Simplifying Multiparty Computation: A Client-Driven Metaprotocol for Federated Secure Computing.

INTRODUCTION: Secure Multi-Party Computation (SMPC) offers a powerful tool for collaborative healthcare research while preserving patient data privacy. STATE OF THE ART: However, existing SMPC frameworks often require separate executions for each desired computation and measurement period, limiting user flexibility. CONCEPT: This research explores the potential of a client-driven metaprotocol for the Federated Secure Computing (FSC) framework and its SImple Multiparty ComputatiON (SIMON) protocol as a step towards more flexible SMPC solutions. IMPLEMENTATION: This client-driven metaprotocol empowers users to specify and execute multiple calculations across diverse measurement periods within a single client-side code execution. This eliminates the need for repeated code executions and streamlines the analysis process. The metaprotocol offers a user-friendly interface, enabling researchers with limited cryptography expertise to leverage the power of SMPC for complex healthcare analyses. LESSONS LEARNED: We evaluate the performance of the client-driven metaprotocol against a baseline iterative approach. Our evaluation demonstrates performance improvements compared to traditional iterative approaches, making this metaprotocol a valuable tool for advancing secure and efficient collaborative healthcare research.

A Comparative Analysis of Federated and Centralized Learning for SpO2 Prediction in Five Critical Care Databases.

This study explores the potential of federated learning (FL) to develop a predictive model of hypoxemia in intensive care unit (ICU) patients. Centralized learning (CL) and local learning (LL) approaches have been limited by the localized nature of data, which restricts CL approaches to the available data due to data privacy regulations. A CL approach that combines data from different institutions, could offer superior performance compared to a single-institution approach. However, the use of this method raises ethical and regulatory concerns. In this context, FL presents a promising middle ground, enabling collaborative model training on geographically dispersed ICU data without compromising patient confidentiality. This study is the first to use all five public ICU databases combined. The findings demonstrate that FL achieved comparable or even slightly improved performance compared to local or centralized learning approaches.

Optimizing Data Extraction: Harnessing RAG and LLMs for German Medical Documents.

In the field of medical data analysis, converting unstructured text documents into a structured format suitable for further use is a significant challenge. This study introduces an automated local deployed data privacy secure pipeline that uses open-source Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG) architecture to convert medical German language documents with sensitive health-related information into a structured format. Testing on a proprietary dataset of 800 unstructured original medical reports demonstrated an accuracy of up to 90% in data extraction of the pipeline compared to data extracted manually by physicians and medical students. This highlights the pipeline's potential as a valuable tool for efficiently extracting relevant data from unstructured sources.

Electronic data capture in resource-limited settings using the lightweight clinical data acquisition and recording system.

Our prototype system designed for clinical data acquisition and recording of studies is a novel electronic data capture (EDC) software for simple and lightweight data capture in clinical research. Existing software tools are either costly or suffer from very limited features. To overcome these shortcomings, we designed an EDC software together with a mobile client. We aimed at making it easy to set-up, modifiable, scalable and thereby facilitating research. We wrote the software in R using a modular approach and implemented existing data standards along with a meta data driven interface and database structure. The prototype is an adaptable open-source software, which can be installed locally or in the cloud without advanced IT-knowledge. A mobile web interface and progressive web app for mobile use and desktop computers is added. We show the software's capability, by demonstrating four clinical studies with over 1600 participants and 679 variables per participant. We delineate a simple deployment approach for a server-installation and indicate further use-cases. The software is available under the MIT open-source license. Conclusively the software is versatile, easily deployable, highly modifiable, and extremely scalable for clinical studies. As an open-source R-software it is accessible, open to community-driven development and improvement in the future.

Predicting Successful Weaning from Mechanical Ventilation by Reduction in Positive End-expiratory Pressure Level Using Machine Learning.

