Medical innovation can be viewed through two lenses: high-cost technological breakthroughs or accessibility-driven solutions. Although the former grabs headlines, the latter addresses the economic barriers that keep modern tools out of reach. Improving access might be less trendy, but it is often the more transformative way to elevate the quality of care. Miniaturised ultrasound devices improve access to echocardiographic evaluations, and machine learning algorithms, now built into modern pocket devices, have the potential to level the playing field, ensuring high diagnostic and measurement accuracy regardless of the operator’s experience. Integrating pulse contour algorithms into standard bedside monitors could democratise access to cardiac output monitoring for high-risk surgical patients without the need for additional, expensive equipment. In the future, continuous blood pressure monitoring could be possible by analysing pulse oximeter waveforms. Because pulse oximeters are already used for every surgical and ICU patient, this approach could become a seamless, invisible layer of added safety. By prioritising integration and affordability, the medical community could bridge the equity gap and ensure that every patient, regardless of geographic location or economic status, benefits from modern haemodynamic insights.

Several artificial intelligence (AI)-driven tools have emerged for the hemodynamic evaluation of critically ill and surgical patients. This article provides an overview of current developments and potential clinical applications of machine learning (ML) for blood pressure measurements, hypotension prediction, hemodynamic profiling, and echocardiography. ML algorithms have shown promise in enabling continuous, non-invasive blood pressure monitoring by analyzing pulse oximetry waveforms, though these methods require periodic calibration with traditional oscillometric brachial cuffs. Additionally, a variety of ML models have been trained to forecast impending hypotension. However, clinical research indicates that these algorithms often primarily rely on mean arterial pressure, leading to questions about their added predictive value. The issue of false-positive alerts is also significant and can result in unwarranted clinical interventions. In terms of hemodynamic profiling, ML algorithms have been proposed to automatically classify patients into specific hemodynamic endotypes. However, current evidence suggests these models tend to replicate conventional hemodynamic profiles found in medical textbooks or depicted on advanced hemodynamic monitors. This raises questions about their practical clinical utility, especially given occasional discrepancies that could impact treatment decisions. Point-of-care ultrasound (POCUS) has gained traction for evaluating cardiac function in patients experiencing circulatory shock. ML algorithms now embedded in some POCUS systems can assist by recognizing ultrasound images, guiding users for optimal imaging, automating and reducing the variability of key echocardiographic measurements. These capabilities are especially beneficial for novice operators, potentially enhancing accuracy and confidence in clinical decision-making. In conclusion, while several AI-based technologies show promise for refining hemodynamic assessment in both critically ill and surgical patients, their clinical value varies. Comprehensive validation studies and real-world testing are essential to identify which innovations will genuinely contribute to improving the quality of care.

Machine learning (ML) algorithms hold significant potential for extracting valuable clinical information from big data, surpassing the processing capabilities of the human brain. However, it would be naïve to believe that ML algorithms can consistently transform data into actionable insights. Clinical studies suggest that in some instances, they tell clinicians what they already know or can plainly see. Additionally, ML algorithms might not be necessary for analysing ‘small data’, such as a limited number of haemodynamic variables. In this respect, whether haemodynamic profiling with an ML algorithm offers advantages over straightforward classification tables or simple visual decision support tools remains unclear.

The advancements in cardiovascular imaging over the past two decades have been significant. The miniaturization of ultrasound devices has greatly contributed to their widespread adoption in operating rooms and intensive care units. The integration of AI-enabled tools has further transformed the field by simplifying echocardiographic evaluations and enhancing the reproducibility of hemodynamic measurements, even for less experienced operators. Speckle tracking echocardiography offers a direct, visual, and quantitative assessment of myocardial shortening, serving as a compelling alternative to traditional methods for evaluating right and left ventricular systolic function. In critically ill patients, sublingual microcirculation imaging has revealed a high prevalence of microvascular alterations, which are markers of disease severity. The use of handheld vital microscopes enables the quantification of several key parameters, including vessel density, perfusion, red blood cell velocity, and the perfused vascular density. Such metrics are useful for evaluating microcirculatory health. The development of automated software marks a significant advance toward real-time bedside microvascular assessment. These advancements could eventually allow shock resuscitation to be tailored based on microvascular responses. In parallel with imaging advances, cardiac output monitors have evolved significantly. Once cumbersome devices displaying basic numerical data in tabular form, they now feature sleek, touch-screen interfaces integrated with visual decision-support tools. These tools synthesize hemodynamic data into intuitive graphical formats, allowing clinicians to quickly grasp the determinants of circulatory shock. This visual clarity supports more efficient and accurate decision-making, which may ultimately lead to improved patient care and outcomes.

