Search

Slender catalytic packed bed reactors are widely used in the chemical industry for chemical processes accompanied with significant heat liberation necessitating efficient heat removal. The employed catalyst particles are often cylindrical shaped, because these are easily manufactured via extrusion. The packing of these cylindrical particles results in a random structure of the bed causing flow maldistribution which reduces the reactor efficiency. The goal of this work is to capture the flow maldistribution with particle-resolved Computational Fluid Dynamics simulations and Magnetic Resonance Imaging experiments. By utilising the packing configuration reconstructed from the experiments, we are able to study the average fluid flow profiles and flow distribution in the bed as well as the local velocity profiles. It can be concluded that there is an overall good correspondence between the experiments and simulations. The minor deviations in the experimental and computed velocity distributions can be attributed to experimental error and slight differences in the experimental and reconstructed packing.

Novel prediction methods should always be compared to a baseline to know how well they perform. Without this frame of reference, the performance score of a model is basically meaningless. What does it mean when a model achieves an F1 of 0.8 on a test set? A proper baseline is needed to evaluate the `goodness' of a performance score. Comparing with the latest state-of-the-art model is usually insightful. However, being state-of-the-art can change rapidly when newer models are developed. Contrary to an advanced model, a simple dummy classifier could be used. However, the latter could be beaten too easily, making the comparison less valuable. This paper presents a universal baseline method for all binary classification models, named the Dutch Draw (DD). This approach weighs simple classifiers and determines the best classifier to use as a baseline. We theoretically derive the DD baseline for many commonly used evaluation measures and show that in most situations it reduces to (almost) always predicting either zero or one. Summarizing, the DD baseline is: (1) general, as it is applicable to all binary classification problems; (2) simple, as it is quickly determined without training or parameter-tuning; (3) informative, as insightful conclusions can be drawn from the results. The DD baseline serves two purposes. First, to enable comparisons across research papers by this robust and universal baseline. Secondly, to provide a sanity check during the development process of a prediction model. It is a major warning sign when a model is outperformed by the DD baseline.

As telecommunication systems evolve to meet increasing demands, integrating deep neural networks (DNNs) has shown promise in enhancing performance. However, the trade-off between accuracy and flexibility remains challenging when replacing traditional receivers with DNNs. This paper introduces a novel probabilistic framework that allows a single DNN demapper to demap multiple QAM and APSK constellations simultaneously. We also demonstrate that our framework allows exploiting hierarchical relationships in families of constellations. The consequence is that we need fewer neural network outputs to encode the same function without an increase in Bit Error Rate (BER). Our simulation results confirm that our approach approaches the optimal demodulation error bound under an Additive White Gaussian Noise (AWGN) channel for multiple constellations. Thereby, we address multiple important issues in making DNNs flexible enough for practical use as receivers.

After a revision surgery, approximately 1–2 % of patients will develop a periprosthetic joint infection (PJI). During the revision surgery, the infected prosthesis is removed, a debridement is performed and a new or temporary spacer is placed. Additionally, patients are treated with antibiotics during and after the surgery. Adequate exposure of the administered antibiotic to the pathogen is of crucial importance during the treatment of any infection. Inadequately low concentrations are associated with an increase in antibiotic resistance, antibiotic related side effects, treatment failures and prolonged infections. While high concentrations may lead to serious adverse events and potential lasting damage. Despite the importance of optimal dosing, there is a lack of knowledge with respect to the correlation between the plasma concentrations and target site concentrations of the antibiotics. Two of the commonly administered antimicrobial agents during the arthroplasty exchange are cefuroxime and flucloxacillin. Therefore, an accurate, specific, and sensitive quantification method is required in order to assess pharmacokinetics of cefuroxime and flucloxacillin in synovial tissue and bone. The aim of this study is to develop and validate a quantification method for the measurement of cefuroxime and flucloxacillin in human synovial tissue and bone using the UPC2-MS/MS conform Food and Drug Administration guidelines. The method was found linear for both compounds in both matrices (r2 > 0.990) from 1 µg/g to 20 µg/g, except for cefuroxime in bone, which was validated from 1 µg/g to 15 µg/g. We developed and validated a quantification method for cefuroxime and flucloxacillin in synovial tissue and bone using a simple sample preparation and a short analysis run time of 5.0 min, which has been already successfully applied in a clinical study. To our knowledge, no methods have been described earlier for the simultaneous quantification of cefuroxime a

