University of Applied Sciences and Arts Northwestern Switzerland

Probabilistic Stimulation Maps (PSMs) are increasingly employed to identify brain regions associated with optimal therapeutic outcomes in Deep Brain Stimulation (DBS). However, their reliability and generalizability are challenged by the limited size of most patient cohorts and the inherent variability introduced by different statistical methods and input data configurations. This study aimed to investigate the geometrical variability of Probabilistic Sweet Spots (PSS) as a function of both the number of patients (nPat) and the number of stimulations per patient (nStim), and to model a stability boundary defining the minimum data requirements for obtaining geometrically stable PSS. Three statistical approaches-Bayesian t-test, Wilcoxon test with False Discovery Rate (FDR) correction, and Wilcoxon test with nonparametric permutation correction-were applied to two patient cohorts: a primary cohort of 36 patients undergoing DBS for Parkinson's Disease (PD), and a secondary cohort of 61 patients treated for Essential Tremor (ET), used to assess generalizability. Stimulation test data was collected intra-operatively for the first cohort and post-operatively for the second one. Geometric stability was evaluated based on variability in PSS volume extent and centroid location. The analysis revealed a non-linear trade-off between nPat and nStim to yield stable PSS. A stability boundary was defined, representing the minimum combinations of nPat-nStim required for anatomically robust PSS. Among the tested methods, the Bayesian t-test achieved stability with smaller sample sizes (∼15 patients) and demonstrated a consistent performance across both cohorts. In contrast, the Wilcoxon-based methods showed variable behavior between cohorts, which differed in symptom type and testing phase (intra-operative testing vs. post-operative screening). The proposed PSS stability boundary provides a practical reference for designing DBS studies and stimulation screening protocols aimed at probabilistic mapping. The Bayesian t-test emerged as a reliable method across both cohorts, supporting its potential in studies with limited sample sizes and scenarios where the method needs to be readily generalized to varying symptoms. These findings underscore the importance of considering both cohort size and stimulation count in probabilistic DBS mapping and call for further investigation into method-specific sensitivities to clinical and procedural factors.

Context. Solar hard X-ray (HXR) observations provide diagnostics of the hottest plasmas and of nonthermal electron populations present during solar flares and coronal mass ejections. HXR images of specific energy ranges often contain overlapping contributions of these components, complicating their interpretation. This is even more challenging as HXR imagers generally use an indirect imaging system. Aims. Our work aims to separately image individual spectral components, such as thermal loops, superhot sources, and nonthermal footpoint sources, rather than obtaining images of specific energy ranges that show a combination of all components. Methods. We introduced a new method called “spectral component imaging” and applied it to observations provided by the Spectrometer/Telescope for Imaging X-rays (STIX) aboard Solar Orbiter. First, the flare integrated HXR spectrum was fit with individual spectral components to get the relative contributions (“weights”) of each component in each native STIX energy channel. In a second step, a set of linear equations was created based on these weights and the observed, energy-dependent STIX visibilities. The visibilities of the individual spectral components were derived by means of a linear least-squares approach and were subsequently utilized for image reconstructions. Results. We demonstrate the effectiveness of spectral component imaging on four different flares observed by STIX. This method provides powerful diagnostics, particularly for flares with hot and superhot components, allowing us to spatially separate these two thermal components. We applied our methodology to the nonthermal peak of the X7.1 flare SOL2024-10-01, and we find that the superhot component is located 4.8 Mm away from the hot thermal loops. The thermal energy of the superhot component is approximately 20% of the energy content of the hot component, highlighting the significance of superhot components in the total flare energy budget. Conclusions. Spectral component imaging provides a powerful tool to image individual spectral components (i.e., thermal and nonthermal X-ray sources), rather than creating images over fixed energy ranges. Because there is no need to select an energy range, spectral component imaging has the potential to automate the image reconstruction process and to establish a robust STIX image database once the spectral components have been defined.

