A passive leg raising (PLR) test is positive if the cardiac index (CI) increased by > 10%, but it requires a direct measurement of CI. On the oxygen saturation plethysmographic signal, the perfusion index (PI) is the ratio between the pulsatile and the non-pulsatile portions. We hypothesised that the changes in PI could predict a positive PLR test and thus preload responsiveness in a totally non-invasive way. In patients with acute circulatory failure, we measured PI (Radical-7) and CI (PiCCO2) before and during a PLR test and, if decided, before and after volume expansion (500-mL saline). Three patients were excluded because the plethysmography signal was absent and 3 other ones because it was unstable. Eventually, 72 patients were analysed. In 34 patients with a positive PLR test (increase in CI ≥ 10%), CI and PI increased during PLR by 21 ± 10% and 54 ± 53%, respectively. In the 38 patients with a negative PLR test, PI did not significantly change during PLR. In 26 patients in whom volume expansion was performed, CI and PI increased by 28 ± 14% and 53 ± 63%, respectively. The correlation between the PI and CI changes for all interventions was significant (r = 0.64, p < 0.001). During the PLR test, if PI increased by > 9%, a positive response of CI (≥ 10%) was diagnosed with a sensitivity of 91 (76–98%) and a specificity of 79 (63–90%) (area under the receiver operating characteristics curve 0.89 (0.80–0.95), p < 0.0001). An increase in PI during PLR by 9% accurately detects a positive response of the PLR test. ID RCB 2016-A00959-42. Registered 27 June 2016.
3 October 2024 in Anaesthesia Critical Care & Pain Medicine
Integrating machine learning (ML) into intensive care units (ICUs) can significantly enhance patient care and operational efficiency. ML algorithms can analyze vast amounts of data from electronic health records, physiological monitoring systems, and other medical devices, providing real-time insights and predictive analytics to assist clinicians in decision-making. ML has shown promising results in predictive modeling for patient outcomes, early detection of sepsis, optimizing ventilator settings, and resource allocation. For instance, predictive algorithms have demonstrated high accuracy in forecasting patient deterioration, enabling timely interventions and reducing mortality rates. Despite these advancements, challenges such as data heterogeneity, integration with existing clinical workflows, and the need for transparency and interpretability of ML models persist. The deployment of ML in ICUs also raises ethical and legal considerations regarding patient privacy and the potential for algorithmic biases. For clinicians interested in the early embracing of AI-driven changes in clinical practice, in this review, we discuss the challenges of integrating AI and ML tools in the ICU environment in several steps and issues: 1. Main categories of ML algorithms; 2. From data enabling to ML development; 3. Decision-support systems that will allow patient stratification, accelerating the foresight of adequate individual care; 4. Improving patient outcomes and healthcare efficiency, with positive society and research implications; 5. Risks and barriers to AI-ML application to the healthcare system, including transparency, privacy, and ethical concerns.
Myocardial dysfunction is common in patients admitted to the intensive care unit (ICU). Septic disease frequently results in cardiac dysfunction, and sepsis represents the most common cause of admission and death in the ICU. The association between left ventricular (LV) systolic dysfunction and mortality is not clear for critically ill patients. Conversely, LV diastolic dysfunction (DD) seems increasingly recognized as a factor associated with poor outcomes, not only in sepsis but also more generally in critically ill patients. Despite recent attempts to simplify the diagnosis and grading of DD, this remains relatively complex, with the need to use several echocardiographic parameters. Furthermore, the current guidelines have several intrinsic limitations when applied to the ICU setting. In this manuscript, we discuss the challenges in DD classification when applied to critically ill patients, the importance of left atrial pressure estimates for the management of patients in ICU, and whether the study of cardiac dysfunction spectrum during critical illness may benefit from the integration of left ventricular and left atrial strain data to improve diagnostic accuracy and implications for the treatment and prognosis.
Ultrasound (US) imaging is used in the diagnosis and monitoring of COVID-19 and breast cancer. The presence of Speckle Noise (SN) is a downside to its usage since it decreases lesion conspicuity. Filters can be used to remove SN, but they involve time-consuming computation and parameter tuning. Several researchers have been developing complex Deep Learning (DL) models (150,000–500,000 parameters) for the removal of simulated added SN, without focusing on the real-world application of removing naturally occurring SN from original US images. Here, a simpler (<30,000 parameters) Convolutional Neural Network Autoencoder (CNN-AE) to remove SN from US images of the breast and lung is proposed. In order to do so, simulated SN was added to such US images, considering four different noise levels ( = 0.05, 0.1, 0.2, 0.5). The original US images (N = 1227, breast + lung) were given as targets, while the noised US images served as the input. The Structural Similarity Index Measure (SSIM) and Peak Signal-to-Noise Ratio (PSNR) were used to compare the output of the CNN-AE and of the Median and Lee filters with the original US images. The CNN-AE outperformed the use of these classic filters for every noise level. To see how well the model removed naturally occurring SN from the original US images and to test its real-world applicability, a CNN model that differentiates malignant from benign breast lesions was developed. Several inputs were used to train the model (original, CNN-AE denoised, filter denoised, and noised US images). The use of the original US images resulted in the highest Matthews Correlation Coefficient (MCC) and accuracy values, while for sensitivity and negative predicted values, the CNN-AE-denoised US images (for higher values) achieved the best results. Our results demonstrate that the application of a simpler DL model for SN removal results in fewer misclassifications of malignant breast lesions in comparison to the use of original US images and the application of the Median filter. This shows that the use of a less-complex model and the focus on clinical practice applicability are relevant and should be considered in future studies.
Hyponatraemia is the most prevalent electrolyte disorder in the neurocritical care setting and is associated with a significant morbimortality. Cerebral salt wasting and inappropriate antidiuretic hormone secretion syndrome have been classically described as the two most frequent entities responsible for hyponatraemia in neurocritical care patients. An accurate aetiological diagnosis of hypotonic hyponatraemia requires a proper volume status assessment. Nevertheless, determination of volume status based on physical examination, laboratory findings and imaging modalities have several limitations and can lead to improperly diagnosis and hyponatraemia mismanagement. Point-of-care ultrasound (POCUS), specifically Venous Excess UltraSound (VExUS) score, is a fast and valuable tool to evaluate venous congestion at the bedside and identify hypervolaemia, helping the physicians in therapeutic decision making in a patient with hyponatraemia. We report a case where the use of POCUS, and more specifically VExUS, can be helpful in volume status assessment, complementing the complex management of multifactorial hyponatraemia in a neurocritical patient.