Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal

Development and validation of a risk prediction model for hospital admission in COVID-19 patients presenting to primary care

BACKGROUND

There is a paucity of prognostic models for COVID-19 that are usable for in-office patient assessment in general practice (GP).

OBJECTIVES

To develop and validate a risk prediction model for hospital admission with readily available predictors.

METHODS

A retrospective cohort study linking GP records from 8 COVID-19 centres and 55 general practices in the Netherlands to hospital admission records. The development cohort spanned March to June 2020, the validation cohort March to June 2021. The primary outcome was hospital admission within 14 days. We used geographic leave-region-out cross-validation in the development cohort and temporal validation in the validation cohort.

RESULTS

In the development cohort, 4,806 adult patients with COVID-19 consulted their GP (median age 56, 56% female); in the validation cohort 830 patients did (median age 56, 52% female). In the development and validation cohort respectively, 292 (6.1%) and 126 (15.2%) were admitted to the hospital within 14 days, respectively. A logistic regression model based on sex, smoking, symptoms, vital signs and comorbidities predicted hospital admission with a c-index of 0.84 (95% CI 0.83 to 0.86) at geographic cross-validation and 0.79 (95% CI 0.74 to 0.83) at temporal validation, and was reasonably well calibrated (intercept -0.08, 95% CI -0.98 to 0.52, slope 0.89, 95% CI 0.71 to 1.07 at geographic cross-validation and intercept 0.02, 95% CI -0.21 to 0.24, slope 0.82, 95% CI 0.64 to 1.00 at temporal validation).

CONCLUSION

We derived a risk model using readily available variables at GP assessment to predict hospital admission for COVID-19. It performed accurately across regions and waves. Further validation on cohorts with acquired immunity and newer SARS-CoV-2 variants is recommended.

Longitudinal deep neural networks for assessing metastatic brain cancer on a large open benchmark

The detection and tracking of metastatic cancer over the lifetime of a patient remains a major challenge in clinical trials and real-world care. Advances in deep learning combined with massive datasets may enable the development of tools that can address this challenge. We present NYUMets-Brain, the world's largest, longitudinal, real-world dataset of cancer consisting of the imaging, clinical follow-up, and medical management of 1,429 patients. Using this dataset we developed Segmentation-Through-Time, a deep neural network which explicitly utilizes the longitudinal structure of the data and obtained state-of-the-art results at small (<10 mm3) metastases detection and segmentation. We also demonstrate that the monthly rate of change of brain metastases over time are strongly predictive of overall survival (HR 1.27, 95%CI 1.18-1.38). We are releasing the dataset, codebase, and model weights for other cancer researchers to build upon these results and to serve as a public benchmark.

Sex hormones and the total testosterone:estradiol ratio as predictors of severe acute respiratory syndrome coronavirus 2 infection in hospitalized men

BACKGROUND

The predictive ability of the early determination of sex steroids and the total testosterone:estradiol ratio for the risk of severe coronavirus disease 2019 or the potential existence of a biological gradient in this relationship has not been evaluated.

OBJECTIVES

To assess the relationship of sex steroid levels and the total testosterone:estradiol ratio with the risk of severe acute respiratory syndrome coronavirus 2 infection in men, defined as the need for intensive care unit admission or death, and the predictive ability of each biomarker.

MATERIALS AND METHODS

This was a prospective observational study. We included all consecutive adult men with severe acute respiratory syndrome coronavirus 2 infections in a single center admitted to a general hospital ward or to the intensive care unit. Sex steroids were evaluated at the centralized laboratory of our hospital.

RESULTS

We recruited 98 patients, 54 (55.1%) of whom developed severe coronavirus disease in 2019. Compared to patients with nonsevere coronavirus disease 2019, patients with severe coronavirus disease 2019 had significantly lower serum levels of total testosterone (111 ± 89 vs. 191 ± 143 ng/dL; p < 0.001), dehydroepiandrosterone (1.69 ± 1.26 vs. 2.96 ± 2.64 ng/mL; p < 0.001), and dehydroepiandrosterone sulfate (91.72 ± 76.20 vs. 134.28 ± 98.261 μg/dL; p = 0.009), significantly higher levels of estradiol (64.61 ± 59.35 vs. 33.78 ± 13.78 pg/mL; p = 0.001), and significantly lower total testosterone:estradiol ratio (0.28 ± 0.31 vs. 0.70 ± 0.75; p < 0.001). The lower the serum level of androgen and the lower the total testosterone:estradiol ratio values, the higher the likelihood of developing severe coronavirus disease 2019, with the linear trend in the adjusted analyses being statistically significant for all parameters except for androstenedione (p = 0.064). In the receiver operating characteristic analysis, better predictive performance was shown by the total testosterone:estradiol ratio, with an area under the curve of 0.77 (95% confidence interval 0.68-0.87; p < 0.001).

DISCUSSION AND CONCLUSION

Our results suggest that men with severe acute respiratory syndrome coronavirus 2 infection, decreased androgen levels and increased estradiol levels have a higher likelihood of developing an unfavorable outcome. The total testosterone:estradiol ratio showed the best predictive ability.

Machine learning approaches to predict the need for intensive care unit admission among Iranian COVID-19 patients based on ICD-10: A cross-sectional study

Background & Aim

Timely identification of the patients requiring intensive care unit admission (ICU) could be life-saving. We aimed to compare different machine learning algorithms to predict the requirements for ICU admission in COVID-19 patients.

Methods

We screened all patients with COVID-19 at six academic hospitals in Tehran comprising our study population. A total of 44,112 COVID-19 patients (≥18 years old) were included, among which 7722 patients were hospitalized. We used a Random Forest algorithm to select significant variables. Then, prediction models were developed using the Support Vector Machine, Naıve Bayes, logistic regression, lightGBM, decision tree, and K-Nearest Neighbor algorithms. Sensitivity, specificity, accuracy, F1 score, and receiver operating characteristic-Area Under the Curve (AUC) were used to compare the prediction performance of different models.

Results

Based on random Forest, the following predictors were selected: age, cardiac disease, cough, hypertension, diabetes, influenza & pneumonia, malignancy, and nervous system disease. Age was found to have the strongest association with ICU admission among COVID-19 patients. All six models achieved an AUC greater than 0.60. Naıve Bayes achieved the best predictive performance (AUC = 0.71).

Conclusion

Naïve Bayes and lightGBM demonstrated promising results in predicting ICU admission needs in COVID-19 patients. Machine learning models could help quickly identify high-risk patients upon entry and reduce mortality and morbidity among COVID-19 patients.

Prediction of COVID-19 Hospitalization and Mortality Using Artificial Intelligence

BACKGROUND

COVID-19 has had a substantial influence on healthcare systems, requiring early prognosis for innovative therapies and optimal results, especially in individuals with comorbidities. AI systems have been used by healthcare practitioners for investigating, anticipating, and predicting diseases, through means including medication development, clinical trial analysis, and pandemic forecasting. This study proposes the use of AI to predict disease severity in terms of hospital mortality among COVID-19 patients.

METHODS

A cross-sectional study was conducted at King Abdulaziz University, Saudi Arabia. Data were cleaned by encoding categorical variables and replacing missing quantitative values with their mean. The outcome variable, hospital mortality, was labeled as death = 0 or survival = 1, with all baseline investigations, clinical symptoms, and laboratory findings used as predictors. Decision trees, SVM, and random forest algorithms were employed. The training process included splitting the data set into training and testing sets, performing 5-fold cross-validation to tune hyperparameters, and evaluating performance on the test set using accuracy.

RESULTS

The study assessed the predictive accuracy of outcomes and mortality for COVID-19 patients based on factors such as CRP, LDH, Ferritin, ALP, Bilirubin, D-Dimers, and hospital stay (p-value ≤ 0.05). The analysis revealed that hospital stay, D-Dimers, ALP, Bilirubin, LDH, CRP, and Ferritin significantly influenced hospital mortality (p ≤ 0.0001). The results demonstrated high predictive accuracy, with decision trees achieving 76%, random forest 80%, and support vector machines (SVMs) 82%.

CONCLUSIONS

Artificial intelligence is a tool crucial for identifying early coronavirus infections and monitoring patient conditions. It improves treatment consistency and decision-making via the development of algorithms.

End-to-end reproducible AI pipelines in radiology using the cloud

Artificial intelligence (AI) algorithms hold the potential to revolutionize radiology. However, a significant portion of the published literature lacks transparency and reproducibility, which hampers sustained progress toward clinical translation. Although several reporting guidelines have been proposed, identifying practical means to address these issues remains challenging. Here, we show the potential of cloud-based infrastructure for implementing and sharing transparent and reproducible AI-based radiology pipelines. We demonstrate end-to-end reproducibility from retrieving cloud-hosted data, through data pre-processing, deep learning inference, and post-processing, to the analysis and reporting of the final results. We successfully implement two distinct use cases, starting from recent literature on AI-based biomarkers for cancer imaging. Using cloud-hosted data and computing, we confirm the findings of these studies and extend the validation to previously unseen data for one of the use cases. Furthermore, we provide the community with transparent and easy-to-extend examples of pipelines impactful for the broader oncology field. Our approach demonstrates the potential of cloud resources for implementing, sharing, and using reproducible and transparent AI pipelines, which can accelerate the translation into clinical solutions.

