Materiały źródłowe

• #1 HELMET: A Hybrid Machine Learning Framework for Real-Time Prediction of Edema Trajectory in Large Middle Cerebral Artery Stroke | medRxiv

  https://www.medrxiv.org/content/10.1101/2024.11.13.24317229v1.full-text

  Malignant cerebral edema occurs when brain swelling displaces and compresses vital midline structures within the first week of a large middle cerebral artery stroke. […] A major driver of poor stroke outcomes is malignant cerebral edema, which can result in a 40-80% risk of neurological deterioration and death. […] Early recognition of evolving cerebral edema is imperative as treatments, such as surgical decompression, can reduce the risk of mortality from 80% to 20% in eligible patients. […] Current strategies for monitoring mass effect and other secondary injuries primarily rely on physical examination and confirmatory imaging. […] Data-driven models for edema risk assessment could lead to more personalized and accurate screening policies. […] Improved methods to estimate cerebral edema trajectory are needed to leverage the rich sources of structured and unstructured data currently available in electronic health records.

• #2 HELMET: A Hybrid Machine Learning Framework for Real-Time Prediction of Edema Trajectory in Large Middle Cerebral Artery Stroke | medRxiv

  https://www.medrxiv.org/content/10.1101/2024.11.13.24317229v1.full-text

  Our dynamic time series models leverage non-linear machine learning algorithms to predict the risk of edema and the trajectory of MLS using a comprehensive multi-modal dataset for two distinct future time windows. […] To the best of our knowledge, HELMET-8 and HELMET-24 constitute the first dynamic risk models for predicting the trajectory of cerebral edema on an hourly basis, leveraging data from two healthcare systems. […] Our findings underscore that multi-modal hybrid approaches combining both expert-derived features and raw data appear to significantly enhance outcome prediction accuracy in clinical settings and may inform which features should be routinely included in radiology reports. […] An important implication of our work is that shifting risk prediction from a single baseline assessment to a continuous, longitudinal approach could improve real-time patient triage, optimize imaging resource use, and explore whether early-warning alerts based on these predictions can enhance patient outcomes and care quality.

• #3 Predicting the efficacy of bevacizumab on peritumoral edema based on imaging features and machine learning | Scientific Reports

  https://www.nature.com/articles/s41598-025-00758-0

  This study proposes a novel approach to predict the efficacy of bevacizumab (BEV) in treating peritumoral edema in metastatic brain tumor patients by integrating advanced machine learning (ML) techniques with comprehensive imaging and clinical data. […] Our results highlight the potential of ML-driven predictive models in optimizing BEV treatment selection, reducing unnecessary treatment risks, and improving clinical decision-making in neuro-oncology. […] The analysis revealed that five variables—tumor origin, surgical history, maximum tumor diameter, edema volume, and edema index—were statistically significantly associated with the treatment outcome, and thus were considered potential predictors of efficacy. […] The results indicated that edema volume and edema index emerged as independent risk factors for BEV treatment efficacy. Notably, the multivariate analysis highlighted edema volume as the most critical factor influencing treatment outcomes.

• #4 Follow-up ASPECTS improves prediction of potentially lethal malignant edema in patients with large middle cerebral artery stroke

  https://pmc.ncbi.nlm.nih.gov/articles/PMC11214636/

  Recent studies have shown that follow-up head CT is a strong predictor of functional outcomes in patients with middle cerebral artery stroke and mechanical thrombectomy. […] We sought to determine whether total and/or regional follow-up Alberta Stroke Program Early CT Score (ASPECTSfu) are associated with important clinical outcomes during hospitalization and improve the performance of clinical prediction models of potentially lethal malignant edema (PLME). […] Lower ASPECTSfu and certain locations on regional ASPECTSfu, including deep and anterior areas, were significantly associated with PLME. Including ASPECTSfu information improved discrimination of established edema prediction models and could be used immediately to help facilitate clinical management decisions and prognostication.