Background: Frailty in newly-diagnosed multiple myeloma (NDMM) patients is associated with treatment-related toxicity, which negatively affects health-related quality of life (HRQoL). Currently, data on changes in HRQoL of frail and intermediate-fit MM patients during active treatment and post-treatment follow-up are absent. Methods: The HOVON123 study (NTR4244) was a phase II trial in which NDMM patients ≥ 75 years were treated with nine dose-adjusted cycles of Melphalan-Prednisone-Bortezomib (MPV). Two HRQoL instruments (EORTC QLQ-C30 and -MY20) were obtained before start of treatment, after 3 and 9 months of treatment and 6 and 12 months after treatment for patients who did not yet start second-line treatment. HRQoL changes and/or differences in frail and intermediate-fit patients (IMWG frailty score) were reported only when both statistically significant (p < 0.005) and clinically relevant (>MID). Results: 137 frail and 71 intermediate-fit patients were included in the analysis. Compliance was high and comparable in both groups. At baseline, frail patients reported lower global health status, lower physical functioning scores and more fatigue and pain compared to intermediate-fit patients. Both groups improved in global health status and future perspective; polyneuropathy complaints worsened over time. Frail patients improved over time in physical functioning, fatigue and pain. Improvement in global health status occurred earlier than in intermediate-fit patients. Conclusion: HRQoL improved during anti-myeloma treatment in both intermediate-fit and frail MM patients. In frail patients, improvement occurred faster and, in more domains, which was retained during follow-up. This implies that physicians should not withhold safe and effective therapies from frail patients in fear of HRQoL deterioration.

01 februari 2023, Item does not contain fulltext, Suboptimal guideline adherence in chronic myeloid leukemia (CML) care is associated with worse treatment outcomes. Current study focused on adherence to seven quality indicators (QIs) based on the European Leukemia Network guideline (one diagnostic, one therapeutic, and five monitoring indicators). Data were obtained from population-based registries in the Netherlands of 405 newly diagnosed chronic phase CML patients between January 2008 and April 2013. Compliance rates regarding diagnostic and therapeutic indicator were 83% and 78%, respectively. Monitoring indicators rates were lower: 21-27% for indicators concerning the first year and 58% and 62% for the second and third year, respectively. Noncompliance occurred mostly due to non-timely monitoring. Twenty cases did not comply with any indicator, 6% complied with all indicators. After adjustment for age, overall survival rates did not differ significantly between the groups. Adherence to guideline-based QIs was suboptimal. This demonstrates the evidence-practice gap, shows room for improvement and underscores the need for real-world data.

Item does not contain fulltext

Contains fulltext : 299845.pdf (Publisher’s version ) (Closed access), OBJECTIVES: Acquired missense mutations in the BCR::ABL1 kinase domain (KD) may cause tyrosine kinase inhibitor (TKI) treatment failure. Based on mutation-specific in vitro derived IC50-values, alternative TKI may be selected. We assessed clinical practice of BCR::ABL1 KD mutation testing, clinical response in relation to IC50-values, and clinical outcome of tested patients. METHODS: Patients from six Dutch CML reference centers and a national registry were included once a mutational analysis was performed. Reasons for testing were categorized as suboptimal TKI response, and primary or secondary TKI resistance. RESULTS: Four hundred twenty analyses were performed in 275 patients. Sixty-nine patients harbored at least one mutation. Most analyses were performed because of suboptimal TKI response but with low mutation incidence (4%), while most mutations were found in primary and secondary resistant patients (21% and 51%, respectively). Harboring a BCR::ABL1 mutation was associated with inferior overall survival (HR 3.2 [95% CI, 1.7-6.1; p < .001]). Clinically observed responses to TKI usually corresponded with the predicted TKI sensitivity based on the IC50-values, but a high IC50-value did not preclude a good clinical response per se. CONCLUSIONS: We recommend BCR::ABL1 KD mutation testing in particular in the context of primary or secondary resistance. IC50-values can direct the TKI choice for CML patients, but clinical efficacy can be seen despite adverse in vitro resistance., 01 december 2023

