Deep learning of longitudinal chest X-ray and clinical variables predicts duration on ventilator and mortality in COVID-19 patients

A meta-analysis of the diagnostic test accuracy of artificial intelligence predicting emergency department dispositions

BACKGROUND

The rapid advancement of Artificial Intelligence (AI) has led to its widespread application across various domains, showing encouraging outcomes. Many studies have utilized AI to forecast emergency department (ED) disposition, aiming to forecast patient outcomes earlier and to allocate resources better; however, a dearth of comprehensive review literature exists to assess the objective performance standards of these predictive models using quantitative evaluations. This study aims to conduct a meta-analysis to assess the diagnostic accuracy of AI in predicting ED disposition, encompassing admission, critical care, and mortality.

METHODS

Multiple databases, including Scopus, Springer, ScienceDirect, PubMed, Wiley, Sage, and Google Scholar, were searched until December 31, 2023, to gather relevant literature. Risk of bias was assessed using the Prediction Model Risk of Bias Assessment Tool. Pooled estimates of sensitivity, specificity, and area under the receiver operating characteristic curve (AUROC) were calculated to evaluate AI's predictive performance. Sub-group analyses were performed to explore covariates affecting AI predictive model performance.

RESULTS

The study included 88 articles possessed with 117 AI models, among which 39, 45, and 33 models predicted admission, critical care, and mortality, respectively. The reported statistics for sensitivity, specificity, and AUROC represent pooled summary measures derived from the component studies included in this meta-analysis. AI's summary sensitivity, specificity, and AUROC for predicting admission were 0.81 (95% Confidence Interval [CI] 0.74-0.86), 0.87 (95% CI 0.81-0.91), and 0.87 (95% CI 0.84-0.93), respectively. For critical care, the values were 0.86 (95% CI 0.79-0.91), 0.89 (95% CI 0.83-0.93), and 0.93 (95% CI 0.89-0.95), respectively, and for mortality, they were 0.85 (95% CI 0.80-0.89), 0.94 (95% CI 0.90-0.96), and 0.93 (95% CI 0.89-0.96), respectively. Emergent sample characteristics and AI techniques showed evidence of significant covariates influencing the heterogeneity of AI predictive models for ED disposition.

CONCLUSIONS

The meta-analysis indicates promising performance of AI in predicting ED disposition, with certain potential for improvement, especially in sensitivity. Future research could explore advanced AI techniques such as ensemble learning and cross-validation with hyper-parameter tuning to enhance predictive model efficacy.

TRIAL REGISTRATION

This systematic review was not registered with PROSPERO or any other similar registry because the review was completed prior to the opportunity for registration, and PROSPERO currently does not accept registrations for reviews that are already completed. We are committed to transparency and have adhered to best practices in systematic review methodology throughout this study.

An ensemble model for predicting dispositions of emergency department patients

OBJECTIVE

The healthcare challenge driven by an aging population and rising demand is one of the most pressing issues leading to emergency department (ED) overcrowding. An emerging solution lies in machine learning's potential to predict ED dispositions, thus leading to promising substantial benefits. This study's objective is to create a predictive model for ED patient dispositions by employing ensemble learning. It harnesses diverse data types, including structured and unstructured information gathered during ED visits to address the evolving needs of localized healthcare systems.

METHODS

In this cross-sectional study, 80,073 ED patient records were amassed from a major southern Taiwan hospital in 2018-2019. An ensemble model incorporated structured (demographics, vital signs) and pre-processed unstructured data (chief complaints, preliminary diagnoses) using bag-of-words (BOW) and term frequency-inverse document frequency (TF-IDF). Two random forest base-learners for structured and unstructured data were employed and then complemented by a multi-layer perceptron meta-learner.

RESULTS

The ensemble model demonstrates strong predictive performance for ED dispositions, achieving an area under the receiver operating characteristic curve of 0.94. The models based on unstructured data encoded with BOW and TF-IDF yield similar performance results. Among the structured features, the top five most crucial factors are age, pulse rate, systolic blood pressure, temperature, and acuity level. In contrast, the top five most important unstructured features are pneumonia, fracture, failure, suspect, and sepsis.

CONCLUSIONS

Findings indicate that utilizing ensemble learning with a blend of structured and unstructured data proves to be a predictive method for determining ED dispositions.

SARS-CoV-2 infection increases long-term multiple sclerosis disease activity and all-cause mortality in an underserved inner-city population

BACKGROUND

Although certain subsets patients with multiple sclerosis (MS), an immune-mediated disorder, are at higher risk of worse acute COVID-19 outcomes compared to the general population, it is not clear whether SARS-CoV-2 infection impacts long-term outcomes compared with MS patients without COVID-19 infection.