• #5 Follow-up ASPECTS improves prediction of potentially lethal malignant edema in patients with large middle cerebral artery stroke

  https://pmc.ncbi.nlm.nih.gov/articles/PMC11214636/

  We hypothesized that total ASPECTSfu is a strong predictor and improves existing models of PLME, including the Enhanced Detection of Edema in Malignant Anterior Circulation Stroke (EDEMA) and modified EDEMA models. […] We found that in our multicenter study of 560 patients, lower total ASPECTSfu was significantly associated with PLME, even with adjusting for ASPECTSbl. ASPECTSfu provides an easy to measure quantitative metric of permanent stroke injury and likelihood of edema more than baseline characteristics alone. […] Follow-up ASPECTS was independently associated with and improved prediction of PLME in patients with large middle cerebral artery stroke. Calculation of ASPECTSfu is a feasible and clinically useful addition to existing tools for prognostication and can help clinicians inform management decisions.

• #6 Relative edema volume is a predictor of outcome in patients with hyperacute spontaneous intracerebral hemorrhage – PubMed

  https://pubmed.ncbi.nlm.nih.gov/12411654/

  Relative edema volume is a predictor of outcome in patients with hyperacute spontaneous intracerebral hemorrhage. […] The purpose of this study was to determine whether absolute or relative edema volume (edema volume divided by hematoma volume) predicts mortality or functional outcome in patients with hyperacute spontaneous ICH. […] We hypothesized that increasing baseline relative edema volume is associated with greater probability of poor functional outcome. […] By multivariable logistic regression analysis, baseline relative edema was the strongest independent predictor of functional outcome and was associated with lesser odds of poor 3-month functional outcome (odds ratio, 0.09 per 1.0-unit [100%] increase; 95% CI, 0.01 to 0.64; P=0.016) and 12-week Barthel Index score 85 (odds ratio, 0.12; 95% CI, 0.02 to 0.91; P=0.039) but did not predict mortality. […] Relative edema is strongly predictive of functional outcome in patients with hyperacute supratentorial spontaneous ICH without intraventricular extension.

• #7 Perihematomal edema as predictor of outcome in spontaneous intracerebral hemorrhage – Journal of Neurosciences in Rural Practice

  https://ruralneuropractice.com/perihematomal-edema-as-predictor-of-outcome-in-spontaneous-intracerebral-hemorrhage/

  In SICH, a larger hematoma volume is a predictor of poor outcome and a relative edema is associated with a better functional status. […] The hematoma volume was significantly higher in the poor outcome group (P 0.0001) and the relative edema was associated with a favorable outcome at 12 weeks (P 0.0001). […] The absolute PHE has not been found to be an independent predictor for 30 day mortality as it has been correlated with hematoma expansion which in turn independently predicts 30 day mortality. […] The volume of hematoma correlated with a poor outcome [odds ratio (OR) =1.25, 95% confidence interval (CI) =1.02 – 1.53, P= 0.031] and relative edema correlated with a favorable outcome at 12 weeks (OR = 0.00, 95% CI = 0.00 – 0.08,P= 0.019). […] The relative edema was significantly different in the two groups in our study. A larger ratio of edema to hematoma volume was noticed in the group with favorable outcome (MRS 3) at 12 weeks. […] The relative edema did not seem to affect the survival function in our patient as assessed by Kaplan-Meier survival curve.

• #8 Prognostication after cardiac arrest | Critical Care | Full Text

  https://ccforum.biomedcentral.com/articles/10.1186/s13054-018-2060-7

  Hypoxicischaemic brain injury (HIBI) is the main cause of death in patients who are comatose after resuscitation from cardiac arrest. A poor neurological outcome defined as death from neurological cause, persistent vegetative state, or severe neurological disability can be predicted in these patients by assessing the severity of HIBI. […] Current guidelines recommend performing prognostication no earlier than 72 h after return of spontaneous circulation in all comatose patients with an absent or extensor motor response to pain, after having excluded confounders such as residual sedation that may interfere with clinical examination. A multimodal approach combining multiple prognostication tests is recommended so that the risk of a falsely pessimistic prediction is minimised. […] To avoid premature WLST in patients with a chance of neurological recovery, the risk of a falsely pessimistic prediction should be kept to a minimum.

