Journal of Clinical Monitoring and Computing最新文献

Purpose: Traditional assessments using carboxyhemoglobin (COHb) levels alone often do not adequately predict clinical course of carbon monoxide (CO) poisoning cases. Perfusion index (PI) and pleth variability index (PVI) offer non-invasive, continuous monitoring of peripheral perfusion, potentially improving patient management. The objective of this study is to evaluate whether perfusion indices can assist in triage and monitoring of patients with CO poisoning.

Methods: All patients aged 18 years and older, diagnosed with CO poisoning were consecutively enrolled in this prospective observational study from January 2019 to May 2023. Perfusion indices, COHb and lactate levels were measured at diagnosis (values denoted by 1) and after 60-min hyperbaric or normobaric oxygen therapy (HBOT or NBOT) (values denoted by 2).

Results: PI-1 showed significant moderate negative correlation with COHb-1 levels in all patients and AUC value of PI-1 in predicting the necessity for HBOT was 0.935. Patients requiring HBOT had significantly lower PI-1 and higher COHb-1, lactate-1, and PVI-1 compared to those receiving NBOT. Following treatment, PI increased, and PVI, lactate, and COHb decreased significantly in both treatment groups (p<0.001 for all).

Conclusions:  Perfusion indices, especially PI, may reflect changes in COHb levels and could provide additional information to support triage and monitoring in CO poisoning.

The Analgesia Nociception Index (ANI) is based on respiratory sinus arrhythmia and is a validated surrogate marker of the nociception-antinociception balance. Along with the ANI, the monitor provides a measure of overall heart rate variability modulation named "Energy" and which is closely related to the standard deviation of normal R-R intervals. The objective of the present study was to evaluate variations in "Energy" during general anesthesia, sedation, and spinal anesthesia. We retrospectively analyzed data stored in the anesthesia data warehouse at Lille University Medical Center (Lille, France). Eligible cases involved general anesthesia, spinal anesthesia, or sedation over the period 2012-2024. Patients with arrhythmia or missing baseline data were excluded. Three periods were defined: pre-induction (P1), post-induction (P2), and intraoperative (P3). Linear mixed models were adjusted for age, the American Society of Anesthesiologists score, norepinephrine use, and sex. 2226 procedures were included. The decrease in "Energy" after induction was significantly greater for general anesthesia after adjustment between P1 and P2 (Mean (SD) -0.306 (-0.321; -0.292), p < 0.001) and between P1 and P3 (-0.334 (-0.348; -0.319), p < 0.001). Same results were found for sedation (P1-P2: -0.120 (-0.176; -0.064), p < 0.001; P1-P3: -0.113 (-0.168; -0.056), p < 0.001) and spinal anesthesia (P1-P2: 0.082 (0.017; 0.146), p = 0.012; P1-P3: 0.089 (0.025; 0.153), p = 0.006) after adjustment. Changes during sedation and spinal anesthesia were not clinically relevant. "Energy" decreases after the induction of general anesthesia and sedation and thus reflects a lower degree of autonomic modulation.

Vegetative reactions are common during neurosurgical procedures. Known effects are mainly cardiovascular, including tachy- and bradyarrhythmia, hyper- and hypotonia as well as cardiac arrest. Computer-assisted real-time analysis of heart rate variability (HRV), baroreflex-sensitivity (BRS) allows for continuous evaluation of the autonomic nervous system (ANS). We analyzed ANS parameters during intracranial neurosurgical procedures. In this pilot study, we aim to provide proof-of-concept that ANS monitoring during surgery is feasible and yields stable results.We included 129 consecutive patients undergoing neurosurgery for intracranial pathologies over a period of four months. Heart rate (HR) and mean arterial pressure (MAP) were continuously monitored during routine anesthesiology care. Data were recorded via ICM + software. HRV, BRS and other vegetative parameters were calculated continuously. Intraoperative events such as hypo-/hypertonia or brady-/tachycardia were monitored.Mean age was 47.2 ± 17.7 years. Of all patients, 54.3% were male (n = 70). For every patient, four intraoperative episodes were defined: start of anesthesia until incision - start of incision until craniotomy - craniotomy until end of resection or intracranial manipulation - end phase until skin closure. BRS continuously decreased during cranial surgery, indicating stabilized autonomic function. Furthermore, blood pressure variability was increased during semi-sitting surgery.Autonomic system monitoring during neurosurgical procedures is safe and feasible. Intraoperatively, an increasing sympathetic activity has been observed without clear disctinction between surgical or anesthesiological events as underlying cause. Monitoring results are reproducible and may be of importance for the detection and prevention of intraoperative cardiovascular events.

