Blood membrane interactions in hemodialysis have been shown to trigger complement (C) activation. As indicators of C-activation the anaphylatoxins (C3a and C5a) are problematical because of methodological difficulties and their kinetic properties. We developed a sensitive and specific micro-ELISA using a monoclonal antibody against neoantigens on the terminal complement complex (TCC); highly purified human TCC served as standard. Concentrations of TCC were measured in single-path perfusion systems (in vitro) and in the blood lines (arterial inlet; venous outlet) of patients on hemodialysis using steam-sterilized or ETO-sterilized dialyzers with the following membranes: cuprophan (CU), hemophan (HE) and polysulfone F6 (PS), respectively. All dialyzers with identical geometry were run under identical conditions. All membranes tested caused continuously ongoing net generation of TCC. In vitro, contact of serum with CU minidialyzers resulted in fivefold higher net release of TCC compared with HE and PS. In vivo TCC concentration-time profiles differed significantly between membranes in the rank order CU much much greater than HE greater than PS (mean basal concentration 58 x 10(-11) M; peak increase over baseline with CU 40-fold, HE fourfold, PS threefold). In addition, more TCC was generated from the same dialyzers with ETO than steam sterilization. TCC differed from C3a and C5a in the following respects: (i) lower detection limit (4 x 10(-11) vs. less than 5 x 10(-9) M for both C-anaphylatoxins); (ii) higher relative increment (inlet) during CU dialysis (25-fold vs. eightfold and twofold, respectively); (iii) C-anaphylatoxins yielded the same ranking (CU much greater than HE greater than PS), but TCC concentrations were not a linear function of C3a or C5a concentrations, respectively. Kinetic analysis (Bateman function) showed significant differences of invasion constants between membranes, that is, CU 0.088 min-1, HE 0.09, PS 0.168. The net amount of TCC released from the dialyzer was calculated under certain assumptions. It was 75.5 mg/4 hr for CU, 7.3 for HE and 5.0 for PS. The elimination constant was also dependent on the type of membrane. Using flow cytofluorometry and immunohistochemical methods (APAAP), TCC was demonstrated on membranes of granulocytes obtained during dialysis; this is compatible with potential in vivo cell activation. Generation of PGE2 and TNF alpha by adherent monocytes induced by cuprophan was C8 dependent: levels were significantly increased by addition of C8 to C8 deficient human serum concomitantly with generation of TCC.(ABSTRACT TRUNCATED AT 250 WORDS)

No abstract

The detection of hate speech online has become an important task, as offensive language such as hurtful, obscene and insulting content can harm marginalized people or groups. This paper presents TU Berlin team experiments and results on the task 1A and 1B of the shared task on hate speech and offensive content identification in Indo-European languages 2021. The success of different Natural Language Processing models is evaluated for the respective subtasks throughout the competition. We tested different models based on recurrent neural networks in word and character levels and transfer learning approaches based on Bert on the provided dataset by the competition. Among the tested models that have been used for the experiments, the transfer learning-based models achieved the best results in both subtasks.

No abstract

Preservation of private user data is of paramount importance for high Quality of Experience (QoE) and acceptability, particularly with services treating sensitive data, such as ITbased health services. Whereas anonymization techniques were shown to be prone to data re-identification, synthetic data generation has gradually replaced anonymization since it is relatively less time and resource-consuming and more robust to data leakage. Generative Adversarial Networks (GANs) have been used for generating synthetic datasets, especially GAN frameworks adhering to the differential privacy phenomena. This research compares state-of-the-art GAN-based models for synthetic data generation to generate time-series synthetic medical records of dementia patients which can be distributed without privacy concerns. Predictive modeling, autocorrelation, and distribution analysis are used to assess the Quality of Generating (QoG) of the generated data. The privacy preservation of the respective models is assessed by applying membership inference attacks to determine potential data leakage risks. Our experiments indicate the superiority of the privacy-preserving GAN (PPGAN) model over other models regarding privacy preservation while maintaining an acceptable level of QoG. The presented results can support better data protection for medical use cases in the future.