Contains fulltext : 299611.pdf (Publisher’s version ) (Closed access), BACKGROUND: Many older patients with acute myeloid leukaemia die or cannot undergo allogeneic haematopoietic stem-cell transplantation (HSCT) due to toxicity caused by intensive chemotherapy. We hypothesised that replacing intensive chemotherapy with decitabine monotherapy could improve outcomes. METHODS: This open-label, randomised, controlled, phase 3 trial was conducted at 54 hospitals in nine European countries. Patients aged 60 years and older who were newly diagnosed with acute myeloid leukaemia and had not yet been treated were enrolled if they had an Eastern Cooperative Oncology Group performance status of 2 or less and were eligible for intensive chemotherapy. Patients were randomly assigned (1:1) to receive decitabine or standard chemotherapy (known as 3 + 7). For the decitabine group, decitabine (20 mg/m(2)) was administered for the first 10 days in the first 28-day cycle, followed by 28-day cycles consisting of 5 days or 10 days of decitabine. For the 3 + 7 group, daunorubicin (60 mg/m(2)) was administered over the first 3 days and cytarabine (200 mg/m(2)) over the first 7 days, followed by 1-3 additional chemotherapy cycles. Allogeneic HSCT was strongly encouraged. Overall survival in the intention-to-treat population was the primary endpoint. Safety was assessed in all patients who received the allocated treatment. This trial is registered at ClinicalTrials.gov, NCT02172872, and is closed to new participants. FINDINGS: Between Dec 1, 2014, and Aug 20, 2019, 606 patients were randomly assigned to the decitabine (n=303) or 3 + 7 (n=303) group. Following an interim analysis which showed futility, the IDMC recommended on May 22, 2019, that the study continued as planned considering the risks and benefits for the patients participating in the study. The cutoff date for the final analysis presented here was June 30, 2021. At a median follow-up of 4·0 years (IQR 2·9-4·8), 4-year overall survival was 26% (95% CI 21-32) in the decitabine group versus 30% (24-35, 01 november 2023

Prediction of preterm birth is a difficult task for clinicians. By examining an electrohysterogram, electrical activity of the uterus that can lead to preterm birth can be detected. Since signals associated with uterine activity are difficult to interpret for clinicians without a background in signal processing, machine learning may be a viable solution. We are the first to employ Deep Learning models, a long-short term memory and temporal convolutional network model, on electrohysterography data using the Term-Preterm Electrohysterogram database. We show that end-to-end learning achieves an AUC score of 0.58, which is comparable to machine learning models that use handcrafted features. Moreover, we evaluate the effect of adding clinical data to the model and conclude that adding the available clinical data to electrohysterography data does not result in a gain in performance. Also, we propose an interpretability framework for time series classification that is well-suited to use in case of limited data, as opposed to existing methods that require large amounts of data. Clinicians with extensive work experience as gynaecologist used our framework to provide insights on how to link our results to clinical practice and stress that in order to decrease the number of false positives, a dataset with patients at high risk of preterm birth should be collected. All code is made publicly available.

Encryption on the internet with the shift to HTTPS has been an important step to improve the privacy of internet users. However, there is an increasing body of work about extracting information from encrypted internet traffic without having to decrypt it. Such attacks bypass security guarantees assumed to be given by HTTPS and thus need to be understood. Prior works showed that the variable bitrates of video streams are sufficient to identify which video someone is watching. These works generally have to make trade-offs in aspects such as accuracy, scalability, robustness, etc. These trade-offs complicate the practical use of these attacks. To that end, we propose a deep metric learning framework based on the triplet loss method. Through this framework, we achieve robust, generalisable, scalable and transferable encrypted video stream detection. First, the triplet loss is better able to deal with video streams not seen during training. Second, our approach can accurately classify videos not seen during training. Third, we show that our method scales well to a dataset of over 1000 videos. Finally, we show that a model trained on video streams over Chrome can also classify streams over Firefox. Our results suggest that this side-channel attack is more broadly applicable than originally thought. We provide our code alongside a diverse and up-to-date dataset for future research.

A major problem in cybersecurity research is the correct labeling of up-to-date datasets. It relies on the availability of human experts, and is as such very cumbersome. Motivated by this, two techniques have been proposed for efficient labeling: Active Learning (AL) and Semi-Supervised Learning (SeSL). In this paper, we introduce Plusmine: an intrusion detection method that combines the benefits of AL and SeSL to efficiently automate classification. We develop new techniques for both components. Moreover, we empirically show that Plusmine obtains good and more robust results than benchmark methods.

