Multiparameter Intelligent Monitoring in Intensive Care II: a public-access intensive care unit database

Intelligent Bio-Impedance System for Personalized Continuous Blood Pressure Measurement

Continuous blood pressure (BP) measurement is crucial for long-term cardiovascular monitoring, especially for prompt hypertension detection. However, most of the continuous BP measurements rely on the pulse transit time (PTT) from multiple-channel physiological acquisition systems that impede wearable applications. Recently, wearable and smart health electronics have become significant for next-generation personalized healthcare progress. This study proposes an intelligent single-channel bio-impedance system for personalized BP monitoring. Compared to the PTT-based methods, the proposed sensing configuration greatly reduces the hardware complexity, which is beneficial for wearable applications. Most of all, the proposed system can extract the significant BP features hidden from the measured bio-impedance signals by an ultra-lightweight AI algorithm, implemented to further establish a tailored BP model for personalized healthcare. In the human trial, the proposed system demonstrates the BP accuracy in terms of the mean error (ME) and the mean absolute error (MAE) within 1.7 ± 3.4 mmHg and 2.7 ± 2.6 mmHg, respectively, which agrees with the criteria of the Association for the Advancement of Medical Instrumentation (AAMI). In conclusion, this work presents a proof-of-concept for an AI-based single-channel bio-impedance BP system. The new wearable smart system is expected to accelerate the artificial intelligence of things (AIoT) technology for personalized BP healthcare in the future.

Using High-Fidelity Avatars to Advance Camera-based Cardiac Pulse Measurement

Non-contact physiological measurement has the potential to provide low-cost, non-invasive health monitoring. However, machine vision approaches are often limited by the availability and diversity of annotated video datasets resulting in poor generalization to complex real-life conditions. To address these challenges, this work proposes the use of synthetic avatars that display facial blood flow changes and allow for systematic generation of samples under a wide variety of conditions. Our results show that training on both simulated and real video data can lead to performance gains under challenging conditions. We show strong performance on three large benchmark datasets and improved robustness to skin type and motion. These results highlight the promise of synthetic data for training camera-based pulse measurement; however, further research and validation is needed to establish whether synthetic data alone could be sufficient for training models.

Association between circadian variation of heart rate and mortality among critically ill patients: a retrospective cohort study

Background

Heart rate (HR) related parameters, such as HR variability, HR turbulence, resting HR, and nighttime mean HR have been recognized as independent predictors of mortality. However, the influence of circadian changes in HR on mortality remains unclear in intensive care units (ICU). The study is designed to evaluate the relationship between the circadian variation in HR and mortality risk among critically ill patients.

Methods

The present study included 4,760 patients extracted from the Multiparameter Intelligent Monitoring in Intensive Care II database. The nighttime mean HR/daytime mean HR ratio was adopted as the circadian variation in HR. According to the median value of the circadian variation in HR, participants were divided into two groups: group A (≤ 1) and group B (> 1). The outcomes included ICU, hospital, 30-day, and 1-year mortalities. The prognostic value of HR circadian variation was investigated by multivariable logistic regression models and Cox proportional hazards models.

Results

Patients in group B (n = 2,471) had higher mortality than those in group A (n = 2,289). Multivariable models revealed that the higher circadian variation in HR was associated with ICU mortality (odds ratio [OR], 1.393; 95% confidence interval [CI], 1.112–1.745; P = 0.004), hospital mortality (OR, 1.393; 95% CI, 1.112–1.745; P = 0.004), 30-day mortality (hazard ratio, 1.260; 95% CI, 1.064–1.491; P = 0.007), and 1-year mortality (hazard ratio, 1.207; 95% CI, 1.057–1.378; P = 0.005), especially in patients with higher SOFA scores.

Conclusions

The circadian variation in HR might aid in the early identification of critically ill patients at high risk of associated with ICU, hospital, 30-day, and 1-year mortalities.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12871-022-01586-9.

Hybrid Convolutional Networks for End-to-End Event Detection in Concurrent PPG and PCG Signals Affected by Motion Artifacts

The accurate detection of physiologically-related events in photopletismographic (PPG) and phocardiographic (PCG) signals, recorded by wearable sensors, is mandatory to perform the estimation of relevant cardiovascular parameters like the heart rate and the blood pressure. However, the measurement performed in uncontrolled conditions without clinical supervision leaves the detection quality particularly susceptible to noise and motion artifacts. The performed work proposed a new fully-automatic computational framework, based on convolutional networks, to identify and localize fiducial points in time as the foot, maximum slope and peak in PPG signal and the S1 sound in the PCG signal, both acquired by a custom chest sensor, described recently in the literature by our group. The novelty entailing a custom neural architecture to process sequentially the PPG and PCG signals. Tests were performed analysing four different acquisition conditions (rest, cycling, rest recovery and walking). Cross-validation results for the three PPG fiducial points showed identification accuracy greater than 93 % and localization error (RMSE) less than 10 ms. As expected, cycling and walking conditions provided worse results than rest and recovery, however reaching an accuracy greater than 90 % and a localization error lower than 15 ms. Likewise, the identification and localization error for S1 sound were greater than 90 % and lower than 25 ms. Overall, this study showcased the ability of the proposed technique to detect events with high accuracy not only for steady acquisitions but also during subject movements. We also showed that the proposed network outperformed traditional Shannon-energy-envelope method in the detection of S1 sound. Therefore, we argue that coupling chest sensors and deep learning processing techniques may disclose wearable devices to unobtrusively acquire health information, being less affected by noise and motion artifacts.

Hidden in Plain Sight: Clinical Informaticians are the Oncology Subspecialists You Did Not Know You Needed

Clinical informatics is a young, diverse and rapidly growing field. We asked eight clinical informaticians from a variety of oncology specialties, training pathways and careers for personal narratives to illustrate the wide spectrum of clinical informatics careers. Primary clinical specialties included radiation oncology, medical/haematology oncology and palliative care. Training pathways included fellowship, non-fellowship formal training and informal training. Careers included clinical care, research, operations and industry. We summarised common themes and advice for trainees. We hope to raise awareness of clinical informatics among trainees and oncologists to reveal new career opportunities and to avoid inadvertently taking clinical informatics and informaticians for granted.

CoCross: An ICT Platform Enabling Monitoring Recording and Fusion of Clinical Information Chest Sounds and Imaging of COVID-19 ICU Patients

Monitoring and treatment of severely ill COVID-19 patients in the ICU poses many challenges. The effort to understand the pathophysiology and progress of the disease requires high-quality annotated multi-parameter databases. We present CoCross, a platform that enables the monitoring and fusion of clinical information from in-ICU COVID-19 patients into an annotated database. CoCross consists of three components: (1) The CoCross4Pros native android application, a modular application, managing the interaction with portable medical devices, (2) the cloud-based data management services built-upon HL7 FHIR and ontologies, (3) the web-based application for intensivists, providing real-time review and analytics of the acquired measurements and auscultations. The platform has been successfully deployed since June 2020 in two ICUs in Greece resulting in a dynamic unified annotated database integrating clinical information with chest sounds and diagnostic imaging. Until today multisource data from 176 ICU patients were acquired and imported in the CoCross database, corresponding to a five-day average monitoring period including a dataset with 3477 distinct auscultations. The platform is well accepted and positively rated by the users regarding the overall experience.

