Towards an automated data cleaning with deep learning in CRESST.

The CRESST experiment employs cryogenic calorimeters for the sensitive measurement of nuclear recoils induced by dark matter particles. The recorded signals need to undergo a careful cleaning process to avoid wrongly reconstructed recoil energies caused by pile-up and read-out artefacts. We frame this process as a time series classification task and propose to automate it with neural networks. With a data set of over one million labeled records from 68 detectors, recorded between 2013 and 2019 by CRESST, we test the capability of four commonly used neural network architectures to learn the data cleaning task. Our best performing model achieves a balanced accuracy of 0.932 on our test set. We show on an exemplary detector that about half of the wrongly predicted events are in fact wrongly labeled events, and a large share of the remaining ones have a context-dependent ground truth. We furthermore evaluate the recall and selectivity of our classifiers with simulated data. The results confirm that the trained classifiers are well suited for the data cleaning task.

Distal and proximal colon cancers differ in terms of molecular, pathological, and clinical features.

BACKGROUND: Differences exist between the proximal and distal colon in terms of developmental origin, exposure to patterning genes, environmental mutagens, and gut flora. Little is known on how these differences may affect mechanisms of tumorigenesis, side-specific therapy response or prognosis. We explored systematic differences in pathway activation and their clinical implications. MATERIALS AND METHODS: Detailed clinicopathological data for 3045 colon carcinoma patients enrolled in the PETACC3 adjuvant chemotherapy trial were available for analysis. A subset of 1404 samples had molecular data, including gene expression and DNA copy number profiles for 589 and 199 samples, respectively. In addition, 413 colon adenocarcinoma from TCGA collection were also analyzed. Tumor side-effect on anti-epidermal growth factor receptor (EGFR) therapy was assessed in a cohort of 325 metastatic patients. Outcome variables considered were relapse-free survival and survival after relapse (SAR). RESULTS: Proximal carcinomas were more often mucinous, microsatellite instable (MSI)-high, mutated in key tumorigenic pathways, expressed a B-Raf proto-oncogene, serine/threonine kinase (BRAF)-like and a serrated pathway signature, regardless of histological type. Distal carcinomas were more often chromosome instable and EGFR or human epidermal growth factor receptor 2 (HER2) amplified, and more frequently overexpressed epiregulin. While risk of relapse was not different per side, SAR was much poorer for proximal than for distal stage III carcinomas in a multivariable model including BRAF mutation status [N = 285; HR 1.95, 95% CI (1.6-2.4), P < 0.001]. Only patients with metastases from a distal carcinoma responded to anti-EGFR therapy, in line with the predictions of our pathway enrichment analysis. CONCLUSIONS: Colorectal carcinoma side is associated with differences in key molecular features, some immediately druggable, with important prognostic effects which are maintained in metastatic lesions. Although within side significant molecular heterogeneity remains, our findings justify stratification of patients by side for retrospective and prospective analyses of drug efficacy and prognosis.

Treatment motivation of men with ED: what motivates men with ED to seek professional help and how can women support their partners?

Although ED can impair sexual satisfaction as well as the quality of partnership and life, men affected often avoid seeking treatment. There is growing evidence that women have an influence on their partner's help-seeking behavior. This qualitative study examined men with ED and their female partners in order to detect motivational factors for men to seek treatment and motivational actions of the women to support their partners. Twelve couples took part in a semi-structured telephone interview, which was performed separately in men and women. Analysis was on the basis of the Grounded Theory. The identified motivational factors could be divided into extrinsic (for example, media, female partner) and intrinsic (for example, desire to clarify the cause of the ED, hope for improvement) factors. Women can support their partners in treatment-seeking through various motivational actions such as talking with each other, showing interest and dealing actively with the problem, appealing to the male self-esteem, supporting the doctor's visit, forcing the treatment, active cooperation and participation in the treatment or initiating sexual intercourse. On the basis of these findings, recommendations for women were developed to support their partners and increase the probability of help-seeking behavior.