Development and validation of a prediction model to estimate risk of acute pulmonary embolism in deep vein thrombosis patients

Using machine learning to improve the diagnostic accuracy of the modified Duke/ESC 2015 criteria in patients with suspected prosthetic valve endocarditis - a proof of concept study

INTRODUCTION

Prosthetic valve endocarditis (PVE) is a serious complication of prosthetic valve implantation, with an estimated yearly incidence of at least 0.4-1.0%. The Duke criteria and subsequent modifications have been developed as a diagnostic framework for infective endocarditis (IE) in clinical studies. However, their sensitivity and specificity are limited, especially for PVE. Furthermore, their most recent versions (ESC2015 and ESC2023) include advanced imaging modalities, e.g., cardiac CTA and [18F]FDG PET/CT as major criteria. However, despite these significant changes, the weighing system using major and minor criteria has remained unchanged. This may have introduced bias to the diagnostic set of criteria. Here, we aimed to evaluate and improve the predictive value of the modified Duke/ESC 2015 (MDE2015) criteria by using machine learning algorithms.

METHODS

In this proof-of-concept study, we used data of a well-defined retrospective multicentre cohort of 160 patients evaluated for suspected PVE. Four machine learning algorithms were compared to the prediction of the diagnosis according to the MDE2015 criteria: Lasso logistic regression, decision tree with gradient boosting (XGBoost), decision tree without gradient boosting, and a model combining predictions of these (ensemble learning). All models used the same features that also constitute the MDE2015 criteria. The final diagnosis of PVE, based on endocarditis team consensus using all available clinical information, including surgical findings whenever performed, and with at least 1 year follow up, was used as the composite gold standard.

RESULTS

The diagnostic performance of the MDE2015 criteria varied depending on how the category of 'possible' PVE cases were handled. Considering these cases as positive for PVE, sensitivity and specificity were 0.96 and 0.60, respectively. Whereas treating these cases as negative, sensitivity and specificity were 0.74 and 0.98, respectively. Combining the approaches of considering possible endocarditis as positive and as negative for ROC-analysis resulted in an excellent AUC of 0.917. For the machine learning models, the sensitivity and specificity were as follows: logistic regression, 0.92 and 0.85; XGBoost, 0.90 and 0.85; decision trees, 0.88 and 0.86; and ensemble learning, 0.91 and 0.85, respectively. The resulting AUCs were, in the same order: 0.938, 0.937, 0.930, and 0.941, respectively.

DISCUSSION

In this proof-of-concept study, machine learning algorithms achieved improved diagnostic performance compared to the major/minor weighing system as used in the MDE2015 criteria. Moreover, these models provide quantifiable certainty levels of the diagnosis, potentially enhancing interpretability for clinicians. Additionally, they allow for easy incorporation of new and/or refined criteria, such as the individual weight of advanced imaging modalities such as CTA or [18F]FDG PET/CT. These promising preliminary findings warrant further studies for validation, ideally in a prospective cohort encompassing the full spectrum of patients with suspected IE.

Interventional therapies for pulmonary embolism

Pulmonary embolism (PE) is the leading cause of in-hospital death and the third most frequent cause of cardiovascular death. The clinical presentation of PE is variable, and choosing the appropriate treatment for individual patients can be challenging. Traditionally, treatment of PE has involved a choice of anticoagulation, thrombolysis or surgery; however, a range of percutaneous interventional technologies have been developed that are under investigation in patients with intermediate-high-risk or high-risk PE. These interventional technologies include catheter-directed thrombolysis (with or without ultrasound assistance), aspiration thrombectomy and combinations of the aforementioned principles. These interventional treatment options might lead to a more rapid improvement in right ventricular function and pulmonary and/or systemic haemodynamics in particular patients. However, evidence from randomized controlled trials on the safety and efficacy of these interventions compared with conservative therapies is lacking. In this Review, we discuss the underlying pathophysiology of PE, provide assistance with decision-making on patient selection and critically appraise the available clinical evidence on interventional, catheter-based approaches for PE treatment. Finally, we discuss future perspectives and unmet needs.