Sepsis prediction, early detection, and identification using clinical text for machine learning: a systematic review

OBJECTIVE

To determine the effects of using unstructured clinical text in machine learning (ML) for prediction, early detection, and identification of sepsis.

MATERIALS AND METHODS

PubMed, Scopus, ACM DL, dblp, and IEEE Xplore databases were searched. Articles utilizing clinical text for ML or natural language processing (NLP) to detect, identify, recognize, diagnose, or predict the onset, development, progress, or prognosis of systemic inflammatory response syndrome, sepsis, severe sepsis, or septic shock were included. Sepsis definition, dataset, types of data, ML models, NLP techniques, and evaluation metrics were extracted.

RESULTS

The clinical text used in models include narrative notes written by nurses, physicians, and specialists in varying situations. This is often combined with common structured data such as demographics, vital signs, laboratory data, and medications. Area under the receiver operating characteristic curve (AUC) comparison of ML methods showed that utilizing both text and structured data predicts sepsis earlier and more accurately than structured data alone. No meta-analysis was performed because of incomparable measurements among the 9 included studies.

DISCUSSION

Studies focused on sepsis identification or early detection before onset; no studies used patient histories beyond the current episode of care to predict sepsis. Sepsis definition affects reporting methods, outcomes, and results. Many methods rely on continuous vital sign measurements in intensive care, making them not easily transferable to general ward units.

CONCLUSIONS

Approaches were heterogeneous, but studies showed that utilizing both unstructured text and structured data in ML can improve identification and early detection of sepsis.

A scoping review of artificial intelligence applications in thoracic surgery

OBJECTIVES

Machine learning (ML) has great potential, but there are few examples of its implementation improving outcomes. The thoracic surgeon must be aware of pertinent ML literature and how to evaluate this field for the safe translation to patient care. This scoping review provides an introduction to ML applications specific to the thoracic surgeon. We review current applications, limitations and future directions.

METHODS

A search of the PubMed database was conducted with inclusion requirements being the use of an ML algorithm to analyse patient information relevant to a thoracic surgeon and contain sufficient details on the data used, ML methods and results. Twenty-two papers met the criteria and were reviewed using a methodological quality rubric.

RESULTS

ML demonstrated enhanced preoperative test accuracy, earlier pathological diagnosis, therapies to maximize survival and predictions of adverse events and survival after surgery. However, only 4 performed external validation. One demonstrated improved patient outcomes, nearly all failed to perform model calibration and one addressed fairness and bias with most not generalizable to different populations. There was a considerable variation to allow for reproducibility.

CONCLUSIONS

There is promise but also challenges for ML in thoracic surgery. The transparency of data and algorithm design and the systemic bias on which models are dependent remain issues to be addressed. Although there has yet to be widespread use in thoracic surgery, it is essential thoracic surgeons be at the forefront of the eventual safe introduction of ML to the clinic and operating room.

Association of Red Blood Cell Distribution Width-Albumin Ratio for Acute Myocardial Infarction Patients with Mortality: A Retrospective Cohort Study

BACKGROUND

Red blood cell distribution width (RDW) was a risk factor for poor prognosis in acute myocardial infarction (AMI). Recent reports suggested that combining RDW with other laboratory metrics could provide a better prediction. This retrospective study aimed to investigate whether the RDW-albumin ratio (RAR) may be associated with mortality after an AMI.

METHODS

This cohort study was conducted among adults (over 16 years old) with AMI in the Medical Information Mart for Intensive Care Database III V1.4 (MIMIC-III). The primary outcome was 30-day mortality, and the secondary outcome was 1-year and 3-year mortality. Cox hazard regression model and Kaplan-Meier survival curves were constructed to estimate the effect of biomarkers on mortality. We used three models to adjust for potential bias. Receiver operating characteristic (ROC) curves and the area under the curve (AUC) were analyzed for the excellent performance of RAR on prognosis.

RESULTS

A total of 826 patients were eventually enrolled in our study. In multivariate analysis, RAR was found to be associated with 30-day mortality (Model 3: HR = 1.23, 95% CI = 1.09-1.39, P < .001). In addition, Subgroup analysis showed that the effect of RAR was higher in female patients than in male patients (P for interaction = .026). Kaplan-Meier survival curves showed that patients in the lower RAR quartile tended to have higher survival rates in the short and long term. AMI patients with RAR ≥ 4 had a 122% increase in 3-year mortality. Results of ROC and AUC showed that the prognostic performance of RAR for mortality was the best (30-day mortality: 0.703; 1-year mortality: 0.729; 3-year mortality: 0.737).

CONCLUSIONS

RAR is a simple and stable predictor of prognosis in AMI patients. Our results support RAR = 4.0 as a criterion for prognostic risk stratification of AMI patients.

OUP accepted manuscript

Objective

This systematic review aims to assess how information from unstructured text is used to develop and validate clinical prognostic prediction models. We summarize the prediction problems and methodological landscape and determine whether using text data in addition to more commonly used structured data improves the prediction performance.

Materials and Methods

We searched Embase, MEDLINE, Web of Science, and Google Scholar to identify studies that developed prognostic prediction models using information extracted from unstructured text in a data-driven manner, published in the period from January 2005 to March 2021. Data items were extracted, analyzed, and a meta-analysis of the model performance was carried out to assess the added value of text to structured-data models.

Results

We identified 126 studies that described 145 clinical prediction problems. Combining text and structured data improved model performance, compared with using only text or only structured data. In these studies, a wide variety of dense and sparse numeric text representations were combined with both deep learning and more traditional machine learning methods. External validation, public availability, and attention for the explainability of the developed models were limited.

Conclusion

The use of unstructured text in the development of prognostic prediction models has been found beneficial in addition to structured data in most studies. The text data are source of valuable information for prediction model development and should not be neglected. We suggest a future focus on explainability and external validation of the developed models, promoting robust and trustworthy prediction models in clinical practice.

Association Between Platelet Levels and 28-Day Mortality in Patients With Sepsis: A Retrospective Analysis of a Large Clinical Database MIMIC-IV

BACKGROUND

This research focused on evaluating the correlation between platelet count and sepsis prognosis, and even the dose-response relationship, in a cohort of American adults.

METHOD

Platelet counts were recorded retrospectively after hospitalization for patients admitted to Beth Israel Deaconess Medical Center's intensive care unit between 2008 and 2019. On admission to the intensive care unit, sepsis patients were divided into four categories based on platelet counts (very low < 50 × 10⁹/L, intermediate-low 50 × 10⁹-100 × 10⁹/L, low 100 × 10⁹-150 × 10⁹/L, and normal ≥ 150 × 10⁹/L). A multivariate Cox proportional risk model was used to calculate the 28-day risk of mortality in sepsis based on baseline platelet counts, and a two-piece linear regression model was used to calculate the threshold effect.

RESULTS

The risk of 28-day septic mortality was nearly 2-fold higher in the platelet very low group when compared to the low group (hazard ratios [HRs], 2.24; 95% confidence interval [CI], 1.92-2.6). Further analysis revealed a curvilinear association between platelets and the sepsis risk of death, with a saturation effect predicted at 100 × 10⁹/L. When platelet counts were below 100 × 10⁹/L, the risk of sepsis 28-day death decreased significantly with increasing platelet count levels (HR, 0.875; 95% CI, 0.84-0.90).