• #9 Prognostication after cardiac arrest | Critical Care | Full Text

  https://ccforum.biomedcentral.com/articles/10.1186/s13054-018-2060-7

  The latest version of the Utstein guidelines on outcome reporting after OHCA suggested that when dichotomising neurological outcome the CPC 35 threshold (or mRS 46) should be used for defining poor outcome. […] The timing of neurological outcome assessment also affects its measured values, since in initially comatose cardiac arrest survivors neurological status can improve for up to 6 months after the event. […] The most robust predictors (FPR5% for prediction of poor outcome with narrow confidence interval documented in 5 studies from at least three different groups of investigators) should be evaluated first. These include bilaterally absent pupillary reflexes at 72 h after ROSC and/or a bilaterally absent N2O SSEP wave after rewarming. […] A multimodal approach combining multiple prognostication tests is recommended by current guidelines so that the risk of a falsely pessimistic prediction is minimised.

• #10 Radiomics for Predicting the Development of Brain Edema from Normal-Appearing Early Brain-CT After Cardiac Arrest and Return of Spontaneous Circulation

  https://www.mdpi.com/2075-4418/15/2/119

  Radiomics, particularly focused on the basalganglia area in normal-appearing brain CTs after CA and ROSC, may enhance predictive insights for HIBI and the development of brain edema. […] The model using manual segmentation achieved an average AUC of 0.76, sensitivity of 0.59, and specificity of 0.78. The automated segmentation model showed an average AUC of 0.66, sensitivity of 0.49, and specificity of 0.68. […] Early post-CA CT scans are commonly used to detect acute brain pathologies such as stroke, bleeding, or early signs of brain edema. They also hold promising potential for predicting clinical progression and neurological outcomes. […] The occurrence of brain edema in 47 out of 83 patients during follow-up scans highlights the prevalence of this post-arrest complication, indicating its substantial occurrence in this cohort.

• #11 Radiomics for Predicting the Development of Brain Edema from Normal-Appearing Early Brain-CT After Cardiac Arrest and Return of Spontaneous Circulation

  https://www.mdpi.com/2075-4418/15/2/119

  The model performance analysis indicated varying degrees of predictive capability between the two segmentation approaches. The model utilizing manual basalganglia segmentation exhibited superior performance, with an average AUC of 0.76, a sensitivity of 0.59, and a specificity of 0.78. On the other hand, the model relying on automated whole-brain segmentation showed comparatively lower performance metrics, with an average AUC of 0.66, sensitivity of 0.49, and specificity of 0.68. […] Radiomics features from normal-appearing early brain CTs after CA and ROSC, together with a simple logistic regression model, might be a viable, explainable, and simple method to add additional information for predicting the development of brain edema.

• #12 Predictive model for estimating the risk of high-altitude pulmonary edema: a single-centre retrospective outcome-reporting study | BMJ Open

  https://bmjopen.bmj.com/content/13/11/e074161

  Objective To develop the first prediction model based on the common clinical symptoms of high-altitude pulmonary edema (HAPE), enabling early identification and an easy-to-execute self-risk prediction tool. […] The recommended prediction model (nomogram) could estimate the risk of HAPE with good precision, high discrimination and possible clinical applications for patients with HAPE. More importantly, it is an easy-to-execute scoring tool for individuals without medical professionals support. […] This study also reflected the satisfactory performance of the model in its modest predictive power, with AUC0.85 in both models. In addition, calibration analyses performed in both cohorts showed that the predicted occurrence of HAPE was highly consistent with the diagnosed HAPE proportion. In both models, clinical decision curve analysis showed a net clinical benefit rate of90% for patients with HAPE. This provides a practical scoring system with accurate assessment even for settings where medical instrumentation is not immediately available.