Weaning patients from mechanical ventilation (MV) is a critical and resource intensive process in the Intensive Care Unit (ICU) that impacts patient outcomes and healthcare expenses. Weaning methods vary widely among providers. Prolonged MV is associated with adverse events and higher healthcare expenses. Predicting weaning readiness is a non-trivial process in which the positive end-expiratory pressure (PEEP), a crucial component of MV, has potential to be indicative but has not yet been used as the target. We aimed to predict successful weaning from mechanical ventilation by targeting changes in the PEEP-level using a supervised machine learning model. This retrospective study included 12,153 mechanically ventilated patients from Medical Information Mart for Intensive Care (MIMIC-IV) and eICU collaborative research database (eICU-CRD). Two machine learning models (Extreme Gradient Boosting and Logistic Regression) were developed using a continuous PEEP reduction as target. The data is splitted into 80% as training set and 20% as test set. The model's predictive performance was reported using 95% confidence interval (CI), based on evaluation metrics such as area under the receiver operating characteristic (AUROC), area under the precision-recall curve (AUPRC), F1-Score, Recall, positive predictive value (PPV), and negative predictive value (NPV). The model's descriptive performance was reported as the variable ranking using SHAP (SHapley Additive exPlanations) algorithm. The best model achieved an AUROC of 0.84 (95% CI 0.83-0.85) and an AUPRC of 0.69 (95% CI 0.67-0.70) in predicting successful weaning based on the PEEP reduction. The model demonstrated a Recall of 0.85 (95% CI 0.84-0.86), F1-score of 0.86 (95% CI 0.85-0.87), PPV of 0.87 (95% CI 0.86-0.88), and NPV of 0.64 (95% CI 0.63-0.66). Most of the variables that SHAP algorithm ranked to be important correspond with clinical intuition, such as duration of MV, oxygen saturation (SaO2), PEEP, and Glasgow Coma Score (GCS) components. This study demonstrates the potential application of machine learning in predicting successful weaning from MV based on continuous PEEP reduction. The model's high PPV and moderate NPV suggest that it could be a useful tool to assist clinicians in making decisions regarding ventilator management.

Predicting Hypoxia Using Machine Learning: Systematic Review.

Background: Hypoxia is an important risk factor and indicator for the declining health of inpatients. Predicting future hypoxic events using machine learning is a prospective area of study to facilitate time-critical interventions to counter patient health deterioration. Objective: This systematic review aims to summarize and compare previous efforts to predict hypoxic events in the hospital setting using machine learning with respect to their methodology, predictive performance, and assessed population. Methods: A systematic literature search was performed using Web of Science, Ovid with Embase and MEDLINE, and Google Scholar. Studies that investigated hypoxia or hypoxemia of hospitalized patients using machine learning models were considered. Risk of bias was assessed using the Prediction Model Risk of Bias Assessment Tool. Results: After screening, a total of 12 papers were eligible for analysis, from which 32 models were extracted. The included studies showed a variety of population, methodology, and outcome definition. Comparability was further limited due to unclear or high risk of bias for most studies (10/12, 83%). The overall predictive performance ranged from moderate to high. Based on classification metrics, deep learning models performed similar to or outperformed conventional machine learning models within the same studies. Models using only prior peripheral oxygen saturation as a clinical variable showed better performance than models based on multiple variables, with most of these studies (2/3, 67%) using a long short-term memory algorithm. Conclusions: Machine learning models provide the potential to accurately predict the occurrence of hypoxic events based on retrospective data. The heterogeneity of the studies and limited generalizability of their results highlight the need for further validation studies to assess their predictive performance.

Mechanical Power Density Predicts Prolonged Ventilation Following Double Lung Transplantation.