Item does not contain fulltext

One of the reasons that the deployment of network intrusion detection methods falls short is the lack of realistic labeled datasets, which makes it challenging to develop and compare techniques. It is caused by the large amounts of effort that it takes for a cyber expert to classify network connections. This has raised the need for methods that learn from both labeled and unlabeled data which observations are best to present to the human expert. Hence, Active Learning (AL) methods are of interest. In this paper, we propose a new hybrid AL method called Jasmine. Firstly, it uses the uncertainty score and anomaly score to determine how suitable each observation is for querying, i.e., how likely it is to enhance classification. Secondly, Jasmine introduces dynamic updating. This allows the model to adjust the balance between querying uncertain, anomalous and randomly selected observations. To this end, Jasmine is able to learn the best query strategy during the labeling process. This is in contrast to the other AL methods in cybersecurity that all have static, predetermined query functions. We show that dynamic updating, and therefore Jasmine, is able to consistently obtain good and more robust results than querying only uncertainties, only anomalies or a fixed combination of the two.

Contains fulltext : 243950.pdf (Publisher’s version ) (Open Access), INTRODUCTION: Efficacy of lenalidomide plus rituximab (R-LEN) compared to rituximab monotherapy (R-mono) for patients with previously treated follicular lymphoma (FL) was investigated in AUGMENT (NCT01938001). Our aim was to evaluate the cost-effectiveness of R-LEN versus R-mono in this setting from a Dutch perspective. AREAS COVERED: Cost-effectiveness was assessed through a partitioned survival model from three perspectives (i.e. societal, healthcare, and societal, including future non-medical costs). Patient-level data from AUGMENT informed effectiveness parameters (i.e. long-term survival) and health state utilities. Resource use and prices were based on AUGMENT and the literature. Clinical experts validated efficacy input parameters and results. Uncertainty was explored through sensitivity and scenario analyses. EXPERT OPINION: R-LEN resulted in 1.7 incremental discounted quality-adjusted life years (QALYs). Total incremental discounted costs were 67,161 EUR from a societal perspective. In conclusion, R-LEN was cost-effective at a willingness-to-pay (WTP) threshold of 50,000 EUR/QALY in the base-case analyses(incremental cost-effectiveness ratio = 40,493 EUR/QALY). Scenario and sensitivity analyses indicated some level of uncertainty regarding this conclusion, depending on the chosen WTP-threshold and perspective.

Contains fulltext : 244701.pdf (Publisher’s version ) (Open Access), OBJECTIVES: As coronavirus disease 2019 is a novel disease, treatment strategies continue to be debated. This provides the intensive care community with a unique opportunity as the population of coronavirus disease 2019 patients requiring invasive mechanical ventilation is relatively homogeneous compared with other ICU populations. We hypothesize that the novelty of coronavirus disease 2019 and the uncertainty over its similarity with noncoronavirus disease 2019 acute respiratory distress syndrome resulted in substantial practice variation between hospitals during the first and second waves of coronavirus disease 2019 patients. DESIGN: Multicenter retrospective cohort study. SETTING: Twenty-five hospitals in the Netherlands from February 2020 to July 2020, and 14 hospitals from August 2020 to December 2020. PATIENTS: One thousand two hundred ninety-four critically ill intubated adult ICU patients with coronavirus disease 2019 were selected from the Dutch Data Warehouse. Patients intubated for less than 24 hours, transferred patients, and patients still admitted at the time of data extraction were excluded. MEASUREMENTS AND MAIN RESULTS: We aimed to estimate between-ICU practice variation in selected ventilation parameters (positive end-expiratory pressure, Fio(2), set respiratory rate, tidal volume, minute volume, and percentage of time spent in a prone position) on days 1, 2, 3, and 7 of intubation, adjusted for patient characteristics as well as severity of illness based on Pao(2)/Fio(2) ratio, pH, ventilatory ratio, and dynamic respiratory system compliance during controlled ventilation. Using multilevel linear mixed-effects modeling, we found significant (p ≤ 0.001) variation between ICUs in all ventilation parameters on days 1, 2, 3, and 7 of intubation for both waves. CONCLUSIONS: This is the first study to clearly demonstrate significant practice variation between ICUs related to mechanical ventilation parameters that are under direct control by intensivists.