CONCLUSION

When platelet count was less than 100 × 10⁹/L, it was a strong predictor of the potential risk of sepsis death, which is declined by 13% for every 10 × 10⁹/L growth in platelets. When platelet counts reach up to 100 × 10⁹/L, the probability of dying to sepsis within 28 days climbs by 1% for every 10 × 10⁹/L increase in platelet count.

Artificial Intelligence for Detection of Cardiovascular-Related Diseases from Wearable Devices: A Systematic Review and Meta-Analysis

PURPOSE

Several artificial intelligence (AI) models for the detection and prediction of cardiovascular-related diseases, including arrhythmias, diabetes, and sleep apnea, have been reported. This systematic review and meta-analysis aimed to identify AI models developed for or applicable to wearable and mobile devices for diverse cardiovascular-related diseases.

MATERIALS AND METHODS

The searched databases included Medline, Embase, and Cochrane Library. For AI models for atrial fibrillation (AF) detection, a meta-analysis of diagnostic accuracy was performed to summarize sensitivity and specificity.

RESULTS

A total of 102 studies were included in the qualitative review. There were AI models for the detection of arrythmia (n=62), followed by sleep apnea (n=11), peripheral vascular diseases (n=6), diabetes mellitus (n=5), hyper/hypotension (n=5), valvular heart disease (n=4), heart failure (n=3), myocardial infarction and cardiac arrest (n=2), and others (n=4). For quantitative analysis of 26 studies reporting AI models for AF detection, meta-analyzed sensitivity was 94.80% and specificity was 96.96%. Deep neural networks showed superior performance [meta-analyzed area under receiver operating characteristics curve (AUROC) of 0.981] compared to conventional machine learning algorithms (meta-analyzed AUROC of 0.961). However, AI models tested with proprietary dataset (meta-analyzed AUROC of 0.972) or data acquired from wearable devices (meta-analyzed AUROC of 0.977) showed inferior performance than those with public dataset (meta-analyzed AUROC of 0.986) or data from in-hospital devices (meta-analyzed AUROC of 0.983).

CONCLUSION

This review found that AI models for diverse cardiovascular-related diseases are being developed, and that they are gradually developing into a form that is suitable for wearable and mobile devices.

AMMU: A survey of transformer-based biomedical pretrained language models

Transformer-based pretrained language models (PLMs) have started a new era in modern natural language processing (NLP). These models combine the power of transformers, transfer learning, and self-supervised learning (SSL). Following the success of these models in the general domain, the biomedical research community has developed various in-domain PLMs starting from BioBERT to the latest BioELECTRA and BioALBERT models. We strongly believe there is a need for a survey paper that can provide a comprehensive survey of various transformer-based biomedical pretrained language models (BPLMs). In this survey, we start with a brief overview of foundational concepts like self-supervised learning, embedding layer and transformer encoder layers. We discuss core concepts of transformer-based PLMs like pretraining methods, pretraining tasks, fine-tuning methods, and various embedding types specific to biomedical domain. We introduce a taxonomy for transformer-based BPLMs and then discuss all the models. We discuss various challenges and present possible solutions. We conclude by highlighting some of the open issues which will drive the research community to further improve transformer-based BPLMs. The list of all the publicly available transformer-based BPLMs along with their links is provided at https://mr-nlp.github.io/posts/2021/05/transformer-based-biomedical-pretrained-language-models-list/.

Development of a Lexicon for Pain

Pain has been an area of growing interest in the past decade and is known to be associated with mental health issues. Due to the ambiguous nature of how pain is described in text, it presents a unique natural language processing (NLP) challenge. Understanding how pain is described in text and utilizing this knowledge to improve NLP tasks would be of substantial clinical importance. Not much work has previously been done in this space. For this reason, and in order to develop an English lexicon for use in NLP applications, an exploration of pain concepts within free text was conducted. The exploratory text sources included two hospital databases, a social media platform (Twitter), and an online community (Reddit). This exploration helped select appropriate sources and inform the construction of a pain lexicon. The terms within the final lexicon were derived from three sources—literature, ontologies, and word embedding models. This lexicon was validated by two clinicians as well as compared to an existing 26-term pain sub-ontology and MeSH (Medical Subject Headings) terms. The final validated lexicon consists of 382 terms and will be used in downstream NLP tasks by helping select appropriate pain-related documents from electronic health record (EHR) databases, as well as pre-annotating these words to help in development of an NLP application for classification of mentions of pain within the documents. The lexicon and the code used to generate the embedding models have been made publicly available.

On Missingness Features in Machine Learning Models for Critical Care: Observational Study

Background

Missing data in electronic health records is inevitable and considered to be nonrandom. Several studies have found that features indicating missing patterns (missingness) encode useful information about a patient’s health and advocate for their inclusion in clinical prediction models. But their effectiveness has not been comprehensively evaluated.

Objective

The goal of the research is to study the effect of including informative missingness features in machine learning models for various clinically relevant outcomes and explore robustness of these features across patient subgroups and task settings.

Methods

A total of 48,336 electronic health records from the 2012 and 2019 PhysioNet Challenges were used, and mortality, length of stay, and sepsis outcomes were chosen. The latter dataset was multicenter, allowing external validation. Gated recurrent units were used to learn sequential patterns in the data and classify or predict labels of interest. Models were evaluated on various criteria and across population subgroups evaluating discriminative ability and calibration.

Results

Generally improved model performance in retrospective tasks was observed on including missingness features. Extent of improvement depended on the outcome of interest (area under the curve of the receiver operating characteristic [AUROC] improved from 1.2% to 7.7%) and even patient subgroup. However, missingness features did not display utility in a simulated prospective setting, being outperformed (0.9% difference in AUROC) by the model relying only on pathological features. This was despite leading to earlier detection of disease (true positives), since including these features led to a concomitant rise in false positive detections.

Conclusions

This study comprehensively evaluated effectiveness of missingness features on machine learning models. A detailed understanding of how these features affect model performance may lead to their informed use in clinical settings especially for administrative tasks like length of stay prediction where they present the greatest benefit. While missingness features, representative of health care processes, vary greatly due to intra- and interhospital factors, they may still be used in prediction models for clinically relevant outcomes. However, their use in prospective models producing frequent predictions needs to be explored further.

Cuff-less Blood Pressure Estimation from Photoplethysmography via Visibility Graph and Transfer Learning

This paper presents a new solution that enables the use of transfer learning for cuff-less blood pressure (BP) monitoring via short duration of photoplethysmogram (PPG). The proposed method estimates BP with low computational budget by 1) creating images from segments of PPG via visibility graph (VG) that preserves the temporal information of the PPG waveform, 2) using pre-trained deep convolutional neural network (CNN) to extract feature vectors from VG images, and 3) solving for the weights and bias between the feature vectors and the reference BPs with ridge regression. Using the University of California Irvine (UCI) database consisting of 348 records, the proposed method achieves a best error performance of 0.008.46 mmHg for systolic blood pressure (SBP), and -0.045.36 mmHg for diastolic blood pressure (DBP), respectively, in terms of the mean error (ME) and the standard deviation (SD) of error, ranking grade B for SBP and grade A for DBP under the British Hypertension Society (BHS) protocol. Our novel data-driven method offers a computationally-efficient end-to-end solution for rapid and user-friendly cuff-less PPG-based BP estimation.