The concept of integrating hemodynamic variables to define specific profiles or phenotypes has been established for decades. Describing hemodynamic phenotypes plays a key role in educating healthcare professionals about cardiovascular physiology, enhancing the understanding of shock mechanisms, and informing treatment strategies. Recently, two notable innovations have emerged to support bedside identification of hemodynamic phenotypes: machine learning (ML) algorithms and visual decision support tools. When it comes to "small data," such as a limited set of hemodynamic variables, ML algorithms may not be essential for data integration or interpretation. In addition, the hemodynamic phenotypes identified by ML techniques often mirror traditional textbook profiles, though occasionally with inconsistencies that may impact patient safety. This raises valid questions about the need to integrate complex and proprietary ML algorithms for bedside hemodynamic assessment. By contrast, visual tools leverage clinicians' innate ability to process graphical information rapidly, improving the understanding of cardiovascular physiology and enabling recognition of hemodynamic profiles at a glance. As such, they may offer a practical, accessible, and cost-effective alternative to ML-based solutions. Future studies comparing the clinical impact of visual versus ML-driven phenotyping are now needed to guide further development and implementation.

Background Approximately 75% of the world’s population lives in middle-income countries (MICs), where access to haemodynamic evaluation tools may be limited, exacerbating global healthcare disparities. Methods We conducted an online survey of anaesthetists and intensivists working in MICs, inviting them to complete 15 questions on bedside haemodynamic evaluations and access to haemodynamic monitoring tools. Results We analysed 1593 valid questionnaires from 20 Upper and 19 Lower MICs. Most respondents (66%) worked in academic hospitals, 43% in private hospitals, and 20% in non-academic public hospitals. Respondents worked in ICUs (39%), operating rooms (38%), or both (23%). Nearly all had access to central venous catheters (99%) and invasive radial arterial pressure monitoring (91%). Fewer than two-thirds (63%) reported access to echocardiography, and only 37% had access to cardiac output monitoring systems when needed. The main barriers were the cost of monitors (54%) and the cost of disposable sensors (52%). Notably, 72% indicated they would use cardiac output monitoring equipment more frequently if costs were reduced. Most respondents (89%) reported a routine practice of predicting fluid responsiveness before giving a fluid bolus, most commonly with pulse pressure variation (64%) or ultrasound indices (55%). Tissue perfusion was mainly assessed by clinical evaluation (86%), blood lactate (81%), and capillary refill time (63%). Conclusions In MICs, less than two-thirds of anaesthetists and intensivists reported having access to echocardiography for haemodynamic assessment. Fewer than 40% have access to cardiac output monitoring systems, mainly attributable to economic constraints. As this report represents a potential concerning equity gap in global healthcare, efforts should be made to prioritise and facilitate access to haemodynamic evaluation tools in MICs.

Both over and underdamping of the arterial pressure waveform are frequent during continuous invasive radial pressure monitoring. They may influence systolic blood pressure measurements and the accuracy of cardiac output monitoring with pulse wave analysis techniques. It is therefore recommended to regularly perform fast flush tests to unmask abnormal damping. Smart algorithms have recently been developed for the automatic detection of abnormal damping. In case of overdamping, air bubbles, kinking, and partial obstruction of the arterial catheter should be suspected and eliminated. In the case of underdamping, resonance filters may be necessary to normalize the arterial pressure waveform and ensure accurate hemodynamic measurements.

The early detection of clinical deterioration could be the next significant step in enhancing patient safety in general hospital wards. Most patients do not deteriorate suddenly; instead, their vital signs are often abnormal or trending towards an abnormal range hours before severe adverse events requiring rescue intervention and/or ICU transfer. To date, at least 10 large clinical studies have demonstrated a significant reduction in severe adverse events when heart rate, blood pressure, oxygen saturation and/or respiratory rate are continuously monitored on medical and surgical wards. Continuous, silent, and automatic monitoring of vital signs also presents the opportunity to eliminate unnecessary spot-checks for stable patients. This could lead to a reduction in nurse workload, while significantly improving patient comfort, sleep quality, and overall satisfaction. Wireless and wearable sensors are particularly valuable, as they make continuous monitoring feasible even for ambulatory patients, raising questions about the future relevance of “stay-in-bed” solutions like capnography, bed sensors, and video-monitoring systems. While the number of wearable sensors and mobile monitoring solutions is rapidly growing, independent validation studies on their sensitivity and specificity in detecting abnormal vital signs in actual patients, rather than healthy volunteers, remain limited. Additionally, further research is needed to evaluate the cost-effectiveness of using wireless wearables for vital sign monitoring both within hospital wards and at home.