• #13 Predictive model for estimating the risk of high-altitude pulmonary edema: a single-centre retrospective outcome-reporting study | BMJ Open

  https://bmjopen.bmj.com/content/13/11/e074161

  The main advantage of this model is that it allows for simple and accurate detection of HAPE clinical symptoms at an early onset, which can be easily used by individuals without medical expert advice. Such a practical scoring tool may help high-altitude visitors estimate the probability of developing HAPE and take relevant precautions that could be potentially lifesaving. Additionally, applying this model for high-altitude travel should aid in the early detection of individuals at high risk of HAPE and contribute to efforts to provide clinical decision-making support to decrease potentially adverse health outcomes.

• #14 A deep learning model enables accurate prediction and quantification of pulmonary edema from chest X-rays | Critical Care | Full Text

  https://ccforum.biomedcentral.com/articles/10.1186/s13054-023-04426-5

  A quantitative assessment of pulmonary edema is important because the clinical severity can range from mild impairment to life threatening. […] Deep learning can quantify pulmonary edema as measured by EVLWI with high accuracy. […] In this study, we sequentially developed a deep learning model that accurately quantifies pulmonary edema from chest X-ray images. […] Despite these limitations our study demonstrates, that deep learning is a useful tool for the quantification of pulmonary edema with a meaningful resolution with high accuracy.

• #15 Cardiac Magnetic Resonance—Detected Acute Myocardial Edema as Predictor of Favourable Prognosis: A Comprehensive Review

  https://www.mdpi.com/2308-3425/10/8/319

  Acute myocardial edema (AME) is increased water content in the myocardium and represents the first and transient pathophysiological response to an acute myocardial injury. […] In the clinical setting, early demonstration of AME has a recognized diagnostic value for acute coronary syndromes and acute myocarditis, although its prognostic value is not well established. […] The presence of AME in clinically suspected or biopsy-proven myocarditis was further confirmed despite its extension. […] The favourable prognostic role of AME in clinically suspected or biopsy-proven myocarditis was further confirmed despite its extension. […] The detection of AME is useful also for the clinical management of TTS patients. […] The presence of AME without LGE has favourable prognostic implications.

• #16 Cardiac Magnetic Resonance—Detected Acute Myocardial Edema as Predictor of Favourable Prognosis: A Comprehensive Review

  https://www.mdpi.com/2308-3425/10/8/319

  The detection of AME is useful also for the clinical management of TTS patients. […] Myocardial edema demonstrated to hold both a diagnostic and a prognostic role in patients resuscitated from OHCA, suggesting a potentially great utility of CMR in their clinical management. […] The presence of edema was proven in 87.5% of the 16 myocardial biopsies performed. […] Myocardial edema was recognized in drug induced myocarditis as well, associated with immune checkpoint inhibitor (ICI) drugs and in the context of drug rash with eosinophilia and systemic symptoms (DRESS) syndrome. […] Myocardial edema could be an expression of subclinical inflammation in patient with autoimmune diseases. […] The presence of edema did not emerge as an independent predictor of outcome. […] The progress in cardiovascular magnetic resonance techniques, particularly with T2 weighted sequences and T2 mapping, allows its detection in multiple clinical scenarios. […] Generally, its presence owns a favourable prognostic meaning in widespread common cardiac conditions, such as acute coronary syndromes, acute myocarditis, TTS and OHCA.

• #17 Predicting the efficacy of bevacizumab on peritumoral edema based on imaging features and machine learning | Scientific Reports

  https://www.nature.com/articles/s41598-025-00758-0

  The predictive model developed in this study offers significant clinical utility by stratifying patients with PTE into probable responders and non-responders to BEV treatment based on integrated clinical and imaging characteristics. This stratification enables clinicians to optimize therapeutic strategies: prioritizing BEV for likely responders to accelerate edema remission, while considering alternative approaches for predicted non-responders to avoid unnecessary exposure to BEV’s risks and reduce financial burdens.

Zobacz więcej