Arrhythmia detection and classification using ECG and PPG techniques: a review

Electrocardiogram (ECG) and photoplethysmograph (PPG) are non-invasive techniques that provide electrical and hemodynamic information of the heart, respectively. This information is advantageous in the diagnosis of various cardiac abnormalities. Arrhythmia is the most common cardiovascular disease, manifested as single or multiple irregular heartbeats. However, due to the continuous manual observation, it becomes troublesome for experts sometimes to identify the paroxysmal nature of arrhythmia correctly. Moreover, due to advancements in technology, there is an inclination towards wearable sensors which monitor such patients continuously. Thus, there is a need for automatic detection techniques for the identification of arrhythmia. In the presented work, ECG and PPG-based state-of-the-art methods have been described, including preprocessing, feature extraction, and classification techniques for the detection of various arrhythmias. Additionally, this review exhibits various wearable sensors used in the literature and public databases available for the evaluation of results. The study also highlights the limitations of the current techniques and pragmatic solutions to improvise the ongoing effort.

Synthetic Data for Multi-Parameter Camera-Based Physiological Sensing

Synthetic data is a powerful tool in training data hungry deep learning algorithms. However, to date, camera-based physiological sensing has not taken full advantage of these techniques. In this work, we leverage a high-fidelity synthetics pipeline for generating videos of faces with faithful blood flow and breathing patterns. We present systematic experiments showing how physiologically-grounded synthetic data can be used in training camera-based multi-parameter cardiopulmonary sensing. We provide empirical evidence that heart and breathing rate measurement accuracy increases with the number of synthetic avatars in the training set. Furthermore, training with avatars with darker skin types leads to better overall performance than training with avatars with lighter skin types. Finally, we discuss the opportunities that synthetics present in the domain of camera-based physiological sensing and limitations that need to be overcome.

A Novel Nomogram for Predicting Survival in Patients with Severe Acute Pancreatitis: An Analysis Based on the Large MIMIC-III Clinical Database

BACKGROUND

Severe acute pancreatitis (SAP) can cause various complications. Septic shock is a relatively common and serious complication that causes uncontrolled systemic inflammatory response syndrome, which is one of the main causes of death. This study aimed to develop a nomogram for predicting the overall survival of SAP patients during the initial 24 hours following admission.

MATERIALS AND METHODS

All the data utilized in this study were obtained from the MIMIC-III (Medical Information Mart for Intensive Care III) database. The data were analyzed using multivariate Cox regression, and the performance of the proposed nomogram was evaluated based on Harrell's concordance index (C-index) and the area under the receiver operating characteristic curve (AUC). The clinical value of the prediction model was tested using decision-curve analysis (DCA). The primary outcomes were 28-day, 60-day, and 90-day mortality rates.

RESULTS

The 850 patients included in the analysis comprised 595 in the training cohort and 255 in the validation cohort. The training cohort consisted of 353 (59.3%) males and 242 (40.7%) females with SAP. Multivariate Cox regression showed that weight, sex, insurance status, explicit sepsis, SAPSII score, Elixhauser score, bilirubin, anion gap, creatinine, hematocrit, hemoglobin, RDW, SPO2, and respiratory rate were independent prognostic factors for the survival of SAP patients admitted to an intensive care unit. The predicted values were compared using C-indexes, calibration plots, integrated discrimination improvement, net reclassification improvement, and DCA.

CONCLUSIONS

We have identified some important demographic and laboratory parameters related to the prognosis of patients with SAP and have used them to establish a more accurate and convenient nomogram for evaluating their 28-day, 60-day, and 90-day mortality rates.

Prediction of arterial blood pressure waveforms from photoplethysmogram signals via fully convolutional neural networks

Cardiovascular disease (CVD) is one of the most serious diseases threatening human health. Arterial blood pressure (ABP) waveforms, containing vivid cardiovascular information, are of great significance for the diagnosis and the prevention of CVD. This paper proposes a deep learning model, named ABP-Net, to transform photoplethysmogram (PPG) signals into ABP waveforms that contain vital physiological information related to cardiovascular systems. In order to guarantee the quality of the predicted ABP waveforms, the structure of the network, the input signals and the loss functions are carefully designed. Specifically, a Wave-U-Net, one kind of fully convolutional neural networks (CNN), is taken as the core architecture of the ABP-Net. Besides the original PPG signals, its first derivative and second derivative signals are all utilized as the inputs of the ABP-Net. Additionally, the maximal absolute loss, accompany with the mean squared error loss is employed to ensure the match of the predicted ABP waveform with the reference one. The performance of the proposed ABP network is tested on the public MIMIC II database both in subject-dependent and subject-independent manners. Both results verify the superior performance of the proposed model over those existing methods accordingly. The mean absolute error (MAE) and the root-mean-square error (RMSE) between the predicted waveforms via the ABP-Net and the reference ones are 3.20 mmHg and 4.38 mmHg during the subject-dependent experiments while those are 5.57 mmHg and 7.15 mmHg during the subject-independent experiments. Benefiting from the predicted high-quality ABP waveforms, more ABP related physiological parameters can be better obtained, which effectively expands the application scope of PPG devices.

Atrial fibrillation increases inpatient and 4-year all-cause mortality in critically ill patients with liver cirrhosis

Background

The association between atrial fibrillation (AF) and cirrhosis is unclear. Therefore, the aim of the present study was to determine the association between AF and short-term and 4-year mortality in critically ill patients with cirrhosis using a large database.

Methods

The Medical Information Mart for Intensive Care III (MIMIC III) database was used to identify patients with cirrhosis hospitalized in an intensive care unit from 2001 to 2012. Demographic and clinical data were extracted from the database. Clinical data and demographic information were collected for each patient in our study. Kaplan-Meier analysis and multivariate Cox regression models were performed to examine the relation between atrial fibrillation and in-hospital and 4-year all-cause mortality.

Results

A total of 1,481 patients (mean age: 58 years, 68% male) with liver cirrhosis were included in the analysis, and the prevalence of AF was 14.18%. The inpatient all-cause mortality rate was 26.6%, and patients who died in hospital had a significantly higher rate of AF (21.57% vs. 11.50%, P<0.001). Multivariate Cox regression analysis indicated that AF was significantly associated with inpatient all-cause mortality [hazard ratio (HR): 1.52, 95% confidence interval (CI): 1.19–1.95, P<0.001], and 4-year all-cause mortality (HR: 1.55, 95% CI: 1.12–2.13, P=0.008). Kaplan-Meier survival analysis showed that patients with AF had a significantly higher inpatient and 4-year all-cause mortality.

Conclusions

Critically ill patients with liver cirrhosis have a high rate of AF, and the presence of AF is an independent risk factor for inpatient and 4-year all-cause mortality.

