[ESC guidelines 2019 on diagnostics and management of acute pulmonary embolism]

Establishment of machine learning-based tool for early detection of pulmonary embolism

BACKGROUND AND OBJECTIVES

Pulmonary embolism (PE) is a complex disease with high mortality and morbidity rate, leading to increasing society burden. However, current diagnosis is solely based on symptoms and laboratory data despite its complex pathology, which easily leads to misdiagnosis and missed diagnosis by inexperienced doctors. Especially, CT pulmonary angiography, the gold standard method, is not widely available. In this study, we aim to establish a rapid and accurate screening model for pulmonary embolism using machine learning technology. Importantly, data required for disease prediction are easily accessed, including routine laboratory data and medical record information of patients.

METHODS

We extracted features from patients' routine laboratory results and medical records, including blood routine, biochemical group, blood coagulation routine and other test results, as well as symptoms and medical history information. Samples with a feature loss rate greater than 0.8 were deleted from the original database. Data from 4723 cases were retained, 231 of which were positive for pulmonary embolism. 50 features were retained through the positive and negative statistical hypothesis testing which was used to build the predictive model. In order to avoid identification as majority-class samples caused by the imbalance of sample proportion, we used the method of Synthetic Minority Oversampling Technique (SMOTE) to increase the amount of information on minority samples. Five typical machine learning algorithms were used to model the screening of pulmonary embolism, including Support Vector Machines, Logistic Regression, Random Forest, XGBoost, and Back Propagation Neural Networks. To evaluate model performance, sensitivity, specificity and AUC curve were analyzed as the main evaluation indicators. Furthermore, a baseline model was established using the characteristics of the pulmonary embolism guidelines as a comparison model.

RESULTS

We found that XGBoost showed better performance compared to other models, with the highest sensitivity and specificity (0.99 and 0.99, respectively). Moreover, it showed significant improvement in performance compared to the baseline model (sensitivity and specificity were 0.76 and 0.76 respectively). More important, our model showed low missed diagnosis rate (0.46) and high AUC value (0.992). Finally, the calculation time of our model is only about 0.05 s to obtain the possibility of pulmonary embolism.

CONCLUSIONS

In this study, five machine learning classification models were established to assess the likelihood of patients suffering from pulmonary embolism, and the XGBoost model most significantly improved the precision, sensitivity, and AUC for pulmonary embolism screening. Collectively, we have established an AI-based model to accurately predict pulmonary embolism at early stage.

Streamlining Patient Management of Suspected COVID-19 Patients in the Emergency Department: Incorporation of Pulmonary CT Angiography into the Triaging Algorithm

Purpose: To evaluate the use of pulmonary computed tomography (CT) angiography during initial admission at an emergency department (ED), to identify COVID-19 patients with accompanying pulmonary embolism (PE) and its impact on clinical management. Methods: We performed a retrospective analysis of COVID-19 patients that underwent pulmonary CT angiography at the ED. CT scans were evaluated for the presence and extent of PE and for imaging changes suspicious of COVID-19. Patients were subdivided into two groups: (1) Group A consisted of patients with proven COVID-19 based on real-time polymerase chain reaction (RT-PCR), and (2) Group B of patients suspected for COVID-19, comprising patients positive on RT-PCR and/or COVID-19-suspicious CT findings. To assess the differences between patients with and without pulmonary embolism, Fisher’s exact test was used. Results: A total of 308 patients were admitted to the ED for diagnostic work-up of dyspnea and suspected COVID-19, and 95 patients underwent pulmonary CT angiography. PE was detected in 13.6% (3/22) of patients in Group A and 20.7% (6/29) in Group B. No significant differences were observed between patients with and without PE concerning hospitalization (Group B: 100% (6/6) vs. 91.3% (21/23)), the necessity of oxygen therapy (Group B: 66% (4/6) vs. 43.5% (10/23)), and death (Group B: 33% (2/6) vs. 4.3% (1/23) p > 0.05, respectively). Conclusions: In 20.7% of COVID-19 patients, PE was detected upon admission to the ED. Although the incorporation of early pulmonary CT angiography in patients suspicious of COVID-19 may be beneficial to identify concomitant PE, further studies are necessary to corroborate these findings.

Direct Oral Anticoagulants Versus Warfarin in the Treatment of Cerebral Venous Thrombosis (ACTION-CVT): A Multicenter International Study

BACKGROUND

A small randomized controlled trial suggested that dabigatran may be as effective as warfarin in the treatment of cerebral venous thrombosis (CVT). We aimed to compare direct oral anticoagulants (DOACs) to warfarin in a real-world CVT cohort.

METHODS

This multicenter international retrospective study (United States, Europe, New Zealand) included consecutive patients with CVT treated with oral anticoagulation from January 2015 to December 2020. We abstracted demographics and CVT risk factors, hypercoagulable labs, baseline imaging data, and clinical and radiological outcomes from medical records. We used adjusted inverse probability of treatment weighted Cox-regression models to compare recurrent cerebral or systemic venous thrombosis, death, and major hemorrhage in patients treated with warfarin versus DOACs. We performed adjusted inverse probability of treatment weighted logistic regression to compare recanalization rates on follow-up imaging across the 2 treatments groups.