Evaluating variable selection methods for multivariable regression models: A simulation study protocol

Researchers often perform data-driven variable selection when modeling the associations between an outcome and multiple independent variables in regression analysis. Variable selection may improve the interpretability, parsimony and/or predictive accuracy of a model. Yet variable selection can also have negative consequences, such as false exclusion of important variables or inclusion of noise variables, biased estimation of regression coefficients, underestimated standard errors and invalid confidence intervals, as well as model instability. While the potential advantages and disadvantages of variable selection have been discussed in the literature for decades, few large-scale simulation studies have neutrally compared data-driven variable selection methods with respect to their consequences for the resulting models. We present the protocol for a simulation study that will evaluate different variable selection methods: forward selection, stepwise forward selection, backward elimination, augmented backward elimination, univariable selection, univariable selection followed by backward elimination, and penalized likelihood approaches (Lasso, relaxed Lasso, adaptive Lasso). These methods will be compared with respect to false inclusion and/or exclusion of variables, consequences on bias and variance of the estimated regression coefficients, the validity of the confidence intervals for the coefficients, the accuracy of the estimated variable importance ranking, and the predictive performance of the selected models. We consider both linear and logistic regression in a low-dimensional setting (20 independent variables with 10 true predictors and 10 noise variables). The simulation will be based on real-world data from the National Health and Nutrition Examination Survey (NHANES). Publishing this study protocol ahead of performing the simulation increases transparency and allows integrating the perspective of other experts into the study design.

Accuracy of routine laboratory tests to predict mortality and deterioration to severe or critical COVID-19 in people with SARS-CoV-2

BACKGROUND

Identifying patients with COVID-19 disease who will deteriorate can be useful to assess whether they should receive intensive care, or whether they can be treated in a less intensive way or through outpatient care. In clinical care, routine laboratory markers, such as C-reactive protein, are used to assess a person's health status.

OBJECTIVES

To assess the accuracy of routine blood-based laboratory tests to predict mortality and deterioration to severe or critical (from mild or moderate) COVID-19 in people with SARS-CoV-2.

SEARCH METHODS

On 25 August 2022, we searched the Cochrane COVID-19 Study Register, encompassing searches of various databases such as MEDLINE via PubMed, CENTRAL, Embase, medRxiv, and ClinicalTrials.gov. We did not apply any language restrictions.

SELECTION CRITERIA

We included studies of all designs that produced estimates of prognostic accuracy in participants who presented to outpatient services, or were admitted to general hospital wards with confirmed SARS-CoV-2 infection, and studies that were based on serum banks of samples from people. All routine blood-based laboratory tests performed during the first encounter were included. We included any reference standard used to define deterioration to severe or critical disease that was provided by the authors.

DATA COLLECTION AND ANALYSIS

Two review authors independently extracted data from each included study, and independently assessed the methodological quality using the Quality Assessment of Prognostic Accuracy Studies tool. As studies reported different thresholds for the same test, we used the Hierarchical Summary Receiver Operator Curve model for meta-analyses to estimate summary curves in SAS 9.4. We estimated the sensitivity at points on the SROC curves that corresponded to the median and interquartile range boundaries of specificities in the included studies. Direct and indirect comparisons were exclusively conducted for biomarkers with an estimated sensitivity and 95% CI of ≥ 50% at a specificity of ≥ 50%. The relative diagnostic odds ratio was calculated as a summary of the relative accuracy of these biomarkers.

MAIN RESULTS

We identified a total of 64 studies, including 71,170 participants, of which 8169 participants died, and 4031 participants deteriorated to severe/critical condition. The studies assessed 53 different laboratory tests. For some tests, both increases and decreases relative to the normal range were included. There was important heterogeneity between tests and their cut-off values. None of the included studies had a low risk of bias or low concern for applicability for all domains. None of the tests included in this review demonstrated high sensitivity or specificity, or both. The five tests with summary sensitivity and specificity above 50% were: C-reactive protein increase, neutrophil-to-lymphocyte ratio increase, lymphocyte count decrease, d-dimer increase, and lactate dehydrogenase increase. Inflammation For mortality, summary sensitivity of a C-reactive protein increase was 76% (95% CI 73% to 79%) at median specificity, 59% (low-certainty evidence). For deterioration, summary sensitivity was 78% (95% CI 67% to 86%) at median specificity, 72% (very low-certainty evidence). For the combined outcome of mortality or deterioration, or both, summary sensitivity was 70% (95% CI 49% to 85%) at median specificity, 60% (very low-certainty evidence). For mortality, summary sensitivity of an increase in neutrophil-to-lymphocyte ratio was 69% (95% CI 66% to 72%) at median specificity, 63% (very low-certainty evidence). For deterioration, summary sensitivity was 75% (95% CI 59% to 87%) at median specificity, 71% (very low-certainty evidence). For mortality, summary sensitivity of a decrease in lymphocyte count was 67% (95% CI 56% to 77%) at median specificity, 61% (very low-certainty evidence). For deterioration, summary sensitivity of a decrease in lymphocyte count was 69% (95% CI 60% to 76%) at median specificity, 67% (very low-certainty evidence). For the combined outcome, summary sensitivity was 83% (95% CI 67% to 92%) at median specificity, 29% (very low-certainty evidence). For mortality, summary sensitivity of a lactate dehydrogenase increase was 82% (95% CI 66% to 91%) at median specificity, 60% (very low-certainty evidence). For deterioration, summary sensitivity of a lactate dehydrogenase increase was 79% (95% CI 76% to 82%) at median specificity, 66% (low-certainty evidence). For the combined outcome, summary sensitivity was 69% (95% CI 51% to 82%) at median specificity, 62% (very low-certainty evidence). Hypercoagulability For mortality, summary sensitivity of a d-dimer increase was 70% (95% CI 64% to 76%) at median specificity of 56% (very low-certainty evidence). For deterioration, summary sensitivity was 65% (95% CI 56% to 74%) at median specificity of 63% (very low-certainty evidence). For the combined outcome, summary sensitivity was 65% (95% CI 52% to 76%) at median specificity of 54% (very low-certainty evidence). To predict mortality, neutrophil-to-lymphocyte ratio increase had higher accuracy compared to d-dimer increase (RDOR (diagnostic Odds Ratio) 2.05, 95% CI 1.30 to 3.24), C-reactive protein increase (RDOR 2.64, 95% CI 2.09 to 3.33), and lymphocyte count decrease (RDOR 2.63, 95% CI 1.55 to 4.46). D-dimer increase had higher accuracy compared to lymphocyte count decrease (RDOR 1.49, 95% CI 1.23 to 1.80), C-reactive protein increase (RDOR 1.31, 95% CI 1.03 to 1.65), and lactate dehydrogenase increase (RDOR 1.42, 95% CI 1.05 to 1.90). Additionally, lactate dehydrogenase increase had higher accuracy compared to lymphocyte count decrease (RDOR 1.30, 95% CI 1.13 to 1.49). To predict deterioration to severe disease, C-reactive protein increase had higher accuracy compared to d-dimer increase (RDOR 1.76, 95% CI 1.25 to 2.50). The neutrophil-to-lymphocyte ratio increase had higher accuracy compared to d-dimer increase (RDOR 2.77, 95% CI 1.58 to 4.84). Lastly, lymphocyte count decrease had higher accuracy compared to d-dimer increase (RDOR 2.10, 95% CI 1.44 to 3.07) and lactate dehydrogenase increase (RDOR 2.22, 95% CI 1.52 to 3.26).

AUTHORS' CONCLUSIONS

Laboratory tests, associated with hypercoagulability and hyperinflammatory response, were better at predicting severe disease and mortality in patients with SARS-CoV-2 compared to other laboratory tests. However, to safely rule out severe disease, tests should have high sensitivity (> 90%), and none of the identified laboratory tests met this criterion. In clinical practice, a more comprehensive assessment of a patient's health status is usually required by, for example, incorporating these laboratory tests into clinical prediction rules together with clinical symptoms, radiological findings, and patient's characteristics.

Systematic review of methods used in prediction models with recurrent event data

BACKGROUND

Patients who suffer from chronic conditions or diseases are susceptible to experiencing repeated events of the same type (e.g. seizures), termed 'recurrent events'. Prediction models can be used to predict the risk of recurrence so that intervention or management can be tailored accordingly, but statistical methodology can vary. The objective of this systematic review was to identify and describe statistical approaches that have been applied for the development and validation of multivariable prediction models with recurrent event data. A secondary objective was to informally assess the characteristics and quality of analysis approaches used in the development and validation of prediction models of recurrent event data.

METHODS

Searches were run in MEDLINE using a search strategy in 2019 which included index terms and phrases related to recurrent events and prediction models. For studies to be included in the review they must have developed or validated a multivariable clinical prediction model for recurrent event outcome data, specifically modelling the recurrent events and the timing between them. The statistical analysis methods used to analyse the recurrent event data in the clinical prediction model were extracted to answer the primary aim of the systematic review. In addition, items such as the event rate as well as any discrimination and calibration statistics that were used to assess the model performance were extracted for the secondary aim of the review.

RESULTS

A total of 855 publications were identified using the developed search strategy and 301 of these are included in our systematic review. The Andersen-Gill method was identified as the most commonly applied method in the analysis of recurrent events, which was used in 152 (50.5%) studies. This was closely followed by frailty models which were used in 116 (38.5%) included studies. Of the 301 included studies, only 75 (24.9%) internally validated their model(s) and three (1.0%) validated their model(s) in an external dataset.

CONCLUSIONS

This review identified a variety of methods which are used in practice when developing or validating prediction models for recurrent events. The variability of the approaches identified is cause for concern as it indicates possible immaturity in the field and highlights the need for more methodological research to bring greater consistency in approach of recurrent event analysis. Further work is required to ensure publications report all required information and use robust statistical methods for model development and validation.

PROSPERO REGISTRATION

CRD42019116031.