The Effect of Long-Term Duration Renal Replacement Therapy on Outcomes of Critically Ill Patients with Acute Kidney Injury: A Retrospective Cohort Study

Background

Renal replacement therapy (RRT), as a cornerstone of supportive treatment, has long been performed in critically ill patients with acute kidney injury (AKI). However, the majority of studies may have neglected the effect of the duration of RRT  on the outcome of AKI patients. This paper is aiming to explore the effect of the long duration of RRT  on the outcome of critically ill patients with AKI.

Methods

This retrospective study was conducted by using the Multiparameter Intelligent Monitoring in Intensive Care II (MIMIC-II) database. The primary outcome measure of this study was the mortality at 28 days, 60 days, and 90 days in the long-duration RRT group and the non-long-duration RRT group. The secondary outcomes assessed the difference in clinical outcome in these two groups. Lastly, the effect of the duration of RRT on mortality in AKI patients was determined as the third outcome.

Results

We selected 1,020 patients in total who received RRT according to the MIMIC-II database. According to the inclusion and exclusion criteria, we finally selected 506 patients with AKI: 286 AKI patients in the non-long-duration RRT group and 220 in the long-duration RRT group. After 28 days, there was a significant difference in all-cause mortality between the long-duration RRT group and the non-long-duration RRT group (P=0.001). However, the difference disappeared after 60 days and 90 days (P=0.803 and P=0.925, respectively). The length of ICU stay, length of hospital stay, and duration of mechanical ventilation were significantly longer in the long-duration RRT group than those in the non-long-duration RRT group. Considering 28-day mortality, the longer duration of RRT was shown to be a protective factor (HR = 0.995, 95% CI 0.993-0.997, P < 0.0001), while 60-day and 90-day mortality were not correlated with improved protection.

Conclusions

The long duration of RRT can improve the short-term prognosis of AKI patients, but it does not affect the long-term prognosis of these patients. Prognosis is determined by the severity of the illness itself. This suggests that RRT can protect AKI patients through the most critical time; however, the final outcome cannot be altered.

Using Restricted Cubic Splines to Study the Trajectory of Systolic Blood Pressure in the Prognosis of Acute Myocardial Infarction

Background: Acute myocardial infarction (AMI) is still the most serious manifestation of coronary artery disease. Systolic blood pressure (SBP) is the best predictor of blood pressure in AMI. Thus, its influence on AMI is necessary to be explored. Methods: A total of 4,277 patients with AMI were extracted from the Medical Information Mart for Intensive Care database. Chi-square test or Student's t-test was used to judge differences between groups, and Cox regression was used to identify factors that affect AMI prognosis. SBP was classified as low (<90 mmHg), normal (90-140 mmHg), or high (>140 mmHg), and a non-linear test was performed. Meaningful variables were incorporated into models for sensitivity analysis. Patient age was classified as low and high for subgroup analysis, and the cutoff value of the trajectory was identified. P < 0.05 indicates statistical significance. Results: The effect of SBP on the prognosis of patients with AMI is non-linear. The risks in models 1-3 with low SBP are 6.717, 4.910, and 3.080 times those of the models with normal SBP, respectively. The risks in models 1-3 with high SBP are 1.483, 1.637, and 2.937 times those of the models with normal SBP, respectively. The cutoff point (95% confidence interval) of the trajectory is 114.489 mmHg (111.275-117.702 mmHg, all P < 0.001). Conclusions: SBP has a non-linear effect on AMI prognosis. Low and high SBP show risks, and the risk of low SBP is obviously greater than that of high SBP.

A Novel Switching State-Space Model for Post-ICU Mortality Prediction and Survival Analysis

Predicting mortality risk in patients accurately during and after intensive care unit (ICU) stay is an essential component for supporting critical care decision-making. To date, various scoring systems have been designed for survival analysis and mortality prediction by providing risk scores based on patient's vital signs and lab results. However, it is challenging using these universal scores to represent the overall severity level of illness and to look into patient's deterioration leading to high mortality risk during ICU stay. Thus, a close monitoring of the severity level over time during ICU stay is more preferable. In this study, we design a new switching state-space model by correlating patient's condition dynamics in last hours of ICU stay to the risk probabilities in a short time period (1-6 days) after ICU discharge. More specifically, we propose to integrate a cumulative hazard function estimating survival probability into the autoregressive hidden Markov model using time-interval sequential SAPS II scores as features. We demonstrate the significant improvement of mortality prediction comparing to SAPS I, SAPS II, and SOFA scoring systems for the PhysioNet MIMIC II Challenge data.

Classification of blood pressure in critically ill patients using photoplethysmography and machine learning

OBJECTIVE

The aim of this study was to evaluate the capability of features extracted from photoplethysmography (PPG) based Pulse Rate Variability (PRV) to classify hypertensive, normotensive and hypotensive events, and to estimate mean arterial, systolic and diastolic blood pressure in critically ill patients.

METHODS

Time-domain, frequency-domain and non-linear indices from PRV were extracted from 5-min and 1-min segments obtained from PPG signals. These features were filtered using machine learning algorithms in order to obtain the optimal combination for the classification of hypertensive, hypotensive and normotensive events, and for the estimation of blood pressure.

RESULTS

5-min segments allowed for an improved performance in both classification and estimation tasks. Classification of blood pressure states showed around 70% accuracy and around 75% specificity. The sensitivity, precision and F1 scores were around 50%. In estimating mean arterial, systolic, and diastolic blood pressure, mean absolute errors as low as 2.55 ± 0.78 mmHg, 4.74 ± 2.33 mmHg, and 1.78 ± 0.14 mmHg were obtained, respectively. Bland-Altman analysis and Wilcoxon rank sum tests showed good agreement between real and estimated values, especially for mean and diastolic arterial blood pressures.

CONCLUSION

PRV-based features could be used for the classification of blood pressure states and the estimation of blood pressure values, although including additional features from the PPG waveform could improve the results.

SIGNIFICANCE

PRV contains information related to blood pressure, which may aid in the continuous, noninvasive, non-intrusive estimation of blood pressure and detection of hypertensive and hypotensive events in critically ill subjects.

Systemic Immune-Inflammation Index for Predicting the Prognosis of Critically Ill Patients with Acute Pancreatitis

Background

Systemic immune-inflammation index (SII) has been identified as a prognostic biomarker in various diseases. However, its significance in acute pancreatitis (AP) has not been reported. Therefore, the main aim of this study was to determine the association of SII with clinical outcomes of AP patients, after adjusting for several confounders.

Methods

This retrospective cohort study was conducted using data retrieved from the Medical Information Mart for Intensive Care III database (MIMIC-III). The study only included patients diagnosed with AP. SII was calculated as the platelet counts x neutrophil counts/lymphocyte counts. Cox regression models were employed to assess the impact of SII on the 30- and 90-day mortality of AP patients. Subgroup analysis was carried out to explore the stability of the relationship between SII and AP mortality.

Results

A total of 513 patients were found to be eligible based on the inclusion and exclusion criteria. For 30-day all-cause mortality, in the model adjusted for multiple confounders, the HR (95% CI) for mid-SII group (SII: 75.6−104.2) and high-SII groups (SII: >104.2) were 1.29 (0.65, 2.56) and 2.57 (1.35, 4.88), respectively, compared to the low-SII group (SII: <75.5). A similar trend was observed for 90-day mortality. Subgroup analyses presented a stable relationship between SII and 30-day all-cause mortality of AP patients.