Prolonged mechanical ventilation (PMV) after lung transplantation poses several risks, including higher tracheostomy rates and increased in-hospital mortality. Mechanical power (MP) of artificial ventilation unifies the ventilatory variables that determine gas exchange and may be related to allograft function following transplant, affecting ventilator weaning. We retrospectively analyzed consecutive double lung transplant recipients at a national transplant center, ventilated through endotracheal tubes upon ICU admission, excluding those receiving extracorporeal support. MP and derived indexes assessed up to 36 h after transplant were correlated with invasive ventilation duration using Spearman's coefficient, and we conducted receiver operating characteristic (ROC) curve analysis to evaluate the accuracy in predicting PMV (>72 h), expressed as area under the ROC curve (AUROC). PMV occurred in 82 (35%) out of 237 cases. MP was significantly correlated with invasive ventilation duration (Spearman's ρ = 0.252 [95% CI 0.129-0.369], p < 0.01), with power density (MP normalized to lung-thorax compliance) demonstrating the strongest correlation (ρ = 0.452 [0.345-0.548], p < 0.01) and enhancing PMV prediction (AUROC 0.78 [95% CI 0.72-0.83], p < 0.01) compared to MP (AUROC 0.66 [0.60-0.72], p < 0.01). Mechanical power density may help identify patients at risk for PMV after double lung transplantation.

Towards interoperability in infection control: a standard data model for microbiology.

The COVID-19 pandemic has made it clear: sharing and exchanging data among research institutions is crucial in order to efficiently respond to global health threats. This can be facilitated by defining health data models based on interoperability standards. In Germany, a national effort is in progress to create common data models using international healthcare IT standards. In this context, collaborative work on a data set module for microbiology is of particular importance as the WHO has declared antimicrobial resistance one of the top global public health threats that humanity is facing. In this article, we describe how we developed a common model for microbiology data in an interdisciplinary collaborative effort and how we make use of the standard HL7 FHIR and terminologies such as SNOMED CT or LOINC to ensure syntactic and semantic interoperability. The use of international healthcare standards qualifies our data model to be adopted beyond the environment where it was first developed and used at an international level.

Data suggested hospitalization as critical indicator of the severity of the COVID-19 pandemic, even at its early stages.

COVID-19 has been spreading widely since January 2020, prompting the implementation of non-pharmaceutical interventions and vaccinations to prevent overwhelming the healthcare system. Our study models four waves of the epidemic in Munich over two years using a deterministic, biology-based mathematical model of SEIR type that incorporates both non-pharmaceutical interventions and vaccinations. We analyzed incidence and hospitalization data from Munich hospitals and used a two-step approach to fit the model parameters: first, we modeled incidence without hospitalization, and then we extended the model to include hospitalization compartments using the previous estimates as a starting point. For the first two waves, changes in key parameters, such as contact reduction and increasing vaccinations, were enough to represent the data. For wave three, the introduction of vaccination compartments was essential. In wave four, reducing contacts and increasing vaccinations were critical parameters for controlling infections. The importance of hospitalization data was highlighted, as it should have been included as a crucial parameter from the outset, along with incidence, to avoid miscommunication with the public. The emergence of milder variants like Omicron and a significant proportion of vaccinated people has made this fact even more evident.

Integrative modelling of reported case numbers and seroprevalence reveals time-dependent test efficiency and infectious contacts.

Mathematical models have been widely used during the ongoing SARS-CoV-2 pandemic for data interpretation, forecasting, and policy making. However, most models are based on officially reported case numbers, which depend on test availability and test strategies. The time dependence of these factors renders interpretation difficult and might even result in estimation biases. Here, we present a computational modelling framework that allows for the integration of reported case numbers with seroprevalence estimates obtained from representative population cohorts. To account for the time dependence of infection and testing rates, we embed flexible splines in an epidemiological model. The parameters of these splines are estimated, along with the other parameters, from the available data using a Bayesian approach. The application of this approach to the official case numbers reported for Munich (Germany) and the seroprevalence reported by the prospective COVID-19 Cohort Munich (KoCo19) provides first estimates for the time dependence of the under-reporting factor. Furthermore, we estimate how the effectiveness of non-pharmaceutical interventions and of the testing strategy evolves over time. Overall, our results show that the integration of temporally highly resolved and representative data is beneficial for accurate epidemiological analyses.

A Functional Network Driven by MicroRNA-125a Regulates Monocyte Trafficking in Acute Inflammation.