A predictive model to explore risk factors for severe COVID-19

With the rapid spread of the novel coronavirus (COVID-19), a sustained global pandemic has emerged. Globally, the cumulative death toll is in the millions. The rising number of COVID-19 infections and deaths has severely impacted the lives of people worldwide, healthcare systems, and economic development. We conducted a retrospective analysis of the characteristics of COVID-19 patients. This analysis includes clinical features upon initial hospital admission, relevant laboratory test results, and imaging findings. We aimed to identify risk factors for severe illness and to construct a predictive model for assessing the risk of severe COVID-19. We collected and analyzed electronic medical records of confirmed COVID-19 patients admitted to the Affiliated Hospital of Jiangsu University (Zhenjiang, China) between December 18, 2022, and February 28, 2023. According to the WHO diagnostic criteria for the novel coronavirus, we divided the patients into two groups: severe and non-severe, and compared their clinical, laboratory, and imaging data. Logistic regression analysis, the least absolute shrinkage and selection operator (LASSO) regression, and receiver operating characteristic (ROC) curve analysis were used to identify the relevant risk factors for severe COVID-19 patients. Patients were divided into a training cohort and a validation cohort. A nomogram model was constructed using the "rms" package in R software. Among the 346 patients, the severe group exhibited significantly higher respiratory rates, breathlessness, altered consciousness, neutrophil-to-lymphocyte ratio (NLR), and lactate dehydrogenase (LDH) levels compared to the non-severe group. Imaging findings indicated that the severe group had a higher proportion of bilateral pulmonary inflammation and ground-glass opacities compared to the non-severe group. NLR and LDH were identified as independent risk factors for severe patients. The diagnostic performance was maximized when NLR, respiratory rate (RR), and LDH were combined. Based on the statistical analysis results, we developed a COVID-19 severity risk prediction model. The total score is calculated by adding up the scores for each of the twelve independent variables. By mapping the total score to the lowest scale, we can estimate the risk of COVID-19 severity. In addition, the calibration plots and DCA analysis showed that the nomogram had better discrimination power for predicting the severity of COVID-19. Our results showed that the development and validation of the predictive nomogram had good predictive value for severe COVID-19.

Predictors of mortality in severe pneumonia patients: a systematic review and meta-analysis

BACKGROUND

Severe pneumonia has consistently been associated with high mortality. We sought to identify risk factors for the mortality of severe pneumonia to assist in reducing mortality for medical treatment.

METHODS

Electronic databases including PubMed, Web of Science, EMBASE, Cochrane Library, and Scopus were systematically searched till June 1, 2023. All human research were incorporated into the analysis, regardless of language, publication date, or geographical location. To pool the estimate, a mixed-effect model was used. The Newcastle-Ottawa Scale (NOS) was employed for assessing the quality of included studies that were included in the analysis.

RESULTS

In total, 22 studies with a total of 3655 severe pneumonia patients and 1107 cases (30.29%) of death were included in the current meta-analysis. Significant associations were found between age [5.76 years, 95% confidence interval [CI] (3.43, 8.09), P < 0.00001], male gender [odds ratio (OR) = 1.47, 95% CI (1.07, 2.02), P = 0.02], and risk of death from severe pneumonia. The comorbidity of neoplasm [OR = 3.37, 95% CI (1.07, 10.57), P = 0.04], besides the presence of complications such as diastolic hypotension [OR = 2.60, 95% CI (1.45, 4.67), P = 0.001], ALI/ARDS [OR = 3.63, 95% CI (1.78, 7.39), P = 0.0004], septic shock [OR = 9.43, 95% CI (4.39, 20.28), P < 0.00001], MOF [OR = 4.34, 95% CI (2.36, 7.95), P < 0.00001], acute kidney injury [OR = 2.45, 95% CI (1.14, 5.26), P = 0.02], and metabolic acidosis [OR = 5.88, 95% CI (1.51, 22.88), P = 0.01] were associated with significantly higher risk of death among patients with severe pneumonia. Those who died, compared with those who survived, differed on multiple biomarkers on admission including serum creatinine [Scr: + 67.77 mmol/L, 95% CI (47.21, 88.34), P < 0.00001], blood urea nitrogen [BUN: + 6.26 mmol/L, 95% CI (1.49, 11.03), P = 0.01], C-reactive protein [CRP: + 33.09 mg/L, 95% CI (3.01, 63.18), P = 0.03], leukopenia [OR = 2.63, 95% CI (1.34, 5.18), P = 0.005], sodium < 136 mEq/L [OR = 2.63, 95% CI (1.34, 5.18), P = 0.005], albumin [- 5.17 g/L, 95% CI (- 7.09, - 3.25), P < 0.00001], PaO2/FiO2 [- 55.05 mmHg, 95% CI (- 60.11, - 50.00), P < 0.00001], arterial blood PH [- 0.09, 95% CI (- 0.15, - 0.04), P = 0.0005], gram-negative microorganism [OR = 2.56, 95% CI (1.17, 5.62), P = 0.02], and multilobar or bilateral involvement [OR = 3.65, 95% CI (2.70, 4.93), P < 0.00001].

CONCLUSIONS

Older age and male gender might face a greater risk of death in severe pneumonia individuals. The mortality of severe pneumonia may also be significantly impacted by complications such diastolic hypotension, ALI/ARDS, septic shock, MOF, acute kidney injury, and metabolic acidosis, as well as the comorbidity of neoplasm, and laboratory indicators involving Scr, BUN, CRP, leukopenia, sodium, albumin, PaO2/FiO2, arterial blood PH, gram-negative microorganism, and multilobar or bilateral involvement.

SYSTEMATIC REVIEW REGISTRATION

PROSPERO Protocol Number: CRD 42023430684.

White blood cells scattergram as a valuable tool for COVID-19 screening: A multicentric study

INTRODUCTION

New tools have been developed to distinguish the COVID-19 diagnosis from other viral infections presenting similar symptomatology and mitigate the lack of sensitivity of molecular testing. We previously identified a specific "sandglass" aspect on the white blood cells (WBC) scattergram of COVID-19 patients, as a highly reliable COVID-19 screening test (sensitivity: 85.9%, specificity: 83.5% and positive predictive value: 94.3%). We then decided to validate our previous data in a multicentric study.

METHODS

This retrospective study involved 817 patients with flu-like illness, among 20 centers, using the same CBC instrument (XN analyzer, SYSMEX, Japan). After training, one specialist per center independently evaluated, under the same conditions, the presence of the "sandglass" aspect of the WDF scattergram, likely representing plasmacytoid lymphocytes.

RESULTS

Overall, this approach showed sensitivity: 59.0%, specificity: 72.9% and positive predictive value: 77.7%. Sensitivity improved with subgroup analysis, including in patients with lymphopenia (65.2%), patients presenting symptoms for more than 5 days (72.3%) and in patients with ARDS (70.1%). COVID-19 patients with larger plasmacytoid lymphocyte cluster (>15 cells) more often have severe outcomes (70% vs. 15% in the control group).

CONCLUSION

Our findings confirm that the WBC scattergram analysis could be added to a diagnostic algorithm for screening and quickly categorizing symptomatic patients as either COVID-19 probable or improbable, especially during COVID-19 resurgence and overlapping with future influenza epidemics. The observed large size of the plasmacytoid lymphocytes cluster appears to be a hallmark of COVID-19 patients and was indicative of a severe outcome. Furthers studies are ongoing to evaluate the value of the new hematological parameters in combination with WDF analysis.

A COVID-19 specific multiparametric and ECG-based score for the prediction of in-hospital mortality: ELCOVID score

We aimed to develop and validate a COVID-19 specific scoring system, also including some ECG features, to predict all-cause in-hospital mortality at admission. Patients were retrieved from the ELCOVID study (ClinicalTrials.gov identifier: NCT04367129), a prospective, multicenter Italian study enrolling COVID-19 patients between May to September 2020. For the model validation, we randomly selected two-thirds of participants to create a derivation dataset and we used the remaining one-third of participants as the validation set. Over the study period, 1014 hospitalized COVID-19 patients (mean age 74 years, 61% males) met the inclusion criteria and were included in this analysis. During a median follow-up of 12 (IQR 7-22) days, 359 (35%) patients died. Age (HR 2.25 [95%CI 1.72-2.94], p < 0.001), delirium (HR 2.03 [2.14-3.61], p = 0.012), platelets (HR 0.91 [0.83-0.98], p = 0.018), D-dimer level (HR 1.18 [1.01-1.31], p = 0.002), signs of right ventricular strain (RVS) (HR 1.47 [1.02-2.13], p = 0.039) and ECG signs of previous myocardial necrosis (HR 2.28 [1.23-4.21], p = 0.009) were independently associated to in-hospital all-cause mortality. The derived risk-scoring system, namely EL COVID score, showed a moderate discriminatory capacity and good calibration. A cut-off score of ≥ 4 had a sensitivity of 78.4% and 65.2% specificity in predicting all-cause in-hospital mortality. ELCOVID score represents a valid, reliable, sensitive, and inexpensive scoring system that can be used for the prognostication of COVID-19 patients at admission and may allow the earlier identification of patients having a higher mortality risk who may be benefit from more aggressive treatments and closer monitoring.

Evaluation of Effectiveness of Self-Supervised Learning in Chest X-Ray Imaging to Reduce Annotated Images

A significant challenge in machine learning-based medical image analysis is the scarcity of medical images. Obtaining a large number of labeled medical images is difficult because annotating medical images is a time-consuming process that requires specialized knowledge. In addition, inappropriate annotation processes can increase model bias. Self-supervised learning (SSL) is a type of unsupervised learning method that extracts image representations. Thus, SSL can be an effective method to reduce the number of labeled images. In this study, we investigated the feasibility of reducing the number of labeled images in a limited set of unlabeled medical images. The unlabeled chest X-ray (CXR) images were pretrained using the SimCLR framework, and then the representations were fine-tuned as supervised learning for the target task. A total of 2000 task-specific CXR images were used to perform binary classification of coronavirus disease 2019 (COVID-19) and normal cases. The results demonstrate that the performance of pretraining on task-specific unlabeled CXR images can be maintained when the number of labeled CXR images is reduced by approximately 40%. In addition, the performance was significantly better than that obtained without pretraining. In contrast, a large number of pretrained unlabeled images are required to maintain performance regardless of task specificity among a small number of labeled CXR images. In summary, to reduce the number of labeled images using SimCLR, we must consider both the number of images and the task-specific characteristics of the target images.