Conclusion

SII is a potentially useful prognostic biomarker for AP. However, prospective studies are needed to confirm this finding.

Occlusion-Based Explanations in Deep Recurrent Models for Biomedical Signals

The biomedical field is characterized by an ever-increasing production of sequential data, which often come in the form of biosignals capturing the time-evolution of physiological processes, such as blood pressure and brain activity. This has motivated a large body of research dealing with the development of machine learning techniques for the predictive analysis of such biosignals. Unfortunately, in high-stakes decision making, such as clinical diagnosis, the opacity of machine learning models becomes a crucial aspect to be addressed in order to increase the trust and adoption of AI technology. In this paper, we propose a model agnostic explanation method, based on occlusion, that enables the learning of the input’s influence on the model predictions. We specifically target problems involving the predictive analysis of time-series data and the models that are typically used to deal with data of such nature, i.e., recurrent neural networks. Our approach is able to provide two different kinds of explanations: one suitable for technical experts, who need to verify the quality and correctness of machine learning models, and one suited to physicians, who need to understand the rationale underlying the prediction to make aware decisions. A wide experimentation on different physiological data demonstrates the effectiveness of our approach both in classification and regression tasks.

[Machine learning in anesthesiology]

The application of artificial intelligence (AI) is currently changing very different areas of life. Artificial intelligence involves the emulation of human behavior with the aid of methods from mathematics and informatics. Machine learning (ML) represents a subdivision of AI. Algorithms for ML have the potential to optimize patient care, in that they can be utilized in a supportive way in personalized medicine, decision making and risk prediction. Although the majority of the applications in medicine are still limited to data analysis and research, it is certain that ML will become increasingly more important in scientific and clinical aspects in this supportive function. Therefore, it is necessary for clinicians to have at least a basic understanding of the functional principles, strengths and weaknesses of ML.

MSSET: A high-performance time-frequency analysis method for sparse-spectrum biomedical signal

When processing sparse-spectrum biomedical signals, traditional time-frequency (TF) analysis methods are faced with the defects of blurry energy concentration and low TF resolution caused by the Heisenberg uncertainty principle. The synchrosqueezing-based methods have demonstrated advanced TF performances in recent studies. However, these methods contain at least three drawbacks: (1) existence of non-reassigned points (NRPs), (2) low noise robustness, and (3) low amplitude accuracy. In this study, the novel TF method, termed multi-synchrosqueezing extracting transform (MSSET), is proposed to address these limitations. The proposed MSSET is divided into three steps. First, multisynchrosqueezing transform (MSST) is performed with specific iterations. Second, a synch-extracting is applied to retain the TF distribution of MSST results that relate most to time-varying information of the raw signal; meanwhile, the other smeared TF energy is discarded. Finally, the MSSET result is obtained by rounding the adjacent results at the frequency plane. Numerical verification results show that the proposed MSSET method can effectively solve the NRPs problem and enhance noise robustness. Furthermore, while retaining superior energy concentration and signal reconstruction capability, the MSSET's amplitude accuracy reaches about 90%, significantly higher than other methods. In the same conditions, the MSSET even consumes less time than MSST and IMSST. It also achieves the best composite performance with the least amplitude accuracy-time cost ratio (ATCR) index. Actual application examples in the Bat signal and the electrocardiograph (ECG) signal also validate the excellent performances of our method. To conclude, our proposed MSSET is superior to state-of-the-art methods and is expected to be widely used in the sparse-spectrum biomedical signal.

Mosaic Privacy-Preserving Mechanisms for Healthcare Analytics

The Internet of Things (IoT) has propelled the evolution of medical sensing technologies to greater heights. Thus, traditional health systems have been transformed into new data-rich environments. This provides an unprecedented opportunity to develop new analytical methods and tools towards a new paradigm of smart and interconnected health systems. Nevertheless, there are risks pertinent to increasing levels of system connectivity and data accessibility. Cyber-attacks become more prevalent and complex, leading to greater likelihood of data breaches. These events bring sudden disruptions to routine operations and cause the loss of billions of dollars. Adversaries often attempt to leverage models to learn a target's sensitive attributes or extrapolate its inclusion within a database. As healthcare systems are critical to improving the wellbeing of our society, there is an urgent need to protect the privacy of patients and minimize the risk of model inversion attacks. This paper presents a new approach, named Mosaic Gradient Perturbation (MGP), to preserve privacy in the framework of predictive modeling, which meets the requirement of differential privacy while mitigating the risk of model inversion. MGP is flexible in fine-tuning the trade-offs between model performance and attack accuracy while being highly scalable for large-scale computing. Experimental results show that the proposed MGP method improves upon traditional gradient perturbation to mitigate the risk of model inversion while offering greater preservation of model accuracy. The MGP technique shows strong potential to circumvent paramount costs due to privacy breaches while maintaining the quality of existing decision-support systems, thereby ushering in a privacy-preserving smart health system.

Precision Medicine in Neonates: A Tailored Approach to Neonatal Brain Injury

Despite advances in neonatal care to prevent neonatal brain injury and neurodevelopmental impairment, predicting long-term outcome in neonates at risk for brain injury remains difficult. Early prognosis is currently based on cranial ultrasound (CUS), MRI, EEG, NIRS, and/or general movements assessed at specific ages, and predicting outcome in an individual (precision medicine) is not yet possible. New algorithms based on large databases and machine learning applied to clinical, neuromonitoring, and neuroimaging data and genetic analysis and assays measuring multiple biomarkers (omics) can fulfill the needs of modern neonatology. A synergy of all these techniques and the use of automatic quantitative analysis might give clinicians the possibility to provide patient-targeted decision-making for individualized diagnosis, therapy, and outcome prediction. This review will first focus on common neonatal neurological diseases, associated risk factors, and most common treatments. After that, we will discuss how precision medicine and machine learning (ML) approaches could change the future of prediction and prognosis in this field.

Discrete Heart Rate Values or Continuous Streams? Representation, Variability, and Meaningful Use of Vital Sign Data

Documentation and review of patient heart rate are a fundamental process across a myriad of clinical settings. While historically recorded manually, bedside monitors now provide for the automated collection of such data. Despite the availability of continuous streaming data, patients' charts continue to reflect only a subset of this information as snapshots recorded throughout a hospitalization. Over the past decade, prominent works have explored the implications of such practices and established fundamental differences in the alignment of discrete charted vitals and steaming data captured by monitoring systems. Limited work has examined the temporal properties of these differences, how they manifest, and their relation to clinical applications. The work presented in this article addresses this disparity, providing evidence that differences between charting techniques extend to measures of variability. Our results demonstrate how variability manifests with respect to temporal elements of charting timing and how it can facilitate personalized care by contextualizing deviations in magnitude. This work also highlights the utility of variability metrics with relation to clinical measures including associations to severity scores and a case study utilizing complex variability metrics derived from the complete set of monitor data.

Day-to-day progression of vital-sign circadian rhythms in the intensive care unit

Background

Disrupted vital-sign circadian rhythms in the intensive care unit (ICU) are associated with complications such as immune system disruption, delirium and increased patient mortality. However, the prevalence and extent of this disruption is not well understood. Tools for its detection are currently limited.