During the onset of acute inflammation, rapid trafficking of leukocytes is essential to mount appropriate immune responses towards an inflammatory insult. Monocytes are especially indispensable for counteracting the inflammatory stimulus, neutralising the noxa and reconstituting tissue homeostasis. Thus, monocyte trafficking to the inflammatory sites needs to be precisely orchestrated. In this study, we identify a regulatory network driven by miR-125a that affects monocyte adhesion and chemotaxis by the direct targeting of two adhesion molecules, i.e., junction adhesion molecule A (JAM-A), junction adhesion molecule-like (JAM-L) and the chemotaxis-mediating chemokine receptor CCR2. By investigating monocytes isolated from patients undergoing cardiac surgery, we found that acute yet sterile inflammation reduces miR-125a levels, concomitantly enhancing the expression of JAM-A, JAM-L and CCR2. In contrast, TLR-4-specific stimulation with the pathogen-associated molecular pattern (PAMP) LPS, usually present within the perivascular inflamed area, resulted in dramatically induced levels of miR-125a with concomitant repression of JAM-A, JAM-L and CCR2 as early as 3.5 h. Our study identifies miR-125a as an important regulator of monocyte trafficking and shows that the phenotype of human monocytes is strongly influenced by this miRNA, depending on the type of inflammatory stimulus.

Early Use of Methylene Blue in Vasoplegic Syndrome: A 10-Year Propensity Score-Matched Cohort Study.

BACKGROUND: Vasoplegic syndrome is associated with increased morbidity and mortality in patients undergoing cardiac surgery. This retrospective, single-center study aimed to evaluate the effect of early use of methylene blue (MB) on hemodynamics after an intraoperative diagnosis of vasoplegic syndrome (VS). METHODS: Over a 10-year period, all patients diagnosed with intraoperative VS (hypotension despite treatment with norepinephrine ≥0.3 µg/kg/min and vasopressin ≥1 IE/h) while undergoing heart surgery and cardiopulmonary bypass were identified, and their data were examined. The intervention group received MB (2 mg/kg intravenous) within 15 min after the diagnosis of vasoplegia, while the control group received standard therapy. The two groups were matched using propensity scores. RESULTS: Of the 1022 patients identified with VS, 221 received MB intraoperatively, and among them, 60 patients received MB within 15 min after the diagnosis of VS. After early MB application, mean arterial pressure was significantly higher, and vasopressor support was significantly lower within the first hour (p = 0.015) after the diagnosis of vasoplegia, resulting in a lower cumulative amount of norepinephrine (p = 0.018) and vasopressin (p = 0.003). The intraoperative need of fresh frozen plasma in the intervention group was lower compared to the control group (p = 0.015). Additionally, the intervention group had higher creatinine values in the first three postoperative days (p = 0.036) without changes in dialysis incidence. The 90-day survival did not differ significantly (p = 0.270). CONCLUSION: Our results indicate the additive effects of MB use during VS compared to standard vasopressor therapy only. Early MB administration for VS may significantly improve the patients' hemodynamics with minor side effects.

[SEPSIS-3.0-Is intensive care medicine ready for ICD-11?] / SEPSIS-3.0 ­ Ist die Intensivmedizin bereit für die ICD-11?

BACKGROUND: The 11th revision of the International Classification of Diseases (ICD-11) will come into effect in January 2022. Among other things, The Third International Consensus Definitions for Sepsis and Septic Shock (SEPSIS­3 definition) will be implemented in it. This defines sepsis as a "life-threatening organ dysfunction caused by a dysregulated host response to infection". The aim of the present secondary analysis of a survey on the topic of "sepsis-induced coagulopathy" was to evaluate whether the SEPSIS­3 definition, 4 years after its international introduction, has arrived in everyday clinical practice of intensive care units (ICU) run by anesthesiologists in Germany and thus the requirements for its use of the ICD-11 are given. METHODS: Between October 2019 and May 2020, we carried out a nationwide survey among German medical directors of ICUs. In a separate block of questions we asked about the definition of sepsis used in daily practice. In addition, we asked whether the quick-sequential (sepsis-related) organ failure assessment (qSOFA) score is used in screening for sepsis in the hospital to which to the participating ICU belongs. RESULTS: A total of 50 medical directors from anesthesiological ICUs took part in the survey. In total, the ICUs evaluated stated that they had around 14% of the high-care beds registered in Germany. The SEPSIS­3 definition is integrated into everyday clinical practice at 78.9% of the university hospitals and 84.0% of the participating teaching hospitals. In contrast, the qSOFA screening test is only used by 26.3% of the participating university hospitals, but at least 52% of the teaching hospitals and 80% of the other hospitals. CONCLUSION: The data show that both SEPSIS­3 and qSOFA have become part of everyday clinical practice in German hospitals. The cautious use of qSOFA at university hospitals with simultaneous broad acceptance of the SEPSIS­3 definition can be interpreted as an indication that the search for a suitable screening test for sepsis has not yet been completed.

