devices was performed according to earlier intermittent BP monitoring standards (ANSI/AAMI/ISO 81060-2:2013), which are less rigorous than the latest standards for continuous monitoring, particularly regarding dynamic BP tracking and long-term stability.

Several CNIBP monitoring devices utilizing PWTT technology have been reported. For example, the chest-adhesive device developed by John Smith et al.²³ integrates the monitoring of multiple physiological parameters, including SpO₂, heart rate (HR), BP, RR, body temperature, and hemodynamics. However, this device supports only the “spot-check” mode and does not enable high-frequency continuous BP monitoring, thus limiting its application in the real-time management of critical emergencies. Moreover, Hellou et al.²⁴ applied similar technology to post-cardiac surgery patients, but their validation focused primarily on accuracy and did not systematically assess long-term stability or dynamic tracking in accordance with ISO 81060-3:2022. Similarly, the cuffless PWTT device examined by Krisai et al.²⁵ demonstrated satisfactory short-term accuracy and performed effectively in both outpatient and hypertensive populations. However, it showed systematic overestimation during 24 h monitoring, particularly at night, with deviations of up to 15–18 mmHg compared to reference measurements, limiting its reliability for full-day monitoring.

In contrast, the PWTT-based mWear wrist-worn device evaluated in this study integrates ECG and PPG signals, thereby providing enhanced physiological validity and plausibility, and greater resistance to signal interference. The device enables continuous monitoring at intervals of at least 3 sec. Its wrist-worn design optimizes patient comfort and supports long-term wearability, extending recalibration intervals to 24 h. This study complies with ISO 81060-3:2022 by systematically validating measurement accuracy, long-term stability, and dynamic BP tracking capability in hemodynamically complex ICU patients. These features build upon existing evidence and collectively advance cuffless continuous BP monitoring technology towards standardization and broader clinical applications. However, as with other cuffless technologies, recalibration is necessary after significant changes in vasoactive medication regimens or abrupt alterations in patient positioning. This requirement represents a common challenge in the ongoing development of CNIBP monitoring.

Rigorous statistical methods, implemented in accordance with ISO 81060-3:2022 (such as the calibrated test standard deviation s_corr, intraclass correlation coefficient I_cc, and the number of independent measurements N_ind), enhance the evaluation of device performance. These methods account for data independence, the effects of repeated measurements, and inter- and intra-individual variability. The use of more complex statistical models provides a more comprehensive assessment of errors. This approach ensures that the samples have sufficient statistical power and more accurately reflect the device performance under varying conditions. Consequently, these methodological and statistical advancements offer stronger evidence supporting the clinical use of CNIBP monitoring devices, including the device examined in this study.

This study demonstrated significant improvements in clinical performance assessment by employing a rigorous clinical trial design, thereby providing more reliable evidence for clinical evaluation and application. The validation of the device ability to accurately track dynamic BP changes in real-world ICU settings underscores its potential role to facilitate timely therapeutic interventions, particularly during vasopressor titration and early detection of hemodynamic instability. The device addresses the critical gap between traditional intermittent cuff measurements and IBP monitoring, overcoming the longstanding clinical challenge of reconciling continuous monitoring with non-invasive techniques. The results indicate that this device may be particularly beneficial in several key clinical scenarios. In this study, IBP calibration was used to ensure methodological rigor, whereas NIBP calibration is sufficient for practical clinical application. The device serves as a valuable alternative for patients with contraindications to arterial catheterization, such as peripheral vascular disease or coagulation disorders, in whom invasive monitoring poses a substantial risk. Additionally, it also provides an effective bridge for patients moving from intensive IBP monitoring to general ward care, offering continuous surveillance at a higher resolution than intermittent cuff measurements, without the associated risks of invasive monitoring. Furthermore, its extended hemodynamic profiling capability makes it ideal for cases requiring prolonged BP trend analysis, without the infection risks associated with long-term arterial lines, thereby expanding the options for continuous hemodynamic surveillance across various clinical settings.

Despite the progress achieved, several limitations remain in this study. Firstly, calibration dependency: the CNIBP device requires recalibration following adjustments with vasoactive drugs or changes in the patient’s posture.²⁶ This dependency limits its application in highly dynamic clinical settings, such as rapid titration of vasoactive medications, intra-aortic balloon pump (IABP) support, or arrhythmia management, where continuous monitoring is essential. Secondly, sample size considerations: although the sample size was adequate to validate the primary endpoints, such as accuracy, stability, and dynamic BP tracking, according to ISO standards, it was insufficient for predefined subgroup analyses (e.g., age ≥ 70 years, comorbidity classification, or varying doses of vasoactive drugs). Future large-scale multi-cohort studies are needed to thoroughly evaluate performance heterogeneity across different patient populations. Thirdly, the study was limited to ICU patients, and a short observation period: a 24 h monitoring period, combined with the exclusive focus on ICU patients, may limit the applicability of the findings to non-ICU settings, such as outpatient, perioperative, or step-down care, where continuous BP monitoring is also essential. Fourthly, there was no evaluation of long-term monitoring beyond 24 h: although the 24 h monitoring period meets the ISO 81060-3:2022 requirement for initial stability assessment, it does not capture potential signal drift or performance decline over extended periods. Future studies should include long-term longitudinal assessments (e.g., extending from 24 h to 7 days) to determine the device durability and calibration requirements in long-term monitoring scenarios. While CNIBP may ultimately complement oscillometric devices in selected clinical contexts, IBP monitoring remains essential for comprehensive hemodynamic assessment in critically ill patients, particularly when frequent blood sampling for metabolic parameters (such as pH and lactate) is required.