The Impact of COVID on Health Systems: The Workforce and Telemedicine Perspective

The COVID-19 pandemic significantly strained global health systems, leading to the rapid adoption of telemedicine and changes in workforce management. Previously underused, telemedicine became an essential means of delivering healthcare while adhering to physical distancing guidelines. This transition addressed longstanding barriers like connectivity issues. Simultaneously, the radiology sector innovated by widely implementing remote reading stations, which helped manage exposure risks and conserve human resources. Moreover, the pandemic highlighted the critical role of technological advancements beyond telemedicine, such as the accelerated integration of AI in diagnostics and management. This article examines these comprehensive effects, emphasizing the remote work adaptations and innovations in healthcare systems that have reshaped both healthcare delivery and workforce dynamics during the pandemic.

A Dynamic Prognostic Model for Identifying Vulnerable COVID-19 Patients at High Risk of Rapid Deterioration

PURPOSE

We aimed to validate and, if performance was unsatisfactory, update the previously published prognostic model to predict clinical deterioration in patients hospitalized for COVID-19, using data following vaccine availability.

METHODS

Using electronic health records of patients ≥18 years, with laboratory-confirmed COVID-19, from a large care-delivery network in Massachusetts, USA, from March 2020 to November 2021, we tested the performance of the previously developed prediction model and updated the prediction model by incorporating data after availability of COVID-19 vaccines. We randomly divided data into development (70%) and validation (30%) cohorts. We built a model predicting worsening in a published severity scale in 24 h by LASSO regression and evaluated performance by c-statistic and Brier score.

RESULTS

Our study cohort consisted of 8185 patients (Development: 5730 patients [mean age: 62; 44% female] and Validation: 2455 patients [mean age: 62; 45% female]). The previously published model had suboptimal performance using data after November 2020 (N = 4973, c-statistic = 0.60. Brier score = 0.11). After retraining with the new data, the updated model included 38 predictors including 18 changing biomarkers. Patients hospitalized after Jun 1st, 2021 (when COVID-19 vaccines became widely available in Massachusetts) were younger and had fewer comorbidities than those hospitalized before. The c-statistic and Brier score were 0.77 and 0.13 in the development cohort, and 0.73 and 0.14 in the validation cohort.

CONCLUSION

The characteristics of patients hospitalized for COVID-19 differed substantially over time. We developed a new dynamic model for rapid progression with satisfactory performance in the validation set.

Accelerating AI for science: open data science for science

Aspirations for artificial intelligence (AI) as a catalyst for scientific discovery are growing. High-profile successes deploying AI in domains such as protein folding have highlighted AI's potential to unlock new frontiers of scientific knowledge. However, the pathway from AI innovation to deployment in research is not linear. Those seeking to drive a new wave of scientific progress through the application of AI require a diffusion engine that can enhance AI adoption across disciplines. Lessons from previous waves of technology change, experiences of deploying AI in real-world contexts and an emerging research agenda from the AI for science community suggest a framework for accelerating AI adoption. This framework requires action to build supply chains of ideas between disciplines; rapidly transfer technological capabilities through open research; create AI tools that empower researchers; and embed effective data stewardship. Together, these interventions can cultivate an environment of open data science that deliver the benefits of AI across the sciences.

Development and validation of a model to predict ceiling of care in COVID-19 hospitalized patients

BACKGROUND

Therapeutic ceiling of care is the maximum level of care deemed appropiate to offer to a patient based on their clinical profile and therefore their potential to derive benefit, within the context of the availability of resources. To our knowledge, there are no models to predict ceiling of care decisions in COVID-19 patients or other acute illnesses. We aimed to develop and validate a clinical prediction model to predict ceiling of care decisions using information readily available at the point of hospital admission.

METHODS

We studied a cohort of adult COVID-19 patients who were hospitalized in 5 centres of Catalonia between 2020 and 2021. All patients had microbiologically proven SARS-CoV-2 infection at the time of hospitalization. Their therapeutic ceiling of care was assessed at hospital admission. Comorbidities collected at hospital admission, age and sex were considered as potential factors for predicting ceiling of care. A logistic regression model was used to predict the ceiling of care. The final model was validated internally and externally using a cohort obtained from the Leeds Teaching Hospitals NHS Trust. The TRIPOD Checklist for Prediction Model Development and Validation from the EQUATOR Network has been followed to report the model.

RESULTS

A total of 5813 patients were included in the development cohort, of whom 31.5% were assigned a ceiling of care at the point of hospital admission. A model including age, COVID-19 wave, chronic kidney disease, dementia, dyslipidaemia, heart failure, metastasis, peripheral vascular disease, chronic obstructive pulmonary disease, and stroke or transient ischaemic attack had excellent discrimination and calibration. Subgroup analysis by sex, age group, and relevant comorbidities showed excellent figures for calibration and discrimination. External validation on the Leeds Teaching Hospitals cohort also showed good performance.

CONCLUSIONS

Ceiling of care can be predicted with great accuracy from a patient's clinical information available at the point of hospital admission. Cohorts without information on ceiling of care could use our model to estimate the probability of ceiling of care. In future pandemics, during emergency situations or when dealing with frail patients, where time-sensitive decisions about the use of life-prolonging treatments are required, this model, combined with clinical expertise, could be valuable. However, future work is needed to evaluate the use of this prediction tool outside COVID-19.

The Time-Varying Impact of COVID-19 on the Acute Kidney Disorders: A Historical Matched Cohort Study and Mendelian Randomization Analysis

Background: This study aimed to explore the time-varying impact of COVID-19 on acute kidney disorders, including acute kidney injury and other acute kidney diseases. Methods: From the UK Biobank, 10,121 participants with COVID-19 were matched with up to 3 historically unexposed controls by age, sex, Townsend deprivation index, and the status of hospitalization or receiving critical care. We investigated the association between COVID-19 and incidence of acute kidney disorders, within the first 4 weeks after infection, using conditional and time-varying Cox proportional hazard regression. In addition, one-sample Mendelian randomization, utilizing the polygenic risk score for COVID-19 as an instrumental variable, was conducted to explore the potential causality of the association. Results: In the matched cohort study, we observed a significant association between COVID-19 and acute kidney disorders predominantly within the first 3 weeks. The impact of COVID-19 was time dependent, peaking in the second week (hazard ratio, 12.77; 95% confidence interval, 5.93 to 27.70) and decreasing by the fourth week (hazard ratio, 2.28; 95% confidence interval, 0.75 to 6.93). In subgroup analyses, only moderate to severe COVID-19 cases were associated with acute worsening of renal function in a time-dependent pattern. One-sample Mendelian randomization analyses further showed that COVID-19 might exert a "short-term" causal effect on the risk of acute kidney disorders, primarily confined to the first week after infection. Conclusions: The risk of acute kidney disorders following COVID-19 demonstrates a time-varying pattern. Hazard effects were observed only in patients with moderate or severe but not mild COVID-19.

Modeling for Prediction of Mortality Based on past Medical History in Hospitalized COVID-19 Patients: A Secondary Analysis

Introduction

Although COVID-19 is not currently a public health emergency, it will affect susceptible individuals in the post-COVID-19 era. Hence, the present study aimed to develop a model for Iranian patients to identify at-risk groups based on past medical history (PMHx) and some other factors affecting the death of patients hospitalized with COVID-19.

Methods

A secondary study was conducted with the existing data of hospitalized COVID-19 adult patients in the hospitals covered by Iran University of Medical Sciences. PMHx was extracted from the registered ICD-10 codes. Stepwise logistic regression was used to predict mortality by PMHx and background covariates such as intensive care unit (ICU) admission. Crude population attributable fraction (PAF) as well as crude and adjusted odds ratio (OR) with 95% confidence interval (CI) were reported.

Results

A total of 8879 patients were selected with 19.68% mortality. Infectious and parasitic diseases' history showed the greatest association (OR = 5.72, 95% CI: 4.20, 7.82), while the greatest PAF was for cardiovascular system diseases (20.46%). According to logistic regression modeling, the largest effect, other than ICU admission and age, was for history of infectious and parasitic diseases (OR = 3.089, 95% CI: 2.13, 4.47). A good performance was achieved (area under curve = 0.875).

Conclusion

Considering the prevalence of underlying diseases, many mortality cases of COVID-19 are attributable to the history of cardiovascular disease. Future studies are needed for policy making regarding reduction of COVID-19 mortality in susceptible groups in the post-COVID-19 era.

Differential privacy preserved federated learning for prognostic modeling in COVID-19 patients using large multi-institutional chest CT dataset

BACKGROUND

Notwithstanding the encouraging results of previous studies reporting on the efficiency of deep learning (DL) in COVID-19 prognostication, clinical adoption of the developed methodology still needs to be improved. To overcome this limitation, we set out to predict the prognosis of a large multi-institutional cohort of patients with COVID-19 using a DL-based model.

PURPOSE

This study aimed to evaluate the performance of deep privacy-preserving federated learning (DPFL) in predicting COVID-19 outcomes using chest CT images.

METHODS

After applying inclusion and exclusion criteria, 3055 patients from 19 centers, including 1599 alive and 1456 deceased, were enrolled in this study. Data from all centers were split (randomly with stratification respective to each center and class) into a training/validation set (70%/10%) and a hold-out test set (20%). For the DL model, feature extraction was performed on 2D slices, and averaging was performed at the final layer to construct a 3D model for each scan. The DensNet model was used for feature extraction. The model was developed using centralized and FL approaches. For FL, we employed DPFL approaches. Membership inference attack was also evaluated in the FL strategy. For model evaluation, different metrics were reported in the hold-out test sets. In addition, models trained in two scenarios, centralized and FL, were compared using the DeLong test for statistical differences.

RESULTS

The centralized model achieved an accuracy of 0.76, while the DPFL model had an accuracy of 0.75. Both the centralized and DPFL models achieved a specificity of 0.77. The centralized model achieved a sensitivity of 0.74, while the DPFL model had a sensitivity of 0.73. A mean AUC of 0.82 and 0.81 with 95% confidence intervals of (95% CI: 0.79-0.85) and (95% CI: 0.77-0.84) were achieved by the centralized model and the DPFL model, respectively. The DeLong test did not prove statistically significant differences between the two models (p-value = 0.98). The AUC values for the inference attacks fluctuate between 0.49 and 0.51, with an average of 0.50 ± 0.003 and 95% CI for the mean AUC of 0.500 to 0.501.