Purpose: Intensive care units (ICUs) handle mechanically ventilated patients with life-threatening conditions, who require intensive monitoring and treatment. In a low physician-patient ratio setting, providing consistent care to all patients is challenging. A survival prediction model using machine-learning can potentially improve prognosis evaluation and resource allocation. This study aims to develop a machine-learning model to predict survival/mortality in mechanically ventilated patients using clinical features recorded at the time of ICU admission and compare its performance with the Sequential Organ Failure Assessment (SOFA) score as a standalone predictor.

Methods: A dataset consisting of 660 mechanically ventilated patients and 98 clinical parameters (n = 660, Male: Female = 365:295, Age = 44.45 ± 19.36 years) from three ICUs at AIIMS, Delhi, was retrospectively evaluated after institutional ethical approval. Binary classification models were trained using 10-fold cross-validation with 70% data and 30% reserved for testing. The outcome was based on the survival/death of the patient during their ICU stay.

Results: A total of 39 features were selected using Shapley-Additive-Explanations (SHAP) and Random Forest model. The top three features were SOFA score, International normalized ratio (INR) and respiratory rate with feature importance values of 7.3%, 4.5% and 3.4% respectively. The K-nearest-neighbour (KNN) model using SHAP-selected features achieved the best test performance with an accuracy = 0.80, area-under-receiver-operating-characteristics-curve (AUROC) = 0.84, sensitivity = 0.82, specificity = 0.77, positive-predictive-value (PPV) = 0.78 and negative-predictive-value (NPV) = 0.82, compared to the SOFA-only model showing accuracy = 0.73, AUROC = 0.73, sensitivity = 0.82, specificity = 0.63, PPV = 0.69 and NPV = 0.78.

Conclusion: The automated machine-learning method for prognosis prediction may assist clinicians in the early triage of patients. These models may offer valuable support to ICU physicians for timely alerts and informed clinical judgment. The study also highlights the continued utility of the SOFA score used by clinicians as the first assessment tool in ICUs, while suggesting that carefully developed machine-learning models may offer complementary support in high-risk ICU settings.

Purpose: Stroke volume variation (SVV) is a dynamic parameter used to assess fluid responsiveness in mechanically ventilated patients. This study aimed to evaluate the agreement and trending ability of SVV measurements obtained from bioreactance (Starling SV) and arterial waveform analysis devices (FloTrac and LiDCOrapid) during cardiac surgery.

Methods: This prospective observational method comparison study was conducted in a single university hospital. 18 patients undergoing off-pump coronary artery bypass grafting (OPCAB) were monitored with Starling SV and FloTrac. 20 patients undergoing cardiac surgery with cardiopulmonary bypass (CPB) were monitored with Starling SV and LiDCOrapid. SVV measurements were collected intraoperatively and postoperatively. Agreement and trending ability between devices were assessed using Bland-Altman analysis and four-quadrant plots with error grids and concordance analysis.

Results: A total of 2055 paired SVV measurements were obtained in the OPCAB group and 367 in the CPB group. The mean bias between Starling SV and FloTrac was 2.3%pt (95% CI 2.1 to 2.6) with wide limits of agreement (-14.3 to 20.5%pt). For Starling SV and LiDCOrapid, the bias was 1.5%pt (95% CI 0.9 to 2.2) with very wide limits of agreement (-38.3 to 38.4%pt). Trending ability was poor in all comparisons.

Conclusion: Despite acceptable mean biases, the variability between devices was considerable, and trending analyses indicated only limited concordance. The studied SVV monitors, therefore, cannot be considered interchangeable in the context of cardiac surgery. These findings highlight the limitations and uncertainty of SVV monitoring in this setting.