OBJECTIVES

This study investigated MS disease activity and mortality 3.5 years post SARS-CoV-2 infection and compared with MS patients without COVID-19.

METHODS

This retrospective study evaluated 1,633 patients with MS in the Montefiore Health System in the Bronx from January 2016 to July 2023. This health system serves a large minority population and was an epicenter for the early pandemic and subsequent surges of infection. Positive SARS-CoV-2 infection was determined by a positive polymerase-chain-reaction test. Primary outcomes were all-cause mortality, and optic neuritis post SARS-CoV-2 infection. Secondary outcomes included change in disease-modifying therapy (DMT), treatment with high-dose methylprednisolone, cerebellar deficits, relapse, and all-cause hospitalization post-infection.

RESULTS

MS patients with COVID-19 had similar demographics but higher prevalence of pre-existing major comorbidities (hypertension, type-2 diabetes, chronic obstructive pulmonary disease, congestive heart failure, chronic kidney disease, and coronary artery disease), optic neuritis, and history of high dose steroid treatment for relapses compared to MS patients without COVID-19. MS patients with COVID-19 had greater risk of mortality (adjusted HR=4.34[1.67, 11.30], p < 0.005), greater risk of post infection optic neuritis (adjusted HR=2.97[1.58, 5.58], p < 0.005), higher incidence of methylprednisolone treatment for post infection acute relapse (12.65% vs. 2.54 %, p < 0.001), and more hospitalization (78.92% vs. 66.81 %, p < 0.01), compared to MS patients without COVID-19.

CONCLUSIONS

MS patients who survived COVID-19 infection experienced worse long-term outcomes, as measured by treatment for relapse, hospitalization and mortality. Identifying risk factors for worse long-term outcomes may draw clinical attention to the need for careful follow-up of at-risk individuals post-SARS-CoV-2 infection.

Designing Chinese hospital emergency departments to leverage artificial intelligence-a systematic literature review on the challenges and opportunities

Artificial intelligence (AI) has witnessed rapid advances in the healthcare domain in recent years, especially in the emergency field, where AI is likely to radically reshape medical service delivery. Although AI has substantial potential to enhance diagnostic accuracy and operational efficiency in hospitals, research on its applications in Emergency Department building design remains relatively scarce. Therefore, this study aims to investigate Emergency Department facility design by identifying the challenges and opportunities of using AI. Two systematic literature reviews are combined, one in AI and the other in sensors, to explore their potential application to support decision-making, resource optimisation and patient monitoring. These reviews have then informed a discussion on integrating AI sensors in contemporary Emergency Department designs for use in China to support the evidence base on resuscitation units, emergency operating rooms and Emergency Department Intensive Care Unit (ED-ICU) design. We hope to inform the strategic implementation of AI sensors and how they might transform Emergency Department design to support medical staff and enhance the patient experience.

Short- and Long-Term Chest-CT Findings after Recovery from COVID-19: A Systematic Review and Meta-Analysis

While ground-glass opacity, consolidation, and fibrosis in the lungs are some of the hallmarks of acute SAR-CoV-2 infection, it remains unclear whether these pulmonary radiological findings would resolve after acute symptoms have subsided. We conducted a systematic review and meta-analysis to evaluate chest computed tomography (CT) abnormalities stratified by COVID-19 disease severity and multiple timepoints post-infection. PubMed/MEDLINE was searched for relevant articles until 23 May 2023. Studies with COVID-19-recovered patients and follow-up chest CT at least 12 months post-infection were included. CT findings were evaluated at short-term (1-6 months) and long-term (12-24 months) follow-ups and by disease severity (severe and non-severe). A generalized linear mixed-effects model with random effects was used to estimate event rates for CT findings. A total of 2517 studies were identified, of which 43 met the inclusion (N = 8858 patients). Fibrotic-like changes had the highest event rate at short-term (0.44 [0.3-0.59]) and long-term (0.38 [0.23-0.56]) follow-ups. A meta-regression showed that over time the event rates decreased for any abnormality (β = -0.137, p = 0.002), ground-glass opacities (β = -0.169, p < 0.001), increased for honeycombing (β = 0.075, p = 0.03), and did not change for fibrotic-like changes, bronchiectasis, reticulation, and interlobular septal thickening (p > 0.05 for all). The severe subgroup had significantly higher rates of any abnormalities (p < 0.001), bronchiectasis (p = 0.02), fibrotic-like changes (p = 0.03), and reticulation (p < 0.001) at long-term follow-ups when compared to the non-severe subgroup. In conclusion, significant CT abnormalities remained up to 2 years post-COVID-19, especially in patients with severe disease. Long-lasting pulmonary abnormalities post-SARS-CoV-2 infection signal a future public health concern, necessitating extended monitoring, rehabilitation, survivor support, vaccination, and ongoing research for targeted therapies.