CONCLUSION

The performance of the proposed model was comparable to centralized models while operating on large and heterogeneous multi-institutional datasets. In addition, the model was resistant to inference attacks, ensuring the privacy of shared data during the training process.

Utility of Cardiac Troponin-I in the Prediction of In-Hospital Mortality of Patients With COVID-19: A Retrospective Cohort Study in Central India

Laboratory tests have been used as prognostic markers in various diseases, especially those with infectious etiology, but the information on the role of biochemical parameters in the risk assessment of patients with COVID-19 is limited. We designed this retrospective cohort study to investigate the utility of troponin-I in predicting the in-hospital mortality of patients with COVID-19 admitted to our tertiary care hospital in central India. We strategically recorded the history, findings on physical examination, comorbid conditions, clinical diagnosis, results of the biochemical parameters, and adverse outcomes (in terms of survival or death) in order to assess the utility of troponin-I estimation done within the first 24 hours of admission in predicting the in-hospital mortality of patients with COVID-19. Appropriate statistical methods were used depending on the data generated to justify the aim of our study. P-values ​​less than 0.05 were considered significant. We observed a statistically higher (p=0.004) prevalence of mortality in the patients with higher troponin-I levels. We also observed a statistically significant association of other biochemical parameters with the mortality of these patients. Our study highlights the utility of troponin-I in predicting the in-hospital mortality of patients with COVID-19.

External Validation and Update of the Risk Prediction Model for Denosumab-Induced Hypocalcemia Developed From a Hospital-Based Administrative Database

PURPOSE

Denosumab is used to treat patients with bone metastasis from solid tumors, but sometimes causes severe hypocalcemia, so careful clinical management is important. This study aims to externally validate our previously developed risk prediction model for denosumab-induced hypocalcemia by using data from two facilities with different characteristics in Japan and to develop an updated model with improved performance and generalizability.

METHODS

In the external validation, retrospective data of Kameda General Hospital (KGH) and Miyagi Cancer Center (MCC) between June 2013 and June 2022 were used and receiver operating characteristic (ROC)-AUC was mainly evaluated. A scoring-based updated model was developed using the same data set from a hospital-based administrative database as previously employed. Selection of variables related to prediction of hypocalcemia was based on the results of external validation.

RESULTS

For the external validation, data from 235 KGH patients and 224 MCC patients were collected. ROC-AUC values in the original model were 0.879 and 0.774, respectively. The updated model consisting of clinical laboratory tests (calcium, albumin, and alkaline phosphatase) afforded similar ROC-AUC values in the two facilities (KGH, 0.837; MCC, 0.856).

CONCLUSION

We developed an updated risk prediction model for denosumab-induced hypocalcemia with small interfacility differences. Our results indicate the importance of using data from plural facilities with different characteristics in the external validation of generalized prediction models and may be generally relevant to the clinical application of risk prediction models. Our findings are expected to contribute to improved management of bone metastasis treatment.

Resilient Artificial Intelligence in Health: Synthesis and Research Agenda Toward Next-Generation Trustworthy Clinical Decision Support

Artificial intelligence (AI)-based clinical decision support systems are gaining momentum by relying on a greater volume and variety of secondary use data. However, the uncertainty, variability, and biases in real-world data environments still pose significant challenges to the development of health AI, its routine clinical use, and its regulatory frameworks. Health AI should be resilient against real-world environments throughout its lifecycle, including the training and prediction phases and maintenance during production, and health AI regulations should evolve accordingly. Data quality issues, variability over time or across sites, information uncertainty, human-computer interaction, and fundamental rights assurance are among the most relevant challenges. If health AI is not designed resiliently with regard to these real-world data effects, potentially biased data-driven medical decisions can risk the safety and fundamental rights of millions of people. In this viewpoint, we review the challenges, requirements, and methods for resilient AI in health and provide a research framework to improve the trustworthiness of next-generation AI-based clinical decision support.

Predictive models of Alzheimer's disease dementia risk in older adults with mild cognitive impairment: a systematic review and critical appraisal

BACKGROUND

Mild cognitive impairment has received widespread attention as a high-risk population for Alzheimer's disease, and many studies have developed or validated predictive models to assess it. However, the performance of the model development remains unknown.

OBJECTIVE

The objective of this review was to provide an overview of prediction models for the risk of Alzheimer's disease dementia in older adults with mild cognitive impairment.

METHOD

PubMed, EMBASE, Web of Science, and MEDLINE were systematically searched up to October 19, 2023. We included cohort studies in which risk prediction models for Alzheimer's disease dementia in older adults with mild cognitive impairment were developed or validated. The Predictive Model Risk of Bias Assessment Tool (PROBAST) was employed to assess model bias and applicability. Random-effects models combined model AUCs and calculated (approximate) 95% prediction intervals for estimations. Heterogeneity across studies was evaluated using the I2 statistic, and subgroup analyses were conducted to investigate sources of heterogeneity. Additionally, funnel plot analysis was utilized to identify publication bias.

RESULTS

The analysis included 16 studies involving 9290 participants. Frequency analysis of predictors showed that 14 appeared at least twice and more, with age, functional activities questionnaire, and Mini-mental State Examination scores of cognitive functioning being the most common predictors. From the studies, only two models were externally validated. Eleven studies ultimately used machine learning, and four used traditional modelling methods. However, we found that in many of the studies, there were problems with insufficient sample sizes, missing important methodological information, lack of model presentation, and all of the models were rated as having a high or unclear risk of bias. The average AUC of the 15 best-developed predictive models was 0.87 (95% CI: 0.83, 0.90).

DISCUSSION

Most published predictive modelling studies are deficient in rigour, resulting in a high risk of bias. Upcoming research should concentrate on enhancing methodological rigour and conducting external validation of models predicting Alzheimer's disease dementia. We also emphasize the importance of following the scientific method and transparent reporting to improve the accuracy, generalizability and reproducibility of study results.

REGISTRATION

This systematic review was registered in PROSPERO (Registration ID: CRD42023468780).

Viral and Host Factors Are Associated With Mortality in Hospitalized Patients With COVID-19

BACKGROUND

Persistent mortality in adults hospitalized due to acute COVID-19 justifies pursuit of disease mechanisms and potential therapies. The aim was to evaluate which virus and host response factors were associated with mortality risk among participants in Therapeutics for Inpatients with COVID-19 (TICO/ACTIV-3) trials.

METHODS

A secondary analysis of 2625 adults hospitalized for acute SARS-CoV-2 infection randomized to 1 of 5 antiviral products or matched placebo in 114 centers on 4 continents. Uniform, site-level collection of participant baseline clinical variables was performed. Research laboratories assayed baseline upper respiratory swabs for SARS-CoV-2 viral RNA and plasma for anti-SARS-CoV-2 antibodies, SARS-CoV-2 nucleocapsid antigen (viral Ag), and interleukin-6 (IL-6). Associations between factors and time to mortality by 90 days were assessed using univariate and multivariable Cox proportional hazards models.

RESULTS

Viral Ag ≥4500 ng/L (vs <200 ng/L; adjusted hazard ratio [aHR], 2.07; 1.29-3.34), viral RNA (<35 000 copies/mL [aHR, 2.42; 1.09-5.34], ≥35 000 copies/mL [aHR, 2.84; 1.29-6.28], vs below detection), respiratory support (<4 L O2 [aHR, 1.84; 1.06-3.22]; ≥4 L O2 [aHR, 4.41; 2.63-7.39], or noninvasive ventilation/high-flow nasal cannula [aHR, 11.30; 6.46-19.75] vs no oxygen), renal impairment (aHR, 1.77; 1.29-2.42), and IL-6 >5.8 ng/L (aHR, 2.54 [1.74-3.70] vs ≤5.8 ng/L) were significantly associated with mortality risk in final adjusted analyses. Viral Ag, viral RNA, and IL-6 were not measured in real-time.

CONCLUSIONS

Baseline virus-specific, clinical, and biological variables are strongly associated with mortality risk within 90 days, revealing potential pathogen and host-response therapeutic targets for acute COVID-19 disease.

A precise performance-based reimbursement model for the multi-centre NAPKON cohorts - development and evaluation

Fair allocation of funding in multi-centre clinical studies is challenging. Models commonly used in Germany - the case fees ("fixed-rate model", FRM) and up-front staffing and consumables ("up-front allocation model", UFAM) lack transparency and fail to suitably accommodate variations in centre performance. We developed a performance-based reimbursement model (PBRM) with automated calculation of conducted activities and applied it to the cohorts of the National Pandemic Cohort Network (NAPKON) within the Network of University Medicine (NUM). The study protocol activities, which were derived from data management systems, underwent validation through standardized quality checks by multiple stakeholders. The PBRM output (first funding period) was compared among centres and cohorts, and the cost-efficiency of the models was evaluated. Cases per centre varied from one to 164. The mean case reimbursement differed among the cohorts (1173.21€ [95% CI 645.68-1700.73] to 3863.43€ [95% CI 1468.89-6257.96]) and centres and mostly fell short of the expected amount. Model comparisons revealed higher cost-efficiency of the PBRM compared to FRM and UFAM, especially for low recruitment outliers. In conclusion, we have developed a reimbursement model that is transparent, accurate, and flexible. In multi-centre collaborations where heterogeneity between centres is expected, a PBRM could be used as a model to address performance discrepancies.Trial registration: https://clinicaltrials.gov/ct2/show/NCT04768998 ; https://clinicaltrials.gov/ct2/show/NCT04747366 ; https://clinicaltrials.gov/ct2/show/NCT04679584 .

