Anne-Laure Boulesteix's research while affiliated with Technische Universität München and other places

... However, future investigations using more advanced RT techniques and larger sample sizes are essential to further validate and refine these observations, potentially enhancing treatment efficacy and patient wellbeing. 15 Bodensohn et al recently published their study comparing Stereotactic radiosurgery (SRS) to whole-brain radiotherapy (WBRT) in patients with 4-10 brain metastases. The study found that SRS was well-tolerated, with no grade III toxicities reported. ...

... The MRI DSC perfusion parameters used in our work (flip angle of 90 • without preloading) do not fully align to current recommendations (flip angle of 60 • with preloading or flip angle of 30 • without preloading), which is a limitation due to the retrospective nature of the study. For the same reason, a newer MRI method, contrast clearance analysis (CCA) with treatment response assessment maps, which has shown high accuracies for the differentiation of tumor progression from pseudoprogression, was not assessed [45,46]. CCA, however, is associated with delayed imaging up to 105 min after administration of a gadoliniumbased contrast agent which may be challenging to implement in daily routine. ...

... An appropriate patient selection is essential to achieve survival benefit. According to international recommendations and prognostic score indexes, reirradiation should be considered in young patients with good performance status, and at least 6 months interval from the first course of RT [61,[64][65][66][67][68]. Survival benefit is longer in patients with lower grade gliomas compared with glioblastoma. ...

... Most studies used unsupervised principal component analysis (PCA), which is capable of revealing the separation of data between groups. Through this approach, potential bias can be revealed [33]. ...

... Benchmarking studies may be performed by independent groups interested in systematically comparing existing methods or by authors of new methods to demonstrate performance improvements or other advantages over existing competitors. With regard to studies performed by independent groups, there have been several benchmarking studies of clustering for continuous data only or categorical data only (e.g., Milligan 1980;Meilȃ and Heckerman 2001;Ferreira and Hitchcock 2009;Saraçli et al. 2013;Boulesteix and Hatz 2017;Javed et al. 2020;Hennig 2022), whereas benchmarking studies of clustering for mixed-type data are scarce (Jimeno et al. 2021;Preud'Homme et al. 2021). We can also distinguish benchmarking studies of clustering for mixed-type data that are part of original papers where new methods are proposed (e.g., Ahmad and Dey 2007;Hennig and Liao 2013;Foss et al. 2016). ...

... 36,38,39,45,46 Cancer cells induce the production of IL-22 from Th cells in breast and lung cancer patients. 12,47,48 NLRP3 inflammasomedriven release of IL-1b induces IL-22 production from T cells in the tumor, and both IL-22 + Th cells and an NLRP3-IL-1b signature can be found in tumor samples of breast and lung cancer. 48,49 Here, we set out to delineate a mechanism whereby IL-22 promotes breast and lung cancer progression. ...

... Until recently, genetic risk prediction studies on complex diseases focused on variants identified as significantly associated in genome-wide association studies (GWAS) or candidate gene association studies (Müller et al., 2016). Studies on cardiovascular disease-related events included only a few literature-based confirmed SNPs in their analyses to investigate the predictive value of genetic information in addition to established risk prediction models (Antiochos et al., 2016;Beaney et al., 2017;Iribarren et al., 2016; R. W. Morris et al., 2016;Paynter et al., 2010;Said et al., 2018;Tada et al., 2016;de Vries et al., 2015). ...

... Therefore, Enet and SPLS are two potential powerful methods which are suitable for addressing the NIR spectroscopy data. In fact, many references [27][28][29][30][31][32][33][34][35][36][37][38] have introduced Enet or SPLS to analysis of NIR spectroscopy data. The purpose of this article is to compare the performance of them when dealing with the NIR spectroscopy data. ...

... In biomedical research, for example, measuring the effects of biomedical markers w.r.t. model prediction is as essential as measuring their added value regarding model performance [4]. We use the term feature importance 1 to describe how important the feature was for the predictive performance of the model, regardless of the shape (e.g., linear or nonlinear relationship) or direction of the feature effect. ...

... Besides the commonly tuned ntree, mtry and nodesize hyper-parameters, we also tested different values related to the sampling scheme. Sampling with replacement can lead to biased variable split selection when there are many categorical covariates with varying numbers of levels (Janitza et al., 2016;Strobl et al., 2007). Hence, we tested sampling without replacement because we had many categories that were not balanced, hoping to achieve a less biased use of all levels across the trees in the forest. ...