Contains fulltext : 244124.pdf (Publisher’s version ) (Open Access), BACKGROUND: There has been a cultural shift toward patient engagement in health, with a growing demand from patients to access their results. OBJECTIVE: The Lymphoma Intervention (LIVE) trial is conducted to examine the impact of return of individual patient-reported outcome (PRO) results and a web-based self-management intervention on psychological distress, self-management, satisfaction with information, and health care use in a population-based setting. METHODS: Return of PRO results included comparison with age- and sex-matched peers and was built into the Patient-Reported Outcomes Following Initial Treatment and Long-Term Evaluation of Survivorship registry. The self-management intervention is an adaptation of a fully automated evidence-based intervention for breast cancer survivors. Patients with lymphoma who completed the web-based questionnaire were equally randomized to care as usual, return of PRO results, and return of PRO results plus self-management intervention. Patients completed questionnaires 9 to 18 months after diagnosis (T0; n=227), 4 months (T1; n=190), 12 months (T2; n=170), and 24 months (T3; n=98). RESULTS: Of all invited patients, 51.1% (456/892) responded and web-based participants (n=227) were randomly assigned to care as usual (n=76), return of PRO results (n=74), or return of PRO results and access to Living with lymphoma (n=77). Return of PRO results was viewed by 76.7% (115/150) of those with access. No statistically significant differences were observed for psychological distress, self-management, satisfaction with information provision, and health care use between patients who received PRO results and those who did not (P>.05). Use of the self-management intervention was low (2/76, 3%), and an effect could therefore not be determined. CONCLUSIONS: Return of individual PRO results seems to meet patients' wishes but had no beneficial effects on patient outcome. No negative effects were found when individual PRO results were disclosed

Contains fulltext : 238831.pdf (Publisher’s version ) (Open Access), BACKGROUND: The Coronavirus disease 2019 (COVID-19) pandemic has underlined the urgent need for reliable, multicenter, and full-admission intensive care data to advance our understanding of the course of the disease and investigate potential treatment strategies. In this study, we present the Dutch Data Warehouse (DDW), the first multicenter electronic health record (EHR) database with full-admission data from critically ill COVID-19 patients. METHODS: A nation-wide data sharing collaboration was launched at the beginning of the pandemic in March 2020. All hospitals in the Netherlands were asked to participate and share pseudonymized EHR data from adult critically ill COVID-19 patients. Data included patient demographics, clinical observations, administered medication, laboratory determinations, and data from vital sign monitors and life support devices. Data sharing agreements were signed with participating hospitals before any data transfers took place. Data were extracted from the local EHRs with prespecified queries and combined into a staging dataset through an extract-transform-load (ETL) pipeline. In the consecutive processing pipeline, data were mapped to a common concept vocabulary and enriched with derived concepts. Data validation was a continuous process throughout the project. All participating hospitals have access to the DDW. Within legal and ethical boundaries, data are available to clinicians and researchers. RESULTS: Out of the 81 intensive care units in the Netherlands, 66 participated in the collaboration, 47 have signed the data sharing agreement, and 35 have shared their data. Data from 25 hospitals have passed through the ETL and processing pipeline. Currently, 3464 patients are included in the DDW, both from wave 1 and wave 2 in the Netherlands. More than 200 million clinical data points are available. Overall ICU mortality was 24.4%. Respiratory and hemodynamic parameters were most frequently measured throughout a patient's stay. For each pa

The goals of this study were to examine whether machine-learning algorithms outper-form multivariable logistic regression in the prediction of insufficient response to methotrexate (MTX); secondly, to examine which features are essential for correct prediction; and finally, to in-vestigate whether the best performing model specifically identifies insufficient responders to MTX (combination) therapy. The prediction of insufficient response (3-month Disease Activity Score 28-Erythrocyte-sedimentation rate (DAS28-ESR) > 3.2) was assessed using logistic regression, least absolute shrinkage and selection operator (LASSO), random forest, and extreme gradient boosting (XGBoost). The baseline features of 355 rheumatoid arthritis (RA) patients from the “treatment in the Rotterdam Early Arthritis CoHort” (tREACH) and the U-Act-Early trial were combined for analyses. The model performances were compared using area under the curve (AUC) of receiver operating characteristic (ROC) curves, 95% confidence intervals (95% CI), and sensitivity and specificity. Fi-nally, the best performing model following feature selection was tested on 101 RA patients starting tocilizumab (TCZ)-monotherapy. Logistic regression (AUC = 0.77 95% CI: 0.68–0.86) performed as well as LASSO (AUC = 0.76, 95% CI: 0.67–0.85), random forest (AUC = 0.71, 95% CI: 0.61 = 0.81), and XGBoost (AUC = 0.70, 95% CI: 0.61–0.81), yet logistic regression reached the highest sensitivity (81%). The most important features were baseline DAS28 (components). For all algorithms, models with six features performed similarly to those with 16. When applied to the TCZ-monotherapy group, logistic regression’s sensitivity significantly dropped from 83% to 69% (p = 0.03). In the current dataset, logistic regression performed equally well compared to machine-learning algorithms in the prediction of insufficient response to MTX. Models could be reduced to six features, which are more conducive for clinical implementation. Interestingly