[Practice of pharmaceutical thrombosis prophylaxis and anticoagulation in patients with sepsis and pre-existing anticoagulation or heparin-induced type II thrombocytopenia-Results of a nationwide survey in German intensive care units]. / Praxis der medikamentösen Thromboseprophylaxe und Antikoagulation bei Patienten mit Sepsis und vorbestehender Antikoagulation oder Heparin-induzierter Thrombozytopenie Typ II ­ Ergebnisse einer deutschlandweiten Umfrage auf Intensivstationen.

BACKGROUND: A pre-existing anticoagulation treatment and predisposing diseases for thromboembolic events represent common problems in patients with sepsis or septic shock; however, these conditions are not addressed in current national guidelines for sepsis and septic shock. One of the aims of this nationwide survey in Germany was therefore to determine how intensive care physicians deal with these problems. METHODS: From October 2019 to May 2020, we conducted a nationwide survey among German medical directors of intensive care units (ICU) addressing anticoagulation and drug-based prophylaxis of venous thromboembolism (VTE) in patients with sepsis and sepsis-induced coagulopathy. One focus was the procedure for patients with a pre-existing anticoagulation treatment or a previously known heparin-induced thrombocytopenia (HIT) type 2 (acute symptomatic vs. dating back years). RESULTS: In most of the participating ICUs pre-existing anticoagulation is largely continued with low molecular weight heparin preparations or unfractionated heparin. In patients with pre-existing HIT type 2 both acute symptomatic and dating back years, argatroban represents the drug of choice. There is a high degree of variability in the definition of the target values, usually being well above the range for pure VTE prophylaxis. CONCLUSION: Data on the continuation of anticoagulation beyond VTE prophylaxis with a subsequently increased risk of bleeding in patients with sepsis and septic shock is limited and treatment decisions are in many cases subject to individual consideration by the practitioner. The results of our survey imply the need for a systematic work-up of this topic in order to support daily practice in many ICUs with the required evidence.

Anthropometric analysis of body habitus and outcomes in critically ill COVID-19 patients.

AIMS: This study aimed to determine whether anthropometric markers of thoracic skeletal muscle and abdominal visceral fat tissue correlate with outcome parameters in critically ill COVID-19 patients. METHODS: We retrospectively analysed thoracic CT-scans of 67 patients in four ICUs at a university hospital. Thoracic skeletal muscle (total cross-sectional area (CSA); pectoralis muscle area (PMA)) and abdominal visceral fat tissue (VAT) were quantified using a semi-automated method. Point-biserial-correlation-coefficient, Spearman-correlation-coefficient, Wilcoxon rank-sum test and logistic regression were used to assess the correlation and test for differences between anthropometric parameters and death, ventilator- and ICU-free days and initial inflammatory laboratory values. RESULTS: Deceased patients had lower CSA and PMA values, but higher VAT values (p < 0.001). Male patients with higher CSA values had more ventilator-free days (p = 0.047) and ICU-free days (p = 0.017). Higher VAT/CSA and VAT/PMA values were associated with higher mortality (p < 0.001), but were negatively correlated with ICU length of stay in female patients only (p < 0.016). There was no association between anthropometric parameters and initial inflammatory biomarker levels. Logistic regression revealed no significant independent predictor for death. CONCLUSION: Our study suggests that pathologic body composition assessed by planimetric measurements using thoracic CT-scans is associated with worse outcome in critically ill COVID-19 patients.

