Background and Purpose — The purpose of this study was to determine the incidence, recurrence, and long-term survival rates of ischemic stroke subtypes by a mechanism-based classification scheme (Trial of ORG 10172 in Acute Stroke Treatment, or TOAST). Methods — We identified all 583 residents of the city of Erlangen, Bavaria, Germany, with a first ischemic stroke between 1994 and 1998. Multiple overlapping sources of information were used to ensure completeness of case ascertainment. The cause of ischemic stroke was classified according to the TOAST criteria. Patients were followed up at 3 months and 1 and 2 years after stroke onset. Results — The age-standardized incidence rates for the European population (per 100 000) regarding ischemic stroke subtypes were as follows: cardioembolism, 30.2 (95% CI 25.6 to 35.7); small-artery occlusion, 25.8 (95% CI 21.5 to 30.9); and large-artery atherosclerosis, 15.3 (95% CI 12 to 19.3). When age-adjusted to the European population, the incidence rate for large-artery atherosclerosis was more than twice as high for men than for women (23.6/100 000 versus 9.2/100 000). Two years after onset, patients in the small-artery occlusion subgroup were 3 times more likely to be alive than those with cardioembolism. Ischemic stroke subtype according to the TOAST criteria was a significant predictor for long-term survival, whereas subtype was not a significant predictor of long-term recurrence up to 2 years, both before and after adjustment for age and sex. Conclusions — Epidemiological observational studies that possess wide access to appropriate diagnostic technologies and apply standardized etiologic classifications provide a much better understanding of underlying risk factors for initial stroke, recurrence, and mortality.

Erythropoietin (EPO) and its receptor play a major role in embryonic brain, are weakly expressed in normal postnatal/adult brain and up-regulated upon metabolic stress. EPO protects neurons from hypoxic/ischemic injury. The objective of this trial is to study the safety and efficacy of recombinant human EPO (rhEPO) for treatment of ischemic stroke in man. The trial consisted of a safety part and an efficacy part. In the safety study, 13 patients received rhEPO intravenously (3.3 × 104 IU/50 ml/30 min) once daily for the first 3 days after stroke. In the double-blind randomized proof-of-concept trial, 40 patients received either rhEPO or saline. Inclusion criteria were age <80 years, ischemic stroke within the middle cerebral artery territory confirmed by diffusion-weighted MRI, symptom onset <8 hr before drug administration, and deficits on stroke scales. The study endpoints were functional outcome at day 30 (Barthel Index, modified Rankin scale), NIH and Scandinavian stroke scales, evolution of infarct size (sequential MRI evaluation using diffusion-weighted [DWI] and fluid-attenuated inversion recovery sequences [FLAIR]) and the damage marker S100ß. No safety concerns were identified. Cerebrospinal fluid EPO increased to 60–100 times that of nontreated patients, proving that intravenously administered rhEPO reaches the brain. In the efficacy trial, patients received rhEPO within 5 hr of onset of symptoms (median, range 2:40–7:55). Admission neurologic scores and serum S100β concentrations were strong predictors ofoutcome. Analysis of covariance controlled for these two variables indicated that rhEPO treatment was associated with an improvement in follow-up and outcome scales. A strong trend for reduction in infarct size in rhEPO patients as compared to controls was observed by MRI. Intravenous high-dose rhEPO is well tolerated in acute ischemic stroke and associated with an improvement in clinical outcome at 1 month. A larger scale clinical trial is warranted.

Summary: Background: The concept of boosting emerged from the field of machine learning. The basic idea is to boost the accuracy of a weak classifying tool by combining various instances into a more accurate prediction. This general concept was later adapted to the field of statistical modelling. Nowadays, boosting algorithms are often applied to estimate and select predictor effects in statistical regression models.Objectives: This review article attempts to highlight the evolution of boosting algorithms from machine learning to statistical modelling.Methods: We describe the AdaBoost algorithm for classification as well as the two most prominent statistical boosting approaches, gradient boosting and likelihood-based boosting for statistical modelling. We highlight the methodological background and present the most common software implementations.Results: Although gradient boosting and likelihood-based boosting are typically treated separately in the literature, they share the same methodological roots and follow the same fundamental concepts. Compared to the initial machine learning algorithms, which must be seen as black-box prediction schemes, they result in statistical models with a straight-forward interpretation.Conclusions: Statistical boosting algorithms have gained substantial interest during the last decade and offer a variety of options to address important research questions in modern biomedicine.