Treatment results of AML in elderly patients are unsatisfactory. We hypothesized that addition of tosedostat, an aminopeptidase inhibitor, to intensive chemotherapy may improve outcome in this population. After establishing a safe dose in a run-in phase of the study in 22 patients, 231 eligible patients with AML above 65 years of age (median 70, range 66–81) were randomly assigned in this open label randomized Phase II study to receive standard chemotherapy (3+7) with or without tosedostat at the selected daily dose of 120 mg (n = 116), days 1–21. In the second cycle, patients received cytarabine 1000 mg/m2 twice daily on days 1-6 with or without tosedostat. CR/CRi rates in the 2 arms were not significantly different (69% (95% C.I. 60–77%) vs 64% (55–73%), respectively). At 24 months, event-free survival (EFS) was 20% for the standard arm versus 12% for the tosedostat arm (Cox-p = 0.01) and overall survival (OS) 33% vs 18% respectively (p = 0.006). Infectious complications accounted for an increased early death rate in the tosedostat arm. Atrial fibrillation wa

Contains fulltext : 244677.pdf (Publisher’s version ) (Open Access), INTRODUCTION: Determining the optimal timing for extubation can be challenging in the intensive care. In this study, we aim to identify predictors for extubation failure in critically ill patients with COVID-19. METHODS: We used highly granular data from 3464 adult critically ill COVID patients in the multicenter Dutch Data Warehouse, including demographics, clinical observations, medications, fluid balance, laboratory values, vital signs, and data from life support devices. All intubated patients with at least one extubation attempt were eligible for analysis. Transferred patients, patients admitted for less than 24 h, and patients still admitted at the time of data extraction were excluded. Potential predictors were selected by a team of intensive care physicians. The primary and secondary outcomes were extubation without reintubation or death within the next 7 days and within 48 h, respectively. We trained and validated multiple machine learning algorithms using fivefold nested cross-validation. Predictor importance was estimated using Shapley additive explanations, while cutoff values for the relative probability of failed extubation were estimated through partial dependence plots. RESULTS: A total of 883 patients were included in the model derivation. The reintubation rate was 13.4% within 48 h and 18.9% at day 7, with a mortality rate of 0.6% and 1.0% respectively. The grandient-boost model performed best (area under the curve of 0.70) and was used to calculate predictor importance. Ventilatory characteristics and settings were the most important predictors. More specifically, a controlled mode duration longer than 4 days, a last fraction of inspired oxygen higher than 35%, a mean tidal volume per kg ideal body weight above 8 ml/kg in the day before extubation, and a shorter duration in assisted mode (< 2 days) compared to their median values. Additionally, a higher C-reactive protein and leukocyte count, a lower thrombocyte count, a lower Glasgow coma scale a

Contains fulltext : 238677.pdf (Publisher’s version ) (Open Access), BACKGROUND: The identification of risk factors for adverse outcomes and prolonged intensive care unit (ICU) stay in COVID-19 patients is essential for prognostication, determining treatment intensity, and resource allocation. Previous studies have determined risk factors on admission only, and included a limited number of predictors. Therefore, using data from the highly granular and multicenter Dutch Data Warehouse, we developed machine learning models to identify risk factors for ICU mortality, ventilator-free days and ICU-free days during the course of invasive mechanical ventilation (IMV) in COVID-19 patients. METHODS: The DDW is a growing electronic health record database of critically ill COVID-19 patients in the Netherlands. All adult ICU patients on IMV were eligible for inclusion. Transfers, patients admitted for less than 24 h, and patients still admitted at time of data extraction were excluded. Predictors were selected based on the literature, and included medication dosage and fluid balance. Multiple algorithms were trained and validated on up to three sets of observations per patient on day 1, 7, and 14 using fivefold nested cross-validation, keeping observations from an individual patient in the same split. RESULTS: A total of 1152 patients were included in the model. XGBoost models performed best for all outcomes and were used to calculate predictor importance. Using Shapley additive explanations (SHAP), age was the most important demographic risk factor for the outcomes upon start of IMV and throughout its course. The relative probability of death across age values is visualized in Partial Dependence Plots (PDPs), with an increase starting at 54 years. Besides age, acidaemia, low P/F-ratios and high driving pressures demonstrated a higher probability of death. The PDP for driving pressure showed a relative probability increase starting at 12 cmH(2)O. CONCLUSION: Age is the most important demographic risk factor of ICU mortality, ICU-free days and ve