Rapid Development of a Registry to Accelerate COVID-19 Vaccine Clinical Trials

Background

The unprecedented scientific response to the SARS-Cov-2 pandemic in 2020 required the rapid development and activation of extensive clinical trial networks to study vaccines and therapeutics. The COVID-19 Prevention Network (CoVPN) coordinated hundreds of sites conducting phase 2 and 3 clinical trials of vaccines and antibody therapeutics. To facilitate these clinical trials, the CoVPN Volunteer Screening Registry (VSR) was created to collect volunteer information at scale, identify volunteers at risk of COVID-19 who met enrollment criteria, distribute candidates across clinical trial sites, and enable monitoring of volunteering and enrollment progress.

Methods

We developed a secure database to support three primary web-based interfaces: a national volunteer questionnaire intake form, a clinical trial site portal, and an Administrative Portal. The Site Portal supported filters based on volunteer attributes, visual analytics, enrollment status tracking, geographic search, and clinical risk prediction. The Administrative Portal supported oversight and development with pre-specified reports aggregated by geography, trial, and trial site; charts of volunteer rates over time; volunteer risk score calculation; and dynamic, user-defined reports.

Findings

Over 650,000 volunteers joined the VSR, and 1094 users were trained to utilize the system. The VSR played a key role in recruitment for the Moderna, Oxford-AstraZeneca, Janssen, and Novavax vaccine clinical trials, provided support to the Pfizer and Sanofi vaccine and prophylactic antibody clinical trials, and enhanced the diversity of trial participants. Clinical trial sites selected 166,729 volunteer records for follow-up screening, and of these 47·7% represented groups prioritized for increased enrollment. Despite the unprecedented urgency of its development, the system maintained 99·99% uptime.

Interpretation

The success of the VSR demonstrates that information tools can be rapidly yet safely developed through a public-private partnership and integrated into a distributed and accelerated clinical trial setting. We further summarize the requirements, design, and development of the system, and discuss lessons learned for future pandemic preparedness.

Palliative prognostic scores for survival prediction of cancer patients: a systematic review and meta-analysis

BACKGROUND

The palliative prognostic score is the most widely validated prognostic tool for cancer survival prediction, with modified versions available. A systematic evaluation of palliative prognostic score tools is lacking. This systematic review and meta-analysis aimed to evaluate the performance and prognostic utility of palliative prognostic score, delirium-palliative prognostic score, and palliative prognostic score without clinician prediction in predicting 30-day survival of cancer patients and to compare their performance.

METHODS

Six databases were searched for peer-reviewed studies and grey literature published from inception to June 2, 2023. English studies must assess palliative prognostic score, delirium-palliative prognostic score, or palliative prognostic score without clinician-predicted survival for 30-day survival in adults aged 18 years and older with any stage or type of cancer. Outcomes were pooled using the random effects model or summarized narratively when meta-analysis was not possible.

RESULTS

A total of 39 studies (n = 10 617 patients) were included. Palliative prognostic score is an accurate prognostic tool (pooled area under the curve [AUC] = 0.82, 95% confidence interval [CI] = 0.79 to 0.84) and outperforms palliative prognostic score without clinician-predicted survival (pooled AUC = 0.74, 95% CI = 0.71 to 0.78), suggesting that the original palliative prognostic score should be preferred. The meta-analysis found palliative prognostic score and delirium-palliative prognostic score performance to be comparable. Most studies reported survival probabilities corresponding to the palliative prognostic score risk groups, and higher risk groups were statistically significantly associated with shorter survival.

CONCLUSIONS

Palliative prognostic score is a validated prognostic tool for cancer patients that can enhance clinicians' confidence and accuracy in predicting survival. Future studies should investigate if accuracy differs depending on clinician characteristics. Reporting of validation studies must be improved, as most studies were at high risk of bias, primarily because calibration was not assessed.

Basic host response parameters to classify mortality risk in COVID-19 and community-acquired pneumonia

Improved phenotyping in pneumonia is necessary to strengthen risk assessment. Via a feasible and multidimensional approach with basic parameters, we aimed to evaluate the effect of host response at admission on severity stratification in COVID-19 and community-acquired pneumonia (CAP). Three COVID-19 and one CAP multicenter cohorts including hospitalized patients were recruited. Three easily available variables reflecting different pathophysiologic mechanisms-immune, inflammation, and respiratory-were selected (absolute lymphocyte count [ALC], C-reactive protein [CRP] and, SpO2/FiO2). In-hospital mortality and intensive care unit (ICU) admission were analyzed as outcomes. A multivariable, penalized maximum likelihood logistic regression was performed with ALC (< 724 lymphocytes/mm3), CRP (> 60 mg/L), and, SpO2/FiO2 (< 450). A total of 1452, 1222 and 462 patients were included in the three COVID-19 and 1292 in the CAP cohort for the analysis. Mortality ranged between 4 and 32% (0 to 3 abnormal biomarkers) and 0-9% in SARS-CoV-2 pneumonia and CAP, respectively. In the first COVID-19 cohort, adjusted for age and sex, we observed an increased odds ratio for in-hospital mortality in COVID-19 with elevated biomarkers altered (OR 1.8, 3, and 6.3 with 1, 2, and 3 abnormal biomarkers, respectively). The model had an AUROC of 0.83. Comparable findings were found for ICU admission, with an AUROC of 0.76. These results were confirmed in the other COVID-19 cohorts Similar OR trends were reported in the CAP cohort; however, results were not statistically significant. Assessing the host response via accessible biomarkers is a simple and rapidly applicable approach for pneumonia.

The dynamics and outcomes of AKI progression during the COVID-19 pandemic

PURPOSE

Acute kidney injury (AKI) associated with COVID-19 is associated with poor prognosis. This study assessed the hitherto uninvestigated impact of COVID-19 on the progression and clinical outcomes of patients with AKI.

METHODS

Data from 576 patients with AKI admitted between 13/3/20 and 13/5/20 were studied. Increasingly complex analyses, from logistic regressions to competing-risk and multi-state models, have revealed insights into AKI progression dynamics associated with PCR-confirmed COVID-19 acquisition and death. Meta-analyses of case fatality ratios among patients with AKI were also conducted.

RESULTS

The overall case-fatality ratio was 0.33 [95% CI (0.20-0.36)]; higher in COVID-19 positive (COVID+) patients 0.52 [95% CI (0.46-0.58)] than in their negative (COVID-) counterparts 0.16 [95% CI (0.12-0.20)]. In AKI Stage-3 patients, that was 0.71 [95% CI (0.64-0.79)] among COVID+ patients with 45% dead within 14 days and 0.35 [95% CI (0.25-0.44)] in the COVID- group and 28% died within 14 days. Among patients diagnosed with AKI Stage-1 within 24 h, the probability of progression to AKI Stage-3 on day 7 post admission was 0.22 [95% CI (0.17-0.27)] among COVID+ patients, and 0.06 [95% CI (0.03, 0.09)] among those who tested negative. The probability of discharge by day 7 was 0.71 [95% CI (0.66, 0.75)] in COVID- patients, and 0.27 [95% CI (0.21, 0.32)] in COVID+ patients. By day 14, in AKI Stage-3 COVID+ patients, that was 0.35 [95% CI (0.25, 0.44)] with little change by day 10, that is, 0.38 [95% CI (0.29, 0.47)].

CONCLUSION

These results are consistent with either a rapid progression in severity, prolonged hospital care, or high case fatality ratio among AKI Stage-3 patients, significantly exacerbated by COVID-19 infection.

The Short and Long of COVID-19: A Review of Acute and Chronic Radiologic Pulmonary Manifestations of SARS-2-CoV and Their Clinical Significance

Coronavirus disease 2019 (COVID-19) pneumonia has had catastrophic effects worldwide. Radiology, in particular computed tomography (CT) imaging, has proven to be valuable in the diagnosis, prognostication, and longitudinal assessment of those diagnosed with COVID-19 pneumonia. This article will review acute and chronic pulmonary radiologic manifestations of COVID-19 pneumonia with an emphasis on CT and also highlighting histopathology, relevant clinical details, and some notable challenges when interpreting the literature.

Liability of Health Professionals Using Sensors, Telemedicine and Artificial Intelligence for Remote Healthcare

In the last few decades, there has been an ongoing transformation of our healthcare system with larger use of sensors for remote care and artificial intelligence (AI) tools. In particular, sensors improved by new algorithms with learning capabilities have proven their value for better patient care. Sensors and AI systems are no longer only non-autonomous devices such as the ones used in radiology or surgical robots; there are novel tools with a certain degree of autonomy aiming to largely modulate the medical decision. Thus, there will be situations in which the doctor is the one making the decision and has the final say and other cases in which the doctor might only apply the decision presented by the autonomous device. As those are two hugely different situations, they should not be treated the same way, and different liability rules should apply. Despite a real interest in the promise of sensors and AI in medicine, doctors and patients are reluctant to use it. One important reason is a lack clear definition of liability. Nobody wants to be at fault, or even prosecuted, because they followed the advice from an AI system, notably when it has not been perfectly adapted to a specific patient. Fears are present even with simple sensors and AI use, such as during telemedicine visits based on very useful, clinically pertinent sensors; with the risk of missing an important parameter; and, of course, when AI appears "intelligent", potentially replacing the doctors' judgment. This paper aims to provide an overview of the liability of the health professional in the context of the use of sensors and AI tools in remote healthcare, analyzing four regimes: the contract-based approach, the approach based on breach of duty to inform, the fault-based approach, and the approach related to the good itself. We will also discuss future challenges and opportunities in the promising domain of sensors and AI use in medicine.