Conclusions

In conclusion, this trial demonstrated that the PWTT-based CNIBP device is a viable, ISO-compliant alternative for continuous BP monitoring in ICU settings. This approach bridges the critical gap between intermittent cuff-based methods and invasive techniques. Future research should focus on longitudinal, multi-center validation and explore hybrid monitoring systems that combine the non-invasive benefits of CNIBP with the selective use of IBP in high-acuity cases.

Supplemental material

Supplemental material - Clinical performance evaluation of a cuffless continuous non-invasive blood pressure monitoring device in ICU patients

Supplemental material for Clinical performance evaluation of a cuffless continuous non-invasive blood pressure monitoring device in ICU patients by Yuting Chen, Zhikun Xu, Yonghui Ma, Huaisheng Chen, Hongbo Xu, Yuan Ye, Fang Xue, Haigang Zhang, Shengyuan Su and Xueyan Liu in Digital Health.

Supplemental material

Supplemental material - Clinical performance evaluation of a cuffless continuous non-invasive blood pressure monitoring device in ICU patients

Footnotes

Acknowledgments

The clinical trial-related device was sponsored by Shenzhen Mindray Bio-Medical Electronics Co., Ltd.Com.

ORCID iD

Yuting Chen

Ethical considerations

This study was conducted with the understanding and the consent of each participant. The study protocol complied with the Declaration of Helsinki and was approved by the local ethics committee (approval number: Shenzhen People’s Hospital, SYL-2023-083-01). Shenzhen Bao’an District People’s Hospital, QX-2023-4. Huazhong University of Science and Technology Union Shenzhen Hospital, I-2023-09-01), registered (ChiCTR2300079037), and monitored externally.

Consent to participate

Informed consent was obtained from all subjects or their legal guardians. All patients provided written informed consent.

Author Contributions

Conception and design: Yuting Chen and Xueyan Liu. Collection and assembly of data: Zhikun Xu, Yonghui Ma, Hongbo Xu, Yuan Ye, Fang Xue, and Shengyuan Su. Data analysis and interpretation: Huaisheng Chen, Shengyuan Su and Haigang Zhang. Manuscript writing: Yuting Chen and Xueyan Liu. Final approval of the manuscript: All authors. Accountability for all aspects of the work: All authors.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by Shenzhen Key Laboratory of Prevention and Treatment of Severe Infections (ZDSYS20200811142804014) and Major Science and Technology Projects of the Shenzhen Science and Technology Innovation Commission (KCXFZ20230731093559005).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request. Supplementary materials may also be shared to support transparency and reproducibility, subject to institutional review board approval and compliance with patient consent agreements. All shared data will be de-identified to protect participant privacy.*

Supplemental material

Supplemental material for this article is available online.

Appendix

References

Yoon

Jeanselme

Dubrawski

, et al. Prediction of hypotension events with physiologic vital sign signatures in the intensive care unit. Crit Care 2020; 24(1): 661. https://doi.org/10.1186/s13054-020-03379-3

Seamon

Feather

Smith

, et al. Just one drop: the significance of a single hypotensive blood pressure reading during trauma resuscitations. J Trauma 2010; 68(6): 1289–1294. discussion 1294-1295. https://doi.org/10.1097/TA.0b013e3181db05dc

Borg

Trapani

. The accuracy of radial artery applanation tonometry and intra-arterial blood pressure monitoring in critically ill patients: An evidence-based review. Nurs Crit Care 2024; 29(6): 1751–1757. https://doi.org/10.1111/nicc.13006

Sonti

Fong

Hummel

, et al. Invasive blood pressure monitoring does not improve survival in septic critically ill patients. Am J Emerg Med 2025; 94: 173–178. https://doi.org/10.1016/j.ajem.2025.04.060

Michard

. Towards the automatic detection and correction of abnormal arterial pressure waveforms. J Clin Monit Comput 2024; 38(4): 749–752. https://doi.org/10.1007/s10877-024-01152-3

Russell

Joel

Hudson

, et al. Prospective evaluation of radial and femoral artery catheterization sites in critically ill adults. Crit Care Med 1983; 11(12): 936–939. https://doi.org/10.1097/00003246-198312000-00007

Baranger

Villemain

Goudot

, et al. The fundamental mechanisms of the Korotkoff sounds generation. Sci Adv 2023; 9(40): eadi4252. https://doi.org/10.1126/sciadv.adi4252

Sharman

Tan

Stergiou

, et al. Automated 'oscillometric' blood pressure measuring devices: how they work and what they measure. J Hum Hypertens 2023; 37(2): 93–100. https://doi.org/10.1038/s41371-022-00693-x

Meiyan

Zheng

Ying

, et al. Influence of cuff size on the accuracy of supine blood pressure measurement. Blood Press Monit 2025; 30(4): 169–174. https://doi.org/10.1097/MBP.0000000000000752

10.