Methods

This paper evaluated and compared vital-sign circadian rhythms in systolic blood pressure, heart rate, respiratory rate and temperature. Comparisons were made between the cohort of patients who recovered from the ICU and those who did not, across three large, publicly available clinical databases. This comparison included a qualitative assessment of rhythm profiles, as well as quantitative metrics such as peak–nadir excursions and correlation to a demographically matched ‘recovered’ profile.

Results

Circadian rhythms were present at the cohort level in all vital signs throughout an ICU stay. Peak–nadir excursions and correlation to a ‘recovered’ profile were typically greater throughout an ICU stay in the cohort of patients who recovered, compared to the cohort of patients who did not.

Conclusions

These results suggest that vital-sign circadian rhythms are typically present at the cohort level throughout an ICU stay and that quantitative assessment of these rhythms may provide information of prognostic use in the ICU.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-021-03574-w.

Determining respiratory rate from photoplethysmogram and electrocardiogram signals using respiratory quality indices and neural networks

Continuous and non-invasive respiratory rate (RR) monitoring would significantly improve patient outcomes. Currently, RR is under-recorded in clinical environments and is often measured by manually counting breaths. In this work, we investigate the use of respiratory signal quality quantification and several neural network (NN) structures for improved RR estimation. We extract respiratory modulation signals from the electrocardiogram (ECG) and photoplethysmogram (PPG) signals, and calculate a possible RR from each extracted signal. We develop a straightforward and efficient respiratory quality index (RQI) scheme that determines the quality of each moonddulation-extracted respiration signal. We then develop NNs for the estimation of RR, using estimated RRs and their corresponding quality index as input features. We determine that calculating RQIs for modulation-extracted RRs decreased the mean absolute error (MAE) of our NNs by up to 38.17%. When trained and tested using 60-sec waveform segments, the proposed scheme achieved an MAE of 0.638 breaths per minute. Based on these results, our scheme could be readily implemented into non-invasive wearable devices for continuous RR measurement in many healthcare applications.

Cuffless blood pressure estimation methods: physiological model parameters versus machine-learned features

Objective.For the first time in the literature, this paper investigates some crucial aspects of blood pressure (BP) monitoring using photoplethysmogram (PPG) and electrocardiogram (ECG). In general, the proposed approaches utilize two types of features: parameters extracted from physiological models or machine-learned features. To provide an overview of the different feature extraction methods, we assess the performance of these features and their combinations. We also explore the importance of the ECG waveform. Although ECG contains critical information, most models merely use it as a time reference. To take this one step further, we investigate the effect of its waveform on the performance.Approach.We extracted 27 commonly used physiological parameters in the literature. In addition, convolutional neural networks (CNNs) were deployed to define deep-learned representations. We applied the CNNs to extract two different feature sets from the PPG segments alone and alongside corresponding ECG segments. Then, the extracted feature vectors and their combinations were fed into various regression models to evaluate our hypotheses.Main results.We performed our evaluations using data collected from 200 subjects. The results were analyzed by the mean difference t-test and graphical methods. Our results confirm that the ECG waveform contains important information and helps us to improve accuracy. The comparison of the physiological parameters and machine-learned features also reveals the superiority of machine-learned representations. Moreover, our results highlight that the combination of these feature sets does not provide any additional information.Significance.We conclude that CNN feature extractors provide us with concise and precise representations of ECG and PPG for BP monitoring.

Relationship between 24-h venous blood glucose variation and mortality among patients with acute respiratory failure

Evidence indicates that glucose variation (GV) plays an important role in mortality of critically ill patients. We aimed to investigate the relationship between the coefficient of variation of 24-h venous blood glucose (24-hVBGCV) and mortality among patients with acute respiratory failure. The records of 1625 patients in the Multiparameter Intelligent Monitoring in Intensive Care II (MIMIC II) database were extracted. The 24-hVBGCV was calculated as the ratio of the standard deviation (SD) to the mean venous blood glucose level, expressed as a percentage. The outcomes included ICU mortality and in-hospital mortality. Participants were divided into three subgroups based on tertiles of 24-hVBGCV. Multivariable logistic regression models were used to evaluate the relationship between 24-hVBGCV and mortality. Sensitivity analyses were also performed in groups of patients with and without diabetes mellitus. Taking the lowest tertile as a reference, after adjustment for all the covariates, the highest tertile was significantly associated with ICU mortality [odds ratio (OR), 1.353; 95% confidence interval (CI), 1.018–1.797] and in-hospital mortality (OR, 1.319; 95% CI, 1.003–1.735), especially in the population without diabetes. The 24-hVBGCV may be associated with ICU and in-hospital mortality in patients with acute respiratory failure in the ICU, especially in those without diabetes.

Validation process of a high-resolution database in a paediatric intensive care unit-Describing the perpetual patient's validation

RATIONALE

High data quality is essential to ensure the validity of clinical and research inferences based on it. However, these data quality assessments are often missing even though these data are used in daily practice and research.

AIMS AND OBJECTIVES

Our objective was to evaluate the data quality of our high-resolution electronic database (HRDB) implemented in our paediatric intensive care unit (PICU).

METHODS

We conducted a prospective validation study of a HRDB in a 32-bed paediatric medical, surgical, and cardiac PICU in a tertiary care freestanding maternal-child health centre in Canada. All patients admitted to the PICU with at least one vital sign monitored using a cardiorespiratory monitor connected to the central monitoring station.

RESULTS

Between June 2017 and August 2018, data from 295 patient days were recorded from medical devices and 4645 data points were video recorded and compared to the corresponding data collected in the HRDB. Statistical analysis showed an excellent overall correlation (R² = 1), accuracy (100%), agreement (bias = 0, limits of agreement = 0), completeness (2% missing data), and reliability (ICC = 1) between recorded and collected data within clinically significant pre-defined limits of agreement. Divergent points could all be explained.

CONCLUSIONS

This prospective validation of a representative sample showed an excellent overall data quality.

Ontology-driven weak supervision for clinical entity classification in electronic health records

In the electronic health record, using clinical notes to identify entities such as disorders and their temporality (e.g. the order of an event relative to a time index) can inform many important analyses. However, creating training data for clinical entity tasks is time consuming and sharing labeled data is challenging due to privacy concerns. The information needs of the COVID-19 pandemic highlight the need for agile methods of training machine learning models for clinical notes. We present Trove, a framework for weakly supervised entity classification using medical ontologies and expert-generated rules. Our approach, unlike hand-labeled notes, is easy to share and modify, while offering performance comparable to learning from manually labeled training data. In this work, we validate our framework on six benchmark tasks and demonstrate Trove's ability to analyze the records of patients visiting the emergency department at Stanford Health Care for COVID-19 presenting symptoms and risk factors.