Development of a long noncoding RNA-based machine learning model to predict COVID-19 in-hospital mortality

Tools for predicting COVID-19 outcomes enable personalized healthcare, potentially easing the disease burden. This collaborative study by 15 institutions across Europe aimed to develop a machine learning model for predicting the risk of in-hospital mortality post-SARS-CoV-2 infection. Blood samples and clinical data from 1286 COVID-19 patients collected from 2020 to 2023 across four cohorts in Europe and Canada were analyzed, with 2906 long non-coding RNAs profiled using targeted sequencing. From a discovery cohort combining three European cohorts and 804 patients, age and the long non-coding RNA LEF1-AS1 were identified as predictive features, yielding an AUC of 0.83 (95% CI 0.82-0.84) and a balanced accuracy of 0.78 (95% CI 0.77-0.79) with a feedforward neural network classifier. Validation in an independent Canadian cohort of 482 patients showed consistent performance. Cox regression analysis indicated that higher levels of LEF1-AS1 correlated with reduced mortality risk (age-adjusted hazard ratio 0.54, 95% CI 0.40-0.74). Quantitative PCR validated LEF1-AS1's adaptability to be measured in hospital settings. Here, we demonstrate a promising predictive model for enhancing COVID-19 patient management.

Association Between Artificial Intelligence Based Chest Computed Tomography and Clinical/Laboratory Characteristics with Severity and Mortality in COVID-19 Hospitalized Patients

Background

Some patients with COVID-19 rapidly develop respiratory failure or mortality, underscoring the necessity for early identification of those prone to severe illness. Numerous studies focus on clinical and lab traits, but only few attend to chest computed tomography. The current study seeks to numerically quantify pulmonary lesions using early-phase CT scans calculated through artificial intelligence algorithms in conjunction with clinical and laboratory helps clinicians to early identify the development of severe illness and death in a group of COVID-19 patients.

Methods

From December 15, 2022, to January 30, 2023, 191 confirmed COVID-19 patients admitted to Xinhua Hospital Affiliated with Shanghai Jiao Tong University School of Medicine were consecutively enrolled. All patients underwent chest CT scans and serum tests within 48 hours prior to admission. Variables significantly linked to critical illness or mortality in univariate analysis were subjected to multivariate logistic regression models post collinearity assessment. Adjusted odds ratio, 95% confidence intervals, sensitivity, specificity, Youden index, receiver-operator-characteristics (ROC) curves, and area under the curve (AUC) were computed for predicting severity and in-hospital mortality.

Results

Multivariate logistic analysis revealed that myoglobin (OR = 1.003, 95% CI 1.001-1.005), APACHE II score (OR = 1.387, 95% CI 1.216-1.583), and the infected CT region percentage (OR = 113.897, 95% CI 4.939-2626.496) independently correlated with in-hospital COVID-19 mortality. Prealbumin stood as an independent safeguarding factor (OR = 0.965, 95% CI 0.947-0.984). Neutrophil counts (OR = 1.529, 95% CI 1.131-2.068), urea nitrogen (OR = 1.587, 95% CI 1.222-2.062), SOFA score(OR = 3.333, 95% CI 1.476-7.522), qSOFA score(OR = 15.197, 95% CI 3.281-70.384), PSI score(OR = 1.053, 95% CI 1.018-1.090), and the infected CT region percentage (OR = 548.221, 95% CI 2.615-114,953.586) independently linked to COVID-19 patient severity.

COVID-19 patient profiles over four waves in Barcelona metropolitan area: A clustering approach

OBJECTIVES

Identifying profiles of hospitalized COVID-19 patients and explore their association with different degrees of severity of COVID-19 outcomes (i.e. in-hospital mortality, ICU assistance, and invasive mechanical ventilation). The findings of this study could inform the development of multiple care intervention strategies to improve patient outcomes.

METHODS

Prospective multicentre cohort study during four different waves of COVID-19 from March 1st, 2020 to August 31st, 2021 in four health consortiums within the southern Barcelona metropolitan region. From a starting point of over 292 demographic characteristics, comorbidities, vital signs, severity scores, and clinical analytics at hospital admission, we used both clinical judgment and supervised statistical methods to reduce to the 36 most informative completed covariates according to the disease outcomes for each wave. Patients were then grouped using an unsupervised semiparametric method (KAMILA). Results were interpreted by clinical and statistician team consensus to identify clinically-meaningful patient profiles.

RESULTS

The analysis included nw1 = 1657, nw2 = 697, nw3 = 677, and nw4 = 787 hospitalized-COVID-19 patients for each of the four waves. Clustering analysis identified 2 patient profiles for waves 1 and 3, while 3 profiles were determined for waves 2 and 4. Patients allocated in those groups showed a different percentage of disease outcomes (e.g., wave 1: 15.9% (Cluster 1) vs. 31.8% (Cluster 2) for in-hospital mortality rate). The main factors to determine groups were the patient's age and number of obese patients, number of comorbidities, oxygen support requirement, and various severity scores. The last wave is also influenced by the massive incorporation of COVID-19 vaccines.

CONCLUSION

Our study suggests that a single care model at hospital admission may not meet the needs of hospitalized-COVID-19 adults. A clustering approach appears to be appropriate for helping physicians to differentiate patients and, thus, apply multiple care intervention strategies, as another way of responding to new outbreaks of this or future diseases.

Evaluating Health Care Professionals' Readiness for e-Health Adoption in the Context of the COVID-19 Pandemic: A Georgian Perspective

Background: The COVID-19 pandemic has accelerated the adoption of Electronic health (e-Health), leveraging technologies such as telemedicine, electronic health records, artificial intelligence, and patient engagement platforms. This transformation underscores e-Health's role in providing efficient, patient-centered care. Our study explores health care professionals' readiness for these technologies, emphasizing the need for tailored education in this evolving landscape. Methods: In our study, conducted between February and March 2023, we administered a questionnaire-based survey to 500 staff members (82.4% female, 17.6% male) aged 25-70 from medical universities in Tbilisi, Georgia. The structured questionnaire covered topics such as computer literacy, telemedicine awareness, patient data security, and ethical considerations. We employed SPSS v21.0 for data analysis, encompassing descriptive statistics and thematic analysis of open-ended responses. Results: Our study included 500 participants categorized into five age groups. Notably, 31% considered themselves computer "experts," while 69% rated their skills as "intermediate" or "advanced." Furthermore, 85% used computers professionally, with 33% having practical computer training. Interestingly, 59% expressed interest in information technology training. Regarding e-Health, 15% believed it involves remote communication between health care professionals and patients, while 42% considered it "correct," and 37% "might be correct." Concerning its application in managing patients, opinions varied. In terms of e-Health's integration into Georgia's health care, responses ranged. Regarding patient data safety, participants exhibited diverse views. Finally, opinions on the necessity of informed consent for e-Health applications varied among participants. Conclusions: Our study explores health care professionals' readiness for e-Health adoption during the COVID-19 pandemic. It reveals varying computer literacy levels, a willingness to learn, differing views on e-Health applications, and mixed opinions on its integration into Georgian health care. These findings emphasize the need for clear e-Health terminology, education, tailored approaches, and a focus on data privacy and informed consent. Overall, e-Health's transformative role in modern health care is underscored.

Prediction model protocols indicate better adherence to recommended guidelines for study conduct and reporting

BACKGROUND AND OBJECTIVE

Protocols are invaluable documents for any research study, especially for prediction model studies. However, the mere existence of a protocol is insufficient if key details are omitted. We reviewed the reporting content and details of the proposed design and methods reported in published protocols for prediction model research.

METHODS

We searched MEDLINE, Embase, and the Web of Science Core Collection for protocols for studies developing or validating a diagnostic or prognostic model using any modeling approach in any clinical area. We screened protocols published between Jan 1, 2022 and June 30, 2022. We used the abstract, introduction, methods, and discussion sections of The Transparent Reporting of a multivariable prediction model of Individual Prognosis Or Diagnosis (TRIPOD) statement to inform data extraction.

RESULTS

We identified 30 protocols, of which 28 were describing plans for model development and six for model validation. All protocols were open access, including a preprint. 15 protocols reported prospectively collecting data. 21 protocols planned to use clustered data, of which one-third planned methods to account for it. A planned sample size was reported for 93% development and 67% validation analyses. 16 protocols reported details of study registration, but all protocols reported a statement on ethics approval. Plans for data sharing were reported in 13 protocols.

CONCLUSION

Protocols for prediction model studies are uncommon, and few are made publicly available. Those that are available were reasonably well-reported and often described their methods following current prediction model research recommendations, likely leading to better reporting and methods in the actual study.

Machine learning-aided algorithm design for prediction of severity from clinical, demographic, biochemical and immunological parameters: Our COVID-19 experience from the pandemic

Background

The severity of laboratory and imaging finding was found to be inconsistent with clinical symptoms in COVID-19 patients, thereby increasing casualties. As compared to conventional biomarkers, machine learning algorithms can learn nonlinear and complex interactions and thus improve prediction accuracy. This study aimed at evaluating role of biochemical and immunological parameters-based machine learning algorithms for severity indexing in COVID-19.

Methods

Laboratory biochemical results of 5715 COVID-19 patients were mined from electronic records including 509 admitted in COVID-19 ICU. Random Forest Classifier (RFC), Support Vector Machine (SVM), Naive Bayesian Classifier (NBC) and K-Nearest Neighbours (KNN) classifier models were used. Lasso regression helped in identifying the most influential parameter. A decision tree was made for subdivided data set, based on randomization.

Results

Accuracy of SVM was highest with 94.18% and RFC with 94.04%. SVM had highest PPV (1.00), and NBC had highest NPV (0.95). QUEST modelling ignored age, urea and total protein, and only C-reactive protein and lactate dehydrogenase were considered to be a part of decision-tree algorithm. The overall percentage of correct classification was 78.31% in the overall algorithm with a sensitivity of 87.95% and an AUC of 0.747.

Conclusion

C-reactive protein and lactate dehydrogenase being routinely performed tests in clinical laboratories in peripheral setups, this algorithm could be an effective predictive tool. SVM and RFC models showed significant accuracy in predicting COVID-19 severity and could be useful for future pandemics.