Respiratory rate is an important early sign of clinical deterioration but the current practice of counting breaths manually is time-consuming and prone to error. We aimed to determine the concordance between manual respiratory rate measurements and automated measurements recorded using a wearable device. We undertook a prospective observational study on three general respiratory wards to compare manual respiratory rate measurements collected during usual clinical care with automated readings from a wearable respiratory rate monitor (RespiraSense, PMD Solutions, Cork, Ireland). Thirty-one patients took part in the study. Manual respiratory rate readings displayed large peaks at 20 and 24 breaths/min, whereas automated readings followed a smooth bell-shaped distribution. Manual and automated respiratory rates were both higher during the day than at night, and this was more marked for automated readings. Automated readings were on average 2.5 (95% confidence interval [CI] 2.2 to 2.8) breaths/minute higher than time-matched manual readings, and the 95% limits of agreement were - 7.9 (95% CI -8.4 to -7.4) and 12.9 (95% CI 12.3 to 13.4) breaths/minute, wider than the clinically acceptable limits of ± 3 breaths/min. Trends in manual and automated respiratory rates were concordant in only 56% of cases. Automated respiratory rate measurements using RespiraSense do not display clinically acceptable agreement with manual measurements in the setting of a respiratory ward.

Functional testers are designed to evaluate select pulse oximeter characteristics but are often misused to validate device accuracy, potentially providing false reassurance. This study evaluated whether the Fluke ProSim8 (FPS8) could accurately predict oximeter performance during human controlled desaturation studies or identify performance differences under low signal conditions. 12 oximeters were tested using two FPS8 protocols: (1) an 'SpO₂ plateau' protocol which mimicked controlled desaturation studies by evaluating device performance over a range of simulated SpO₂ (70-100%), and (2) a 'signal space' protocol designed to assess device accuracy under varying modulation and transmission conditions. Each device also underwent controlled desaturation testing in healthy adults. Six of the 12 oximeters passed (ARMS ≤ 3%) the SpO₂ plateau protocol; however, three of these failed (ARMS > 3%) human testing. At lower simulated saturations, most devices overestimated SpO₂. In the signal space protocol, oximeters performed well under high signal conditions, but many failed to produce readings or showed SpO₂ errors > 3% under low signal conditions. On average, oximeters failed to generate a reading 20.2 ± 7.2 times out of 60 attempts. Ten devices passed (ARMS < 3%) the signal space protocol, but two of these failed human testing. Oximeter performance on the FPS8 did not correlate with human performance (R² = 0.08 for the plateau protocol; R² = 0.01 for the signal space protocol). The FPS8 did not reliably predict oximeter accuracy in human desaturation studies or under low signal conditions; current functional tester protocols are limited in predicting real-world oximeter performance.

Background: Accurately identifying surgical patients who will have an increase in stroke volume following fluid administration remains challenging when utilizing noninvasive bedside methods. This study aims to compare the value of using ultrasound to measure changes in corrected carotid artery flow time (ΔFTc) with that of using invasive measurements of stroke volume variation (ΔSVV) for assessing volume responsiveness in patients under general anesthesia and mechanical ventilation.

Methods: A total of 91 patients undergoing elective abdominal surgery under general anesthesia were enrolled in this prospective observational study. Under general anesthesia and mechanical ventilation, the ΔFTc was measured using noninvasive bedside ultrasound, and the ΔSVV was measured using invasive hemodynamic monitoring, both before and after fluid administration. Fluid responders were defined as an increase in stroke volume of ≥ 10% after the fluid challenge.

Results: A total of 47 (54.0%) patients were fluid responders. The ΔFTc was 14.5 ± 8.3 ms for responders and 5.7 ± 4.9 ms for non-responders, while the ΔSVV was 3.3 ± 1.4% for responders and 1.7 ± 0.7% for non-responders. The areas under the receiver operating characteristic curves for ΔFTc and ΔSVV were 0.85 (95% CI 0.77-0.92; P < 0.05) and 0.84 (95% CI 0.75-0.93; P < 0.05), respectively. The optimal cutoff values were 7.03 ms for ΔFTc (sensitivity 91.5%, specificity 69.8%) and 2.85% for ΔSVV (sensitivity 95.6%, specificity 72.6%). A narrower gray zone for ΔFTc, ranging from 7 ms to 12 ms and covering 27 patients, was observed compared with that for ΔSVV, which ranged from 1% to 3% and covered 41 patients.

Conclusions: Both the ΔFTc and ΔSVV reliably assessed fluid responsiveness in patients undergoing general anesthesia and mechanical ventilation. Noninvasive bedside ultrasound measurement of ΔFTc can serve as an accurate assessment parameter, reflecting the presence of volume responsiveness following rapid fluid administration and showing greater clinical applicability.

Trial registration: www.chictr.org.cn (ChiCTR2500101114); registered 21 April 2025.