RESULTS

Among 1025 CVT patients across 27 centers, 845 patients met our inclusion criteria. Mean age was 44.8 years, 64.7% were women; 33.0% received DOAC only, 51.8% received warfarin only, and 15.1% received both treatments at different times. During a median follow-up of 345 (interquartile range, 140-720) days, there were 5.68 recurrent venous thrombosis, 3.77 major hemorrhages, and 1.84 deaths per 100 patient-years. Among 525 patients who met recanalization analysis inclusion criteria, 36.6% had complete, 48.2% had partial, and 15.2% had no recanalization. When compared with warfarin, DOAC treatment was associated with similar risk of recurrent venous thrombosis (aHR, 0.94 [95% CI, 0.51-1.73]; P=0.84), death (aHR, 0.78 [95% CI, 0.22-2.76]; P=0.70), and rate of partial/complete recanalization (aOR, 0.92 [95% CI, 0.48-1.73]; P=0.79), but a lower risk of major hemorrhage (aHR, 0.35 [95% CI, 0.15-0.82]; P=0.02).

CONCLUSIONS

In patients with CVT, treatment with DOACs was associated with similar clinical and radiographic outcomes and favorable safety profile when compared with warfarin treatment. Our findings need confirmation by large prospective or randomized studies.

Hospitalization for venous thromboembolic disease management: a 10 years Registry in Buenos Aires, Argentina

Venous thromboembolism (VTE) is the most frequent cause of preventable mortality in hospitalized patients and the third leading cause of mortality due to vascular diseases. We aim to describe patients with VTE who required hospitalization, their clinical characteristics, management and evolution after discharge. Prospective cohort which included patients with acute and symptomatic VTE who required hospitalization in a single tertiary center for their management in a 10 years´ period (between 2006 and 2016). Patients were included at the time of VTE diagnosis; we collected baseline characteristics, risk stratification and initial therapeutic approach. They were periodically, prospectively and systematically followed up to evaluate recurrence, bleeding and/or death. Time-to-event analysis was performed for these outcomes related to evolucion after discharge. 3457 confirmed VTE episodes were included for the analysis (1985 DVT, 930 PE and 542 DVT + DVT), corresponding to 3157 patients. Most were elderly, the median age was 68 years (SD 16) with a predominance of women 57% (1955). Most frequent predisposing factors were immobility (45%), active cancer (35%), recent surgery (33%), obesity (26%), recent hospitalization (22%) and previous VTE (21%). One-fourth of pulmonary embolism had sPESI 0 points. Mortality, recurrence and bleeding rates within 30 days were 14%, 10% and 5% respectively. High proportion of VTE require hospitalization for their management, even PE with low sPESI. Mortality and complication remain elevated. We encourage the development and maintenance of more VTE registries in Latin America, to fully understand local characteristics of this disease.

Successful Surgical Embolectomy for a Rapidly Deteriorating Patient with Pulmonary Embolism and Hemorrhagic Stroke

Pulmonary embolism (PE) is a life-threatening condition. High-risk PE is defined as pulmonary embolism with either hemodynamic collapse, persistent hypotension, and/or organ hypoperfusion. The overall mortality rate associated with high-risk PE remains at approximately 30%.

Intermediate–high risk PE is a new term introduced to identify hemodynamically stable PE patients with evidence of right ventricular dysfunction.

Thrombolytics therapy is the first choice for treatment of high-risk PE with hemodynamic instability; however, in a patient who failed thrombolytics or have contraindication to thrombolytics, thrombus removal either with open surgical or catheter embolectomy is a good alternative.

We report a case of a patient who presented with hemorrhagic stroke complicated by intermediate–high-risk PE that rapidly deteriorated before undergoing successful surgical embolectomy.

Predictive Value of Red Blood Cell Distribution Width in Chronic Obstructive Pulmonary Disease Patients with Pulmonary Embolism

Purpose

This study is aimed at investigating the relationship between red cell distribution width (RDW) and chronic obstructive pulmonary disease (COPD) patients with pulmonary embolism (PE).

Methods

We conducted a retrospective study enrolling a total of 125 patients from January 2013 to December 2019. The study group consisted of 40 COPD patients with PE, and the control group had 85 COPD patients without PE. Clinical data including demographic characteristics, comorbidities, and results of imaging examinations and laboratory tests were recorded. Blood biomarkers, including red blood cell distribution width standard deviation (RDW-SD), red blood cell distribution width coefficient of variation (RDW-CV), and D-Dimer, were included.

Results

RDW-SD and RDW-CV were higher in the COPD patients with the PE group (p < 0.001). A higher RDW-SD led to a significantly increased risk of PE than a lower RDW-SD (adjusted odds ratio (OR): 1.188; 95% confidence interval (CI): 1.048-1.348). The area under the curve (AUC) of RDW-SD used for predicting PE was 0.737. Using 44.55 as the cutoff value of RDW-SD, the sensitivity was 80% and the specificity was 64.7%. The prediction accuracy of RDW-SD combined with D-Dimer (AUC = 0.897) was higher than that of RDW-SD or D-Dimer alone. The optimal cutoff value of RDW-SD+D-Dimer for predicting PE was 0.266, which generated a sensitivity of 87.5% and specificity of 83.5%.

Conclusion

RDW is significantly increased in COPD patients with PE and may thus be useful in predicting the occurrence of PE in patients with COPD.