Artificial intelligence in surgery

Artificial intelligence (AI) is rapidly emerging in healthcare, yet applications in surgery remain relatively nascent. Here we review the integration of AI in the field of surgery, centering our discussion on multifaceted improvements in surgical care in the preoperative, intraoperative and postoperative space. The emergence of foundation model architectures, wearable technologies and improving surgical data infrastructures is enabling rapid advances in AI interventions and utility. We discuss how maturing AI methods hold the potential to improve patient outcomes, facilitate surgical education and optimize surgical care. We review the current applications of deep learning approaches and outline a vision for future advances through multimodal foundation models.

Prediction of Morbidity and Mortality After Esophagectomy: A Systematic Review

BACKGROUND

Esophagectomy for esophageal cancer has a complication rate of up to 60%. Prediction models could be helpful to preoperatively estimate which patients are at increased risk of morbidity and mortality. The objective of this study was to determine the best prediction models for morbidity and mortality after esophagectomy and to identify commonalities among the models.

PATIENTS AND METHODS

A systematic review was performed in accordance to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement and was prospectively registered in PROSPERO ( https://www.crd.york.ac.uk/prospero/ , study ID CRD42022350846). Pubmed, Embase, and Clarivate Analytics/Web of Science Core Collection were searched for studies published between 2010 and August 2022. The Prediction model Risk of Bias Assessment Tool was used to assess the risk of bias. Extracted data were tabulated and a narrative synthesis was performed.

RESULTS

Of the 15,011 articles identified, 22 studies were included using data from tens of thousands of patients. This systematic review included 33 different models, of which 18 models were newly developed. Many studies showed a high risk of bias. The prognostic accuracy of models differed between 0.51 and 0.85. For most models, variables are readily available. Two models for mortality and one model for pulmonary complications have the potential to be developed further.

CONCLUSIONS

The availability of rigorous prediction models is limited. Several models are promising but need to be further developed. Some models provide information about risk factors for the development of complications. Performance status is a potential modifiable risk factor. None are ready for clinical implementation.

Exploring the intersection of obesity and gender in COVID-19 outcomes in hospitalized Mexican patients: a comparative analysis of risk profiles using unsupervised machine learning

Introduction

Obesity and gender play a critical role in shaping the outcomes of COVID-19 disease. These two factors have a dynamic relationship with each other, as well as other risk factors, which hinders interpretation of how they influence severity and disease progression. This work aimed to study differences in COVID-19 disease outcomes through analysis of risk profiles stratified by gender and obesity status.

Methods

This study employed an unsupervised clustering analysis, using Mexico's national COVID-19 hospitalization dataset, which contains demographic information and health outcomes of patients hospitalized due to COVID-19. Patients were segmented into four groups by obesity and gender, with participants' attributes and clinical outcome data described for each. Then, Consensus and PAM clustering methods were used to identify distinct risk profiles based on underlying patient characteristics. Risk profile discovery was completed on 70% of records, with the remaining 30% available for validation.

Results

Data from 88,536 hospitalized patients were analyzed. Obesity, regardless of gender, was linked with higher odds of hypertension, diabetes, cardiovascular diseases, pneumonia, and Intensive Care Unit (ICU) admissions. Men tended to have higher frequencies of ICU admissions and pneumonia and higher mortality rates than women. Within each of the four analysis groups (divided based on gender and obesity status), clustering analyses identified four to five distinct risk profiles. For example, among women with obesity, there were four profiles; those with a hypertensive profile were more likely to have pneumonia, and those with a diabetic profile were most likely to be admitted to the ICU.

Conclusion

Our analysis emphasizes the complex interplay between obesity, gender, and health outcomes in COVID-19 hospitalizations. The identified risk profiles highlight the need for personalized treatment strategies for COVID-19 patients and can assist in planning for patterns of deterioration in future waves of SARS-CoV-2 virus transmission. This research underscores the importance of tackling obesity as a major public health concern, given its interplay with many other health conditions, including infectious diseases such as COVID-19.

Accuracy of automated computer-aided risk scoring systems to estimate the risk of COVID-19: a retrospective cohort study

BACKGROUND

In the UK National Health Service (NHS), the patient's vital signs are monitored and summarised into a National Early Warning Score (NEWS) score. A set of computer-aided risk scoring systems (CARSS) was developed and validated for predicting in-hospital mortality and sepsis in unplanned admission to hospital using NEWS and routine blood tests results. We sought to assess the accuracy of these models to predict the risk of COVID-19 in unplanned admissions during the first phase of the pandemic.

METHODS

Adult ( > = 18 years) non-elective admissions discharged (alive/deceased) between 11-March-2020 to 13-June-2020 from two acute hospitals with an index NEWS electronically recorded within ± 24 h of admission. We identified COVID-19 admission based on ICD-10 code 'U071' which was determined by COVID-19 swab test results (hospital or community). We assessed the performance of CARSS (CARS_N, CARS_NB, CARM_N, CARM_NB) for predicting the risk of COVID-19 in terms of discrimination (c-statistic) and calibration (graphically).

RESULTS

The risk of in-hospital mortality following emergency medical admission was 8.4% (500/6444) and 9.6% (620/6444) had a diagnosis of COVID-19. For predicting COVID-19 admissions, the CARS_N model had the highest discrimination 0.73 (0.71 to 0.75) and calibration slope 0.81 (0.72 to 0.89) compared to other CARSS models: CARM_N (discrimination:0.68 (0.66 to 0.70) and calibration slope 0.47 (0.41 to 0.54)), CARM_NB (discrimination:0.68 (0.65 to 0.70) and calibration slope 0.37 (0.31 to 0.43)), and CARS_NB (discrimination:0.68 (0.66 to 0.70) and calibration slope 0.56 (0.47 to 0.64)).

CONCLUSIONS

The CARS_N model is reasonably accurate for predicting the risk of COVID-19. It may be clinically useful as an early warning system at the time of admission especially to triage large numbers of unplanned admissions because it requires no additional data collection and is readily automated.

Predicting outcomes of the acute phase of COVID-19. High sensitive prognostic model, based on the results of the international registry "analysis of chronic non-infectious diseases dynamics after COVID-19 infection in adult patients" (ACTIV)

The aim of this study is to investigate the course of the acute period of COVID-19 and devise a prognostic scale for patients hospitalized.

Materials and methods

The ACTIV registry encompassed both male and female patients aged 18 years and above, who were diagnosed with COVID-19 and subsequently hospitalized. Between June 2020 and March 2021, a total of 9364 patients were enrolled across 26 medical centers in seven countries. Data collected during the patients' hospital stay were subjected to multivariate analysis within the R computational environment. A predictive mathematical model, utilizing the "Random Forest" machine learning algorithm, was established to assess the risk of reaching the endpoint (defined as in-hospital death from any cause). This model was constructed using a training subsample (70% of patients), and subsequently tested using a control subsample (30% of patients).

Results

Out of the 9364 hospitalized COVID-19 patients, 545 (5.8%) died. Multivariate analysis resulted in the selection of eleven variables for the final model: minimum oxygen saturation, glomerular filtration rate, age, hemoglobin level, lymphocyte percentage, white blood cell count, platelet count, aspartate aminotransferase, glucose, heart rate, and respiratory rate. Receiver operating characteristic analysis yielded an area under the curve of 89.2%, a sensitivity of 86.2%, and a specificity of 76.0%. Utilizing the final model, a predictive equation and nomogram (termed the ACTIV scale) were devised for estimating in-hospital mortality amongst COVID-19 patients.

Conclusion

The ACTIV scale provides a valuable tool for practicing clinicians to predict the risk of in-hospital death in patients hospitalized with COVID-19.

A comprehensive benchmark for COVID-19 predictive modeling using electronic health records in intensive care

The COVID-19 pandemic highlighted the need for predictive deep-learning models in health care. However, practical prediction task design, fair comparison, and model selection for clinical applications remain a challenge. To address this, we introduce and evaluate two new prediction tasks-outcome-specific length-of-stay and early-mortality prediction for COVID-19 patients in intensive care-which better reflect clinical realities. We developed evaluation metrics, model adaptation designs, and open-source data preprocessing pipelines for these tasks while also evaluating 18 predictive models, including clinical scoring methods and traditional machine-learning, basic deep-learning, and advanced deep-learning models, tailored for electronic health record (EHR) data. Benchmarking results from two real-world COVID-19 EHR datasets are provided, and all results and trained models have been released on an online platform for use by clinicians and researchers. Our efforts contribute to the advancement of deep-learning and machine-learning research in pandemic predictive modeling.

A Systematic Review of Predictor Composition, Outcomes, Risk of Bias, and Validation of COVID-19 Prognostic Scores

BACKGROUND

Numerous prognostic scores have been published to support risk stratification for patients with coronavirus disease 2019 (COVID-19).

METHODS

We performed a systematic review to identify the scores for confirmed or clinically assumed COVID-19 cases. An in-depth assessment and risk of bias (ROB) analysis (Prediction model Risk Of Bias ASsessment Tool [PROBAST]) was conducted for scores fulfilling predefined criteria ([I] area under the curve [AUC)] ≥ 0.75; [II] a separate validation cohort present; [III] training data from a multicenter setting [≥2 centers]; [IV] point-scale scoring system).

RESULTS

Out of 1522 studies extracted from MEDLINE/Web of Science (20/02/2023), we identified 242 scores for COVID-19 outcome prognosis (mortality 109, severity 116, hospitalization 14, long-term sequelae 3). Most scores were developed using retrospective (75.2%) or single-center (57.1%) cohorts. Predictor analysis revealed the primary use of laboratory data and sociodemographic information in mortality and severity scores. Forty-nine scores were included in the in-depth analysis. The results indicated heterogeneous quality and predictor selection, with only five scores featuring low ROB. Among those, based on the number and heterogeneity of validation studies, only the 4C Mortality Score can be recommended for clinical application so far.

CONCLUSIONS

The application and translation of most existing COVID scores appear unreliable. Guided development and predictor selection would have improved the generalizability of the scores and may enhance pandemic preparedness in the future.