Bilo

Sala

Perego

, et al.

Impact of cuff positioning on blood pressure measurement accuracy: may a specially designed cuff make a difference?

Hypertens Res 2017; 40(6): 573–580. https://doi.org/10.1038/hr.2016.184

11.

Netea

Lenders

Smits

, et al. Both body and arm position significantly influence blood pressure measurement. J Hum Hypertens 2003; 17(7): 459–462. https://doi.org/10.1038/sj.jhh.1001573

12.

Ren

Zhao

, et al. Assessing arterial stiffness using characteristics of Korotkoff sounds. Front Cardiovasc Med 2026; 13: 1654162. https://doi.org/10.3389/fcvm.2026.1654162

13.

Kouz

Wegge

Flick

, et al. Continuous intra-arterial versus intermittent oscillometric arterial pressure monitoring and hypotension during induction of anaesthesia: the AWAKE randomised trial. Br J Anaesth 2022; 129(4): 478–486. https://doi.org/10.1016/j.bja.2022.06.027

14.

Pressman

Newgard

. A TRANSDUCER FOR THE CONTINUOUS EXTERNAL MEASUREMENT OF ARTERIAL BLOOD PRESSURE. IEEE Trans Biomed Eng 1963; 10: 73–81. https://doi.org/10.1109/tbmel.1963.4322794

15.

Penaz

. Photoelectric measurement of blood pressure, volume and flow in the finger. Digest of the 10th International Conference on Medical and Biological Engineering. Dresden, 1973.

16.

Choi

Hourlier-Fargette

, et al. Relation between blood pressure and pulse wave velocity for human arteries. Proc Natl Acad Sci USA 2018; 115(44): 11144–11149. https://doi.org/10.1073/pnas.1814392115

17.

Pielmus

Mühlstef

Bresch

, et al. Surrogate based continuous noninvasive blood pressure measurement. Biomed Tech (Berl) 2021; 66(3): 231–245. https://doi.org/10.1515/bmt-2020-0209

18.

Derendinger

Vischer

Krisai

, et al. Ability of a 24-h ambulatory cuffless blood pressure monitoring device to track blood pressure changes in clinical practice. J Hypertens 2024; 42(4): 662–671. https://doi.org/10.1097/HJH.0000000000003667

19.

Bhatt

Amir

Jones

, et al. Validation of a popular consumer-grade cuffless blood pressure device for continuous 24 h monitoring. Eur Heart J Digit Health 2025; 6(4): 704–712. https://doi.org/10.1093/ehjdh/ztaf044

20.

ISO 81060:-3:2022, Non-invasive sphygmomanometers-Part 3: Clinical investigation of continuous automated measurement type, International Organization for Standardization, 2022.

21.

Almeida

Cortés

Perruchoud

, et al. Aktiia cuffless blood pressure monitor yields equivalent daytime blood pressure measurements compared to a 24-h ambulatory blood pressure monitor: Preliminary results from a prospective single-center study. Hypertens Res 2023; 46(6): 1456–1461. https://doi.org/10.1038/s41440-023-01258-2

22.

Kwon

Stafford

Enfield

, et al. Continuous Noninvasive Blood Pressure Monitoring of Beat-By-Beat Blood Pressure and Heart Rate Using Caretaker Compared With Invasive Arterial Catheter in the Intensive Care Unit. J Cardiothorac Vasc Anesth 2022; 36(7): 2012–2021. https://doi.org/10.1053/j.jvca.2021.09.042

23.

FDA 510(k) Summary Biobeat Platform-2 and BB-613WP Patch. https://www.accessdata.fda.gov/cdrh_docs/pdf22/K222010.pdf. Accessed 07/28/2023.

24.

Hellou

Jamal

Zuroff

, et al. Performance of a cuffless photoplethysmography-based device for continuous monitoring of blood pressure after cardiac surgery: a preliminary validation study. J Hum Hypertens 2025; 39(12): 894–902. https://doi.org/10.1038/s41371-025-01082-w

25.

Krisai

Vischer

Kilian

, et al. Accuracy of 24-hour ambulatory blood pressure monitoring by a novel cuffless device in clinical practice. Heart 2019; 105(5): 399–405. https://doi.org/10.1136/heartjnl-2018-313592

26.

Chen

Kobayashi

Ichikawa

, et al. Continuous estimation of systolic blood pressure using the pulse arrival time and intermittent calibration. Med Biol Eng Comput 2000; 38(5): 569–574. https://doi.org/10.1007/BF02345755

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

1.44 MB

0.58 MB

0.00 MB