Statistica software was used for these calculations

rprints. More detail information can be seen in Method & S1 Fig. doi:10.1371/journal.pone.0122416.g005 12 / 15 PCM Modeling by New Protein Fingerprints and EPIF RMSE vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi uXn 2 u pi ti t i1 n 3 Xn RAE Xi1 n jpi ti j jti j t 4 i1 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi uXn 2 u p ti u u Xi1 i RRSE t n 2 ti t i1 5 MAE represents Mean Absolute Error, RMSE represents Root Mean Squared Error, RAE represents Relative Absolute Error and RRSE represents Root Relative Squared Error. pi represents predicted activity data calculated by different models, ti represents true activity data simulated by Hex server while t represents the mean of true activity data. ~~ Venous thromboembolism is the third most common cardiovascular disease affecting 12 per 1000 adults annually. VTE requires acute treatment with low-molecular-weightheparin followed by at least three months of therapy with oral anticoagulants, such as vitamin K-antagonists according to the American Digitoxin College of Chest Physicians guideline. Although this treatment strategy is fairly effective in preventing VTE recurrences, it causes major bleeding complications with an incidence of 2-7/100 patient years, which are associated with increased morbidity, mortality, and health care costs. Identifying patients at high risk of major bleeding events is therefore of importance and would help physicians targeting bleeding preventive strategies, such as adequate control of hypertension, discouraging the use of non-steroidal anti-inflammatory drugs or platelet-inhibitors, and frequent International Normalized Ratio monitoring. However, externally validated bleeding risk scores with adequate discriminative power are lacking for the VTE population, whereas several algorithms have been developed for patients with atrial fibrillation treated with VKA, including the HAS-BLED score . The HAS-BLED score has been validated in several independent cohorts of patients with atrial fibrillation, but it is currently unknown whether the HAS-BLED score accurately predicts major bleeding events in patients with acute VTE. The aim of our study was to analyse whether the HAS-BLED score accurately identifies patients at high risk of major bleeds during VKA treatment PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19761586 for acute VTE. Methods We identified all patients starting VKA treatment for acute VTE between January 2006 and March 2007 via records of the Leiden anticoagulation clinic. This timeframe was chosen to ensure sufficient follow-up time for included patients and treatment of acute VTE according to current clinical practice. Patients had to be treated for acute VTE by the affiliated academic or one of the two affiliated non-academic teaching hospitals of this anticoagulation clinic to be selected for inclusion for logistic reasons. VTE diagnosis was objectified by computed tomographypulmonary angiography or ultrasound. Patients were managed according to clinical practice with PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19763404 initial low-molecular-weight-heparin followed by long-term VKA therapy. The study was approved for all three participating centers by the ethics committee of the Leiden University Medical Center, Leiden, The Netherlands 2 / 11 HAS-BLED Score in Patients with Acute VTE . Patient information was anonymized prior to analysis. The need for informed consent was waived by the ethics committee. Chart review Medical records from two sources were searched f