During surgery, various haemodynamic variables are monitored and optimised to maintain organ perfusion pressure and oxygen delivery – and to eventually improve outcomes. Important haemodynamic variables that provide an understanding of most pathophysiologic haemodynamic conditions during surgery include heart rate, arterial pressure, central venous pressure, pulse pressure variation/stroke volume variation, stroke volume, and cardiac output. A basic physiologic and pathophysiologic understanding of these haemodynamic variables and the corresponding monitoring methods is essential. We therefore revisit the pathophysiologic rationale for intraoperative monitoring of haemodynamic variables, describe the history, current use, and future technological developments of monitoring methods, and finally briefly summarise the evidence that haemodynamic management can improve patient-centred outcomes.

Anaesthesiologists overwhelmingly favour pulse wave analysis techniques as their primary method to monitor cardiac output during high-risk noncardiac surgery. In patients with a radial arterial catheter in place, pulse wave analysis techniques have the advantage of instantly providing non-operator-dependent and continuous haemodynamic monitoring information. Green pulse wave analysis techniques working with any standard pressure transducer are as reliable as techniques requiring dedicated pressure transducers. They have the advantage of minimising plastic waste and related carbon dioxide emissions, and also significantly reducing hospital costs. The future integration of pulse wave analysis algorithms into multivariable bedside monitors, obviating the need for standalone haemodynamic monitors, could lead to wider use of haemodynamic monitoring solutions by further reducing their cost and carbon footprint.

Retrospective observational studies have reported a significant association between intraoperative hypotension and postoperative morbidity. However, association does not imply causation, and whether preventing intraoperative hypotension can improve patient outcome remains to be demonstrated. In this issue of the British Journal of Anaesthesia, D'Amico and colleagues meta-analysed 10 prospective randomised trials comparing low (≤60 mm Hg) and higher mean arterial pressure targets during anaesthesia and surgery. They did not observe an increase in postoperative morbidity and mortality in the low target group. In contrast, they reported a statistically significant (but not clinically relevant) reduction in postoperative cardiac arrhythmia and hospital length of stay when targeting mean arterial pressure ≤60 mm Hg. These findings suggest that during most surgical cases, intraoperative hypotension is a marker of the severity, frailty, or both rather than a mediator of postoperative complications.

The analysis of arterial pressure waveforms with machine learning algorithms has been proposed to predict intraoperative hypotension. The ability to forecast arterial hypotension 5-15 min ahead of the fall in blood pressure allows clinicians to be pro-active instead of reactive, and could potentially decrease postoperative morbidity. However, the predictive value of machine learning algorithms has been overestimated due to selection bias in several clinical studies, and they might not be superior to mere observation of arterial pressure. Continuous blood pressure monitoring enables immediate detection of hypotension, and giving fluid, vasopressors or inotropes to patients who are not yet (and might never become) hypotensive based on an algorithm is questionable. Finally, recent prospective interventional studies suggest that reducing intraoperative hypotension does not improve postoperative outcomes.