An impedance pneumography signal quality index: Design, assessment and application to respiratory rate monitoring

Impedance pneumography (ImP) is widely used for respiratory rate (RR) monitoring. However, ImP-derived RRs can be imprecise. The aim of this study was to develop a signal quality index (SQI) for the ImP signal, and couple it with a RR algorithm, to improve RR monitoring. An SQI was designed which identifies candidate breaths and assesses signal quality using: the variation in detected breath durations, how well peaks and troughs are defined, and the similarity of breath morphologies. The SQI categorises 32 s signal segments as either high or low quality. Its performance was evaluated using two critical care datasets. RRs were estimated from high-quality segments using a RR algorithm, and compared with reference RRs derived from manual annotations. The SQI had a sensitivity of 77.7 %, and specificity of 82.3 %. RRs estimated from segments classified as high quality were accurate and precise, with mean absolute errors of 0.21 and 0.40 breaths per minute (bpm) on the two datasets. Clinical monitor RRs were significantly less precise. The SQI classified 34.9 % of real-world data as high quality. In conclusion, the proposed SQI accurately identifies high-quality segments, and RRs estimated from those segments are precise enough for clinical decision making. This SQI may improve RR monitoring in critical care. Further work should assess it with wearable sensor data.

Clinician involvement in research on machine learning-based predictive clinical decision support for the hospital setting: A scoping review

OBJECTIVE

The study sought to describe the prevalence and nature of clinical expert involvement in the development, evaluation, and implementation of clinical decision support systems (CDSSs) that utilize machine learning to analyze electronic health record data to assist nurses and physicians in prognostic and treatment decision making (ie, predictive CDSSs) in the hospital.

MATERIALS AND METHODS

A systematic search of PubMed, CINAHL, and IEEE Xplore and hand-searching of relevant conference proceedings were conducted to identify eligible articles. Empirical studies of predictive CDSSs using electronic health record data for nurses or physicians in the hospital setting published in the last 5 years in peer-reviewed journals or conference proceedings were eligible for synthesis. Data from eligible studies regarding clinician involvement, stage in system design, predictive CDSS intention, and target clinician were charted and summarized.

RESULTS

Eighty studies met eligibility criteria. Clinical expert involvement was most prevalent at the beginning and late stages of system design. Most articles (95%) described developing and evaluating machine learning models, 28% of which described involving clinical experts, with nearly half functioning to verify the clinical correctness or relevance of the model (47%).

DISCUSSION

Involvement of clinical experts in predictive CDSS design should be explicitly reported in publications and evaluated for the potential to overcome predictive CDSS adoption challenges.

CONCLUSIONS

If present, clinical expert involvement is most prevalent when predictive CDSS specifications are made or when system implementations are evaluated. However, clinical experts are less prevalent in developmental stages to verify clinical correctness, select model features, preprocess data, or serve as a gold standard.

Do the types and routes of proton pump inhibitor treatments affect clostridium difficile in ICU patients? A retrospective cohort study

Background: : Proton pump inhibitors (PPI) are associated with Clostridium difficile infection (CDI). Impact of the route of administration is unknown.Research Design and Methods: Patients in Multiparameter Intelligent Monitoring in Intensive Care II database (MIMIC-II) from 2001 to 2008, >18 years old, admitted to medical, surgical, or cardiac ICUs were included. PPI exposures were omeprazole, esomeprazole, lansoprazole, and pantoprazole. PPI administration routes were oral or intravenous. Patients who received histamine receptor antagonists (H2RA) were the control arm. CDI was identified using ICD-9 diagnostic code 008.45. Multiple logistic regression analysis was performed to calculate odds ratios (OR).Results: The study included 16,820 patients (57% male) with a mean age of 63 (SD±17) years and hospitalization duration of 10.2 days (SD±11). Pantoprazole was the most common PPI (94%). CDI occurred in 2.4% and more in patients receiving PPIs than H2RAs (3.0% vs. 0.8%, p < 0.001). CDI prevalence increased with intravenous (95%CI = 1.69-3.39, OR 2.4) and oral (95%CI = 1.59-3.27, OR 2.3) PPI use compared to H2RAs. CDI prevalence was not associated with PPI route in the multivariable model (OR 1.07, 95%CI 0.86-1.34).Conclusions: Both intravenous and oral PPI use in the ICU were independently associated with CDI.

The prediction of mortality influential variables in an intensive care unit: a case study

The intensive care units (ICUs) are among the most expensive and essential parts of all hospitals for extremely ill patients. This study aims to predict mortality and explore the crucial factors affecting it. Generally, in the health care systems, having a fast and precise ICU mortality prediction for patients plays a key role in care quality, resulting in reduced costs and improved survival chances of the patients. In this study, we used a medical dataset, including patients' demographic details, underlying diseases, laboratory disorder, and LOS. Since accurate estimates are required to have optimal results, various data pre-processings as the initial steps are used here. Besides, machine learning models are employed to predict the risk of mortality ICU discharge. For AdaBoost model, these measures are considered AUC= 0.966, sensitivity (recall) = 87.88%, Kappa=0.859, F-measure = 89.23% making it, AdaBoost, accounts for the highest rate. Our model outperforms other comparison models by using various scenarios of data processing. The obtained results demonstrate that the high mortality can be caused by underlying diseases such as diabetes mellitus and high blood pressure, moderate Pulmonary Embolism Wells Score risk, platelet blood count less than 100000 (mcl), hypertension (HTN), high level of Bilirubin, smoking, and GCS level between 6 and 9.

Improving sentiment analysis on clinical narratives by exploiting UMLS semantic types

Sentiments associated with assessments and observations recorded in a clinical narrative can often indicate a patient's health status. To perform sentiment analysis on clinical narratives, domain-specific knowledge concerning meanings of medical terms is required. In this study, semantic types in the Unified Medical Language System (UMLS) are exploited to improve lexicon-based sentiment classification methods. For sentiment classification using SentiWordNet, the overall accuracy is improved from 0.582 to 0.710 by using logistic regression to determine appropriate polarity scores for UMLS 'Disorders' semantic types. For sentiment classification using a trained lexicon, when disorder terms in a training set are replaced with their semantic types, classification accuracies are improved on some data segments containing specific semantic types. To select an appropriate classification method for a given data segment, classifier combination is proposed. Using classifier combination, classification accuracies are improved on most data segments, with the overall accuracy of 0.882 being obtained.

[Artificial intelligence in neurocritical care]

Artificial intelligence (AI) has been introduced into medicine and an AI-assisted medicine will be the future that we should help to shape. In particular, supervised, unsupervised, and reinforcement learning will be the main methods to play a role in the implementation of AI. Severely ill patients admitted to the intensive care unit (ICU) are closely monitored in order to be able to quickly respond to any changes. These monitoring data can be used to train AI models to predict critical phases in advance, making an earlier reaction possible. To achieve this a large amount of clinical data are needed in order to train models and an external validation on independent cohorts should take place. Prospective studies with treatment of patients admitted to the ICU with AI assistance should show that they provide a benefit for patients. We present the most important resources from de-identified (anonymized) patient data on open-source use for AI research in intensive care medicine. The focus is on neurological diseases in the ICU, therefore, we provide an overview of existing models for prediction of outcome, vasospasms, intracranial pressure and levels of consciousness. To introduce the advantages of AI in the clinical routine, more AI-based models with larger datasets will be needed. To achieve this international cooperation is absolutely necessary. Clinical centers associated with universities are needed to provide a constant validation of applied models as these models can change during use or a bias can develop during the training. A strong commitment to AI research is important for Germany, not only with respect to academic achievements but also in the light of a rapidly growing influence of AI on the economy.