A Hybrid Decision Tree and Deep Learning Approach Combining Medical Imaging and Electronic Medical Records to Predict Intubation Among Hospitalized Patients With COVID-19: Algorithm Development and Validation

BACKGROUND

Early prediction of the need for invasive mechanical ventilation (IMV) in patients hospitalized with COVID-19 symptoms can help in the allocation of resources appropriately and improve patient outcomes by appropriately monitoring and treating patients at the greatest risk of respiratory failure. To help with the complexity of deciding whether a patient needs IMV, machine learning algorithms may help bring more prognostic value in a timely and systematic manner. Chest radiographs (CXRs) and electronic medical records (EMRs), typically obtained early in patients admitted with COVID-19, are the keys to deciding whether they need IMV.

OBJECTIVE

We aimed to evaluate the use of a machine learning model to predict the need for intubation within 24 hours by using a combination of CXR and EMR data in an end-to-end automated pipeline. We included historical data from 2481 hospitalizations at The Mount Sinai Hospital in New York City.

METHODS

CXRs were first resized, rescaled, and normalized. Then lungs were segmented from the CXRs by using a U-Net algorithm. After splitting them into a training and a test set, the training set images were augmented. The augmented images were used to train an image classifier to predict the probability of intubation with a prediction window of 24 hours by retraining a pretrained DenseNet model by using transfer learning, 10-fold cross-validation, and grid search. Then, in the final fusion model, we trained a random forest algorithm via 10-fold cross-validation by combining the probability score from the image classifier with 41 longitudinal variables in the EMR. Variables in the EMR included clinical and laboratory data routinely collected in the inpatient setting. The final fusion model gave a prediction likelihood for the need of intubation within 24 hours as well.

RESULTS

At a prediction probability threshold of 0.5, the fusion model provided 78.9% (95% CI 59%-96%) sensitivity, 83% (95% CI 76%-89%) specificity, 0.509 (95% CI 0.34-0.67) F1-score, 0.874 (95% CI 0.80-0.94) area under the receiver operating characteristic curve (AUROC), and 0.497 (95% CI 0.32-0.65) area under the precision recall curve (AUPRC) on the holdout set. Compared to the image classifier alone, which had an AUROC of 0.577 (95% CI 0.44-0.73) and an AUPRC of 0.206 (95% CI 0.08-0.38), the fusion model showed significant improvement (P<.001). The most important predictor variables were respiratory rate, C-reactive protein, oxygen saturation, and lactate dehydrogenase. The imaging probability score ranked 15th in overall feature importance.

CONCLUSIONS

We show that, when linked with EMR data, an automated deep learning image classifier improved performance in identifying hospitalized patients with severe COVID-19 at risk for intubation. With additional prospective and external validation, such a model may assist risk assessment and optimize clinical decision-making in choosing the best care plan during the critical stages of COVID-19.

Longitudinal Chest X-ray Scores and their Relations with Clinical Variables and Outcomes in COVID-19 Patients

Background: This study evaluated the temporal characteristics of lung chest X-ray (CXR) scores in COVID-19 patients during hospitalization and how they relate to other clinical variables and outcomes (alive or dead). Methods: This is a retrospective study of COVID-19 patients. CXR scores of disease severity were analyzed for: (i) survivors (N = 224) versus non-survivors (N = 28) in the general floor group, and (ii) survivors (N = 92) versus non-survivors (N = 56) in the invasive mechanical ventilation (IMV) group. Unpaired t-tests were used to compare survivors and non-survivors and between time points. Comparison across multiple time points used repeated measures ANOVA and corrected for multiple comparisons. Results: For general-floor patients, non-survivor CXR scores were significantly worse at admission compared to those of survivors (p < 0.05), and non-survivor CXR scores deteriorated at outcome (p < 0.05) whereas survivor CXR scores did not (p > 0.05). For IMV patients, survivor and non-survivor CXR scores were similar at intubation (p > 0.05), and both improved at outcome (p < 0.05), with survivor scores showing greater improvement (p < 0.05). Hospitalization and IMV duration were not different between groups (p > 0.05). CXR scores were significantly correlated with lactate dehydrogenase, respiratory rate, D-dimer, C-reactive protein, procalcitonin, ferritin, SpO2, and lymphocyte count (p < 0.05). Conclusions: Longitudinal CXR scores have the potential to provide prognosis, guide treatment, and monitor disease progression.