Purpose In patients with a radial arterial catheter, underdamping of the pressure signal is common and responsible for an overestimation of systolic arterial pressure (SAP). The maximum rate of the arterial pressure rise during systole (dP/dt MAX ) has been proposed to assess left ventricular systolic function. The impact of underdamping on dP/dt MAX is likely but has never been quantified. Methods We analyzed data from 70 critically ill patients monitored with a radial catheter in whom underdamping of the arterial pressure waveform was confirmed by the Gardner’s method. Invasive SAP and dP/dt MAX were recorded at baseline and after the correction of underdamping with a resonance filter. Results With resonance filtering, SAP decreased from 159 ± 25 to 139 ± 22 mmHg ( p < 0.001) and dP/dt MAX from 2.0 ± 0.6 to 1.1 ± 0.3 mmHg/ms ( p < 0.001). The underdamping-induced overestimation of SAP (delta-SAP) ranged from 6 to 41 mmHg (mean 21 ± 9 mmHg or + 15%) and the overestimation of dP/dt MAX (delta-dP/dt MAX ) ranged from 0.2 to 2.0 mmHg/ms (mean 0.9 ± 0.4 mmHg/ms or + 84%). A significant linear relationship ( p < 0.001, r = 0.6) was observed between delta-SAP and delta-dP/dt MAX such that the higher was delta-SAP, the higher was delta-dP/dt MAX . Conclusions Radial arterial pressure underdamping has a major impact on dP/dt MAX . In case of underdamping, the overestimation of dP/dt MAX is > fivefold higher than SAP overestimation. Therefore, caution should be exercised before using radial dP/dt MAX as a marker of left ventricular systolic function. Trial registration Registered at ClinicalTrials.gov on December 22, 2021 (NCT05166993). Graphical Abstract

Editor—The portability and cost of ultrasound devices, and operator skills, remain significant barriers to the adoption of point-of-care echocardiography. 1 Vieillard-Baron A. Millington S.J. Sanfilippo F. et al. A decade of progress in critical care echocardiography: a narrative review. Intensive Care Med. 2019; 45: 770-788 Crossref PubMed Scopus (125) Google Scholar , 2 Marbach J.A. Almufleh A. Di Santo P. et al. A shifting paradigm: the role of focused cardiac ultrasound in bedside patient assessment. Chest. 2020; 158: 2107-2118 Abstract Full Text Full Text PDF PubMed Scopus (18) Google Scholar , 3 Mayo P.H. Chew M. Douflé M. et al. Machines that save lives in the intensive care unit: the ultrasonography machine. Intensive Care Med. 2022; 48: 1429-1438 Crossref PubMed Scopus (6) Google Scholar Recent hardware and software ultrasound innovations include crystal-free pocket probes and algorithms designed to facilitate and automate echocardiographic measurements, and connectivity to smartphones. 3 Mayo P.H. Chew M. Douflé M. et al. Machines that save lives in the intensive care unit: the ultrasonography machine. Intensive Care Med. 2022; 48: 1429-1438 Crossref PubMed Scopus (6) Google Scholar ,4 Le M.P.T. Voigt L. Nathanson R. et al. Comparison of four handheld point-of-care ultrasound devices by expert users. Ultrasound J. 2022; 14: 27 Crossref PubMed Scopus (18) Google Scholar We designed the present study to compare automatic measurements of left ventricular ejection fraction (LVEF) taken with a silicon chip ultrasound probe and a smartphone (LVEFSMART) to reference manual measurements (LVEFREF) taken with a high-end ultrasound device.

Background: Continuous and wireless vital sign monitoring is superior to intermittent monitoring in detecting vital sign abnormalities; however, the impact on clinical outcomes has not been established. Methods: We performed a propensity-matched analysis of data describing patients admitted to general surgical wards between January 2018 and December 2019 at a single, tertiary medical centre in the USA. The primary outcome was a composite of in-hospital mortality or ICU transfer during hospitalisation. Secondary outcomes were the odds of individual components of the primary outcome, and heart failure, myocardial infarction, acute kidney injury, and rapid response team activations. Data are presented as odds ratios (ORs) with 95% confidence intervals (CIs) and n (%). Results: We initially screened a population of 34,636 patients (mean age 58.3 (Range 18-101) yr, 16,456 (47.5%) women. After propensity matching, intermittent monitoring (n=12 345) was associated with increased risk of a composite of mortality or ICU admission (OR 3.42, 95% CI 3.19-3.67; P<0.001), and heart failure (OR 1.48, 95% CI 1.21-1.81; P<0.001), myocardial infarction (OR 3.87, 95% CI 2.71-5.71; P<0.001), and acute kidney injury (OR 1.32, 95% CI 1.09-1.57; P<0.001) compared with continuous wireless monitoring (n=7955). The odds of rapid response team intervention were similar in both groups (OR 0.86, 95% CI 0.79-1.06; P=0.726). Conclusions: Patients who received continuous ward monitoring were less likely to die or be admitted to ICU than those who received intermittent monitoring. These findings should be confirmed in prospective randomised trials.

Michard, Frederic; Biais, Matthieu; Futier, Emmanuel; Romagnoli, Stefano Author Information