Repurposing existing medications for coronavirus disease 2019: protocol for a rapid and living systematic review.

BACKGROUND: Coronavirus disease 2019 (COVID-19) has no confirmed specific treatments. However, there might be in vitro and early clinical data as well as evidence from severe acute respiratory syndrome and Middle Eastern respiratory syndrome that could inform clinicians and researchers. This systematic review aims to create priorities for future research of drugs repurposed for COVID-19. METHODS: This systematic review will include in vitro, animal, and clinical studies evaluating the efficacy of a list of 34 specific compounds and 4 groups of drugs identified in a previous scoping review. Studies will be identified both from traditional literature databases and pre-print servers. Outcomes assessed will include time to clinical improvement, time to viral clearance, mortality, length of hospital stay, and proportions transferred to the intensive care unit and intubated, respectively. We will use the GRADE methodology to assess the quality of the evidence. DISCUSSION: The challenge posed by COVID-19 requires not just a rapid review of drugs that can be repurposed but also a sustained effort to integrate new evidence into a living systematic review. TRIAL REGISTRATION: PROSPERO 2020 CRD42020175648.

[Treatment of sepsis-induced coagulopathy : Results of a Germany-wide survey in intensive care units]. / Therapie der sepsisinduzierten Koagulopathie : Ergebnisse einer deutschlandweiten Umfrage auf Intensivstationen.

BACKGROUND: In the context of sepsis and septic shock, coagulopathy often occurs due to the close relationship between coagulation and inflammation. Sepsis-induced coagulopathy (SIC) is the most severe and potentially fatal form. Anticoagulants used in prophylactic or therapeutic doses are discussed to potentially exert beneficial effects in patients with sepsis and/or SIC; however, due to the lack of evidence recent guidelines are limited to recommendations for drug prophylaxis of venous thromboembolism (VTE), while treatment of SIC has not been addressed. METHODS: In order to determine the status quo of VTE prophylaxis as well as treatment of SIC in German intensive care units (ICU), we conducted a Germany-wide online survey among heads of ICUs from October 2019 to May 2020. In April 2020, the survey was supplemented by an additional block of questions on VTE prophylaxis and SIC treatment in coronavirus disease 2019 (COVID-19) patients. RESULTS: A total of 67 senior doctors took part in the survey. The majority (n = 50; 74.6%) of the responses were from ICU under the direction of an anesthesiologist and/or a department of anesthesiology. Most of the participants worked either at a university hospital (n = 31; 47.8%) or an academic teaching hospital (n = 27; 40.3%). The survey results show a pronounced heterogeneity in clinical practice with respect to the prophylaxis of VTE as well as SIC treatment. In an exemplary case of pneumogenic sepsis, low molecular weight heparins (LMWH) were by far the most frequently mentioned group of medications (n = 51; 76.1% of the responding ITS). In the majority of cases (n = 43; 64.2%), anti-FXa activity is not monitored with the use of LMWH in prophylaxis doses. Unfractionated heparin (UFH) was listed as a strategy for VTE prophylaxis in 37.3% of the responses (n = 25). In an exemplary case of abdominal sepsis 54.5% of the participants (n = 36; multiple answers possible) stated the use of UFH or LMWH and UFH with dosage controlled by PTT is used on two participating ICUs. The anti-FXa activity under prophylactic anticoagulation with LMWH is monitored in 7 participating clinics (10.6%) in abdominal sepsis. Systematic screening for sepsis-associated coagulation disorders does not take place in most hospitals and patterns in the use of anticoagulants show significant variability between ICUs. In the case of COVID-19 patients, it is particularly noticeable that in three quarters of the participating ICUs the practice of drug-based VTE prophylaxis and SIC treatment does not differ from that of non-COVID-19 patients. CONCLUSION: The heterogeneity of answers collected in the survey suggests that a systematic approach to this topic via clinical trials is urgently needed to underline individualized patient care with the necessary evidence.