Semi-supervised learning can be applied to datasets that contain both labeled and unlabeled instances and can result in more accurate predictions compared to fully supervised or unsupervised learning in case limited labeled data is available. A subclass of problems, called Positive-Unlabeled (PU) learning, focuses on cases in which the labeled instances contain only positive examples. Given the lack of negatively labeled data, estimating the general performance is difficult. In this paper, we propose a new approach to approximate the F1 score for PU learning. It requires an estimate of what fraction of the total number of positive instances is available in the labeled set. We derive theoretical properties of the approach and apply it to several datasets to study its empirical behavior and to compare it to the most well-known score in the field, LL score. Results show that even when the estimate is quite off compared to the real fraction of positive labels the approximation of the F1 score is significantly better compared with the LL score.

The goals of this study were to examine whether machine-learning algorithms outper-form multivariable logistic regression in the prediction of insufficient response to methotrexate (MTX); secondly, to examine which features are essential for correct prediction; and finally, to in-vestigate whether the best performing model specifically identifies insufficient responders to MTX (combination) therapy. The prediction of insufficient response (3-month Disease Activity Score 28-Erythrocyte-sedimentation rate (DAS28-ESR) > 3.2) was assessed using logistic regression, least absolute shrinkage and selection operator (LASSO), random forest, and extreme gradient boosting (XGBoost). The baseline features of 355 rheumatoid arthritis (RA) patients from the “treatment in the Rotterdam Early Arthritis CoHort” (tREACH) and the U-Act-Early trial were combined for analyses. The model performances were compared using area under the curve (AUC) of receiver operating characteristic (ROC) curves, 95% confidence intervals (95% CI), and sensitivity and specificity. Fi-nally, the best performing model following feature selection was tested on 101 RA patients starting tocilizumab (TCZ)-monotherapy. Logistic regression (AUC = 0.77 95% CI: 0.68–0.86) performed as well as LASSO (AUC = 0.76, 95% CI: 0.67–0.85), random forest (AUC = 0.71, 95% CI: 0.61 = 0.81), and XGBoost (AUC = 0.70, 95% CI: 0.61–0.81), yet logistic regression reached the highest sensitivity (81%). The most important features were baseline DAS28 (components). For all algorithms, models with six features performed similarly to those with 16. When applied to the TCZ-monotherapy group, logistic regression’s sensitivity significantly dropped from 83% to 69% (p = 0.03). In the current dataset, logistic regression performed equally well compared to machine-learning algorithms in the prediction of insufficient response to MTX. Models could be reduced to six features, which are more conducive for clinical implementation. Interesti

Treatment results of AML in elderly patients are unsatisfactory. We hypothesized that addition of tosedostat, an aminopeptidase inhibitor, to intensive chemotherapy may improve outcome in this population. After establishing a safe dose in a run-in phase of the study in 22 patients, 231 eligible patients with AML above 65 years of age (median 70, range 66–81) were randomly assigned in this open label randomized Phase II study to receive standard chemotherapy (3+7) with or without tosedostat at the selected daily dose of 120 mg (n = 116), days 1–21. In the second cycle, patients received cytarabine 1000 mg/m2 twice daily on days 1-6 with or without tosedostat. CR/CRi rates in the 2 arms were not significantly different (69% (95% C.I. 60–77%) vs 64% (55–73%), respectively). At 24 months, event-free survival (EFS) was 20% for the standard arm versus 12% for the tosedostat arm (Cox-p = 0.01) and overall survival (OS) 33% vs 18% respectively (p = 0.006). Infectious complications accounted for an increased early death rate in the tosedostat arm. Atrial fibrillation was more common in the tosedostat arm as well. The results of the present study show that the addition of tosedostat to standard chemotherapy does negatively affect the therapeutic outcome of elderly AML patients.