Risk factors for patients with acute hospital-acquired symptomatic pulmonary thromboembolism

This study aimed to identify independent risk factors for acute hospital-acquired symptomatic pulmonary embolism (HA-SPE) by comparing the clinical data of HA-SPE and acute nonhospital-acquired symptomatic pulmonary embolism (NHA-SPE). A total of 292 patients were included in the analysis and divided into two groups: 191 patients had acute NHA-SPE, and 101 patients had acute HA-SPE. The average age of these 292 patients was 63.2 years, and the sample included 145 males. Multivariate analysis showed that malignant tumour (OR, 3.811; 95% CI [1.914-7.586], P = 0.000), recent surgery (OR, 7.310; 95% CI 3.392-15.755], P = 0.000), previous VTE (OR, 5.973; 95% CI 2.194 16.262], P = 0. 000), and the length of stay (LOS) (OR, 1.075; 95% CI [1.040-1.111], P = 0.000) were independent risk factors for acute HA-AEP. The c-statistic for this model was 0.758 (95% CI [0.698-0.800], P < 0.0001). The K-M curve showed that the hazard ratio (HR) of the HA group to the NHA group in all-cause mortality was 3.807 (95% CI [1.987, 7.295], P = 0.0061). Strengthening the prevention and control of patients with these risk factors may reduce the incidence of acute HA-SPE.

Development and internal validation of a clinical prediction model for the diagnosis of immune thrombocytopenia

BACKGROUND

Immune thrombocytopenia (ITP) is a diagnosis of exclusion that can resemble other thrombocytopenic disorders.

OBJECTIVES

To develop a clinical prediction model (CPM) for the diagnosis of ITP to aid hematogists in investigating patients presenting with undifferentiated thrombocytopenia.

METHODS

We designed a CPM for ITP diagnosis at the time of the initial hematology consultation using penalized logistic regression based on data from patients with thrombocytopenia enrolled in the McMaster ITP registry (n = 523) called the Predict-ITP Tool. The case definition for ITP was a platelet count less than 100 × 109 /L and a platelet count response after high-dose corticosteroids or intravenous immune globulin, defined as the achievement of a platelet count above 50 × 109 /L and at least a doubling of baseline. Internal validation was done using bootstrap resampling. Model discrimination was assessed by the c-statistic, and calibration was assessed by the calibration slope, calibration-in-the-large, and calibration plot.

RESULTS

The final model included the following variables: (1) platelet count variability (based on three or more platelet count values), (2) lowest platelet count value, (3) maximum mean platelet volume, and (4) history of major bleeding (defined by the ITP bleeding scale). The optimism-corrected c-statistic was 0.83, the calibration slope was 0.88, and calibration-in-the-large for all performance measures was <0.001 with standard error <0.001, indicating good discrimination and excellent calibration.

CONCLUSIONS

The Predict-ITP Tool can estimate the likelihood of ITP for a given patient with thrombocytopenia at the time of the initial hematology consultation. The tool had high predictive accuracy for the diagnosis of ITP.

Clinical value of deep vein thrombosis density on pre-contrast and post-contrast lower-extremity CT for prediction of pulmonary thromboembolism

Background

There was a lack of studies assessing the relationship between deep vein thrombosis (DVT) Hounsfield unit (HU) density and pulmonary thromboembolism (PTE).

Purpose

To evaluate the clinical value of DVT density measured on pre- and post-contrast lower-extremity computed tomography (CT) for the prediction of PTE.

Material and Methods

From 2017 to 2021, patients who underwent pulmonary CT angiography within one week after diagnosis of DVT on lower-extremity CT were included in this retrospective study. Then, the patients without PTE were included in “DVT group” and those with both DVT and PTE were included in the “DVT-PTE group.” The DVT HU density was measured by drawing free-hand region of interests (ROIs) within the thrombus at the most proximal filling defect level. A receiver operating characteristic (ROC) analysis was used to evaluate the predictive value of DVT density for the risk of PTE.

Results

This study included a total of 94 patients (DVT group: n=56; DVT-PTE group: m=38). DVT density was significantly higher in the DVT-PTE group than the DVT group in both pre-contrast (53.5 ± 6.2 HU vs. 44.1 ± 7.9 HU; P < 0.001) and post-contrast CT (67.0 ± 8.6 HU vs. 57.1 ± 10.6 HU; P < 0.001). ROC analysis revealed that the area under curve, sensitivity, and specificity for predicting the risk of PTE were 0.739, 71.1%, and 64.2%, respectively, at a DVT density cutoff of 48.2 HU on pre-contrast CT and were 0.779, 73.7%, and 69.6% at a DVT density cutoff of 61.8 HU on post-contrast CT.

Conclusion

The DVT density on both pre- and post-contrast CT could be a predictive factor of PTE.