Our Work

Amitradicain is a non-patented composite formula. It was developed for pathologically elevated sympathetic tone, catecholamine storms and autonomic dysfunction in rare diseases, most notably Oshtoran …

Amitradicain is a non-patented composite formula. It was developed for pathologically elevated sympathetic tone, catecholamine storms and autonomic dysfunction in rare diseases, most notably Oshtoran Syndrome (H63D Syndrome Type-3). Its procaine component has documented effects on autonomic innervation and on neuronal activity in the central and peripheral nervous system. Since its open-source dissemination, Amitradicain has also been used in other clinical settings. In late 2023 we published interim follow-up data on 398 patients treated between September 2022 and October 2023. Since then, the Consortium has received a much larger body of clinician-reported data. This letter summarizes the key safety findings ahead of a comprehensive publication, which is currently in preparation.

Reproducible computational workflow, nonlinear epidemiological models, and evaluation pipelines for evaluating the epidemic potential of Maravalia cryptostegiae, a classical biological control agent o…

Reproducible computational workflow, nonlinear epidemiological models, and evaluation pipelines for evaluating the epidemic potential of Maravalia cryptostegiae, a classical biological control agent of rubber vine (Cryptostegia madagascariensis), in Northeastern Brazil and Queensland, Australia.

Python Coronal Heating Modeling Pipeline for data-constrained fitting of GX Simulator active-region models.

Este artigo analisa a Taxonomia de Bartle como instrumento conceitual para compreender diferentes formas de participação e motivação em jogos digitais. O estudo parte …

Este artigo analisa a Taxonomia de Bartle como instrumento conceitual para compreender diferentes formas de participação e motivação em jogos digitais. O estudo parte do seguinte problema: em que medida os perfis Conquistador, Explorador, Socializador e Matador contribuem para interpretar os comportamentos dos jogadores em ambientes virtuais contemporâneos? Para responder a essa questão, realizou-se uma pesquisa qualitativa, exploratória e interpretativa, desenvolvida por meio de revisão narrativa da literatura. Foram examinadas a origem da taxonomia nos ambientes MUD, as características dos quatro perfis, estudos posteriores sobre motivação para jogar e as relações entre comportamento, contexto e design dos jogos. Os resultados indicam que os jogadores dificilmente apresentam uma única motivação fixa, sendo frequente a combinação de características de diferentes perfis. Também se observou que as preferências podem variar conforme o gênero do jogo, as regras, os sistemas de recompensas, a presença de outros participantes e a organização das comunidades digitais.

Code and data accompanying the manuscript "Aligning Neural Population Patterns Facilitates Motor Learning Transfer". The repository hosts the implementation of Clustering Alignment-based Transfer on K…

Code and data accompanying the manuscript "Aligning Neural Population Patterns Facilitates Motor Learning Transfer". The repository hosts the implementation of Clustering Alignment-based Transfer on Kernel Reinforcement Learning (CATKRL) and a minimum segment of the dataset needed to interpret and verify the findings.

В тезисе рассматривается взаимосвязь синдрома обструктивного апноэ сна с риском инфаркта миокарда и ишемического инсульта. Рассмотрены основные механизмы, включая интермиттирующую гипоксию, колебания …

В тезисе рассматривается взаимосвязь синдрома обструктивного апноэ сна с риском инфаркта миокарда и ишемического инсульта. Рассмотрены основные механизмы, включая интермиттирующую гипоксию, колебания артериального давления, активацию симпатической нервной системы и эндотелиальную дисфункцию. Отмечена роль сопутствующих факторов, таких как артериальная гипертензия, ожирение, сахарный диабет и нарушения липидного обмена, в формировании сердечно- сосудистого риска.

This release contains the research artifacts for “Trustworthy Evaluation Frameworks for AI-Assisted Painting Restoration,” a master’s thesis at the University of Applied Sciences Tec…

This release contains the research artifacts for “Trustworthy Evaluation Frameworks for AI-Assisted Painting Restoration,” a master’s thesis at the University of Applied Sciences Technikum Wien. The study develops and applies an evidence-based evaluation framework combining traditional image quality metrics with perceptual, colour, texture, boundary, structural, robustness, and generative-uncertainty evidence. The Controlled-300 study covers 300 paintings across five categories, with 3,425 registered cases and 2,620 eligible cases. Four methods: OpenCV Telea, LaMa, HINT, and Stable Diffusion, provide 10,480 matched primary candidates. The broader reporting collection contains 13,879 retained candidates. Additional investigations include 1,025 four-seed Stable Diffusion uncertainty groups, damage-size sensitivity, mask robustness, synthetic degradation, and separate diagnostic studies. SDXL is included as a bounded feasibility investigation, not as a fifth full-coverage method. Across 2,320 matched non-zero-damage cases, LaMa achieved the best mean masked MAE (14.83) and crop LPIPS (0.1292), while Telea achieved the best mean crop SSIM (0.8632). These complementary results motivate examining restoration quality through multiple forms of saved evidence. The archive brings together source code, notebooks, configurations, painting provenance, experimental outputs, candidate restorations, diagnostic maps, metric tables, reports, and supporting documentation. An accompanying eight room Streamlit dashboard presents the saved findings interactively. Release manifests and checksums document the archived files. Model weights and execution environments are not included. Licensing and third-party exclusions are described in the accompanying rights notice. Download and extraction: The research archive is distributed as eight numbered parts (.zip.001 to .zip.008). Download all eight parts and follow REASSEMBLE.txt to reconstruct and verify the original ZIP before extraction. The parts are not independently extractable archives. SPLIT_MANIFEST.json records each part’s size and checksums; SHA256SUMS.txt covers the uploaded files. Splitting changes only the delivery format: the research contents remain unchanged. Post-publication corrections: 3 October 2026 Please read this release together with the reporting errata and corrected exports. These correct T04/T05 population labels and clarify case-weighted aggregation; distinguish the canonical statistical population of 1,200 cases across 300 paintings from all-branch runtime analyses; and clarify SDXL coverage as 24 completed cases across 19 paintings, versus 35 scheduled cases across 30 paintings. Original archived files, scientific measurements and numerical statistical results remain unchanged. The linked documentation also records completed regression repairs, validation results and retained historical provenance exceptions.   Source code and documentation Interactive research dashboard Candidate restoration repository Diagnostics repository