(Abridged) The bimodality between star-forming discs and quiescent spheroids requires the existence of two main processes: the galaxy quenching and the morphological transformation. In this paper, we aim to understand the link between these processes and their relation with the stellar mass of galaxies and their local environment. Taking advantage of the first data released by the Euclid Collaboration, covering more than 60 deg2 with space-based imaging and photometry, we analyse a mass-complete sample of nearly one million galaxies in the range 0.2510^{9.5} M_\odot$. We divide the sample into four sub-populations of galaxies, based on their star-formation activity and morphology. We then analyse the physical properties of these populations and their relative abundances in the stellar mass vs. local density plane. Together with confirming the passivity-density relation and the morphology-density relation, we find that quiescent discy galaxies are more abundant in the low-mass regime of high-density environment. At the same time, star-forming bulge-dominated galaxies are more common in field regions, preferentially at high masses. Building on these results and interpreting them through comparison with simulations, we propose a scenario where the evolution of galaxies in the field significantly differs from that in higher-density environments. The morphological transformation in the majority of field galaxies takes place before the onset of quenching and is mainly driven by secular processes taking place within the main sequence, leading to the formation of star-forming bulge-dominated galaxies as intermediate-stage galaxies. Conversely, quenching of star formation precedes morphological transformation for most galaxies in higher-density environments. This causes the formation of quiescent disc-dominated galaxies before their transition into bulge-dominated ones.

This study examines how the integration of structured information affects the performance of large language models (LLMs) in the context of facility management. The aim is to determine to what extent structured data such as maintenance schedules, room information, and asset inventories can improve the accuracy, correctness, and contextual relevance of LLM-generated responses. We focused on scenarios involving function calling of a database with building information. Three use cases were developed to reflect different combinations of structured and unstructured input and output. The research follows a design science methodology and includes the implementation of a modular testing prototype, incorporating empirical experiments using various LLMs (Gemini, Llama, Qwen, and Mistral). The evaluation pipeline consists of three steps: user query translation (natural language into SQL), query execution, and final response (translating the SQL query results into natural language). The evaluation was based on defined criteria such as SQL execution validity, semantic correctness, contextual relevance, and hallucination rate. The study found that the use cases involving function calling are mostly successful. The execution validity improved up to 67% when schema information is provided.

Abstract Deep Brain Stimulation (DBS) is an effective treatment for movement disorders. Optimizing stimulation parameters remains, however, a trial-and-error process. Datadriven models leveraging Probabilistic Mapping have shown promise in predicting DBS outcomes, yet current studies rely on chronic stimulation data. This study explores the feasibility of using intra-operative stimulation test data for DBS effect prediction. Probabilistic volumes of beneficial and adverse effects were computed from intra-operative stimulation test data of 65 patients (23 with Essential Tremor + 42 with Parkinson’s Disease). A prediction dataset was generated including clinical, morphological, stimulation features along with features derived from probabilistic maps and simulated Volumes of Tissue Activated. Three machine learning models (Adaboost, Support Vector Classifier and Naïve Bayes) were implemented to predict stimulation effects in a classification task. The models were validated in a leave-one-out crossvalidation and their performances were compared. All the developed models were able to predict DBS outcome classes. The best predictive performance was achieved by the Adaboost model with a maximum balanced accuracy of 0.71 on 3 classes. These results show that intra-operative stimulation test data can predict DBS effects with a similar approach and comparable accuracy to post-operative monopolar review data.

We present two extensive sets of 3500+1000 simulations of dark matter haloes on the past light cone, and two corresponding sets of simulated (`mock') galaxy catalogues that represent the Euclid spectroscopic sample. The simulations were produced with the latest version of the PINOCCHIO code, and provide the largest, public set of simulated skies. Mock galaxy catalogues were obtained by populating haloes with galaxies using an halo occupation distribution (HOD) model extracted from the Flagship galaxy catalogue provided by Euclid Collaboration. The Geppetto set of 3500 simulated skies was obtained by tiling a 1.2 Gpc/h box to cover a light-cone whose sky footprint is a circle of 30 deg radius, for an area of 2763 deg$^2$ and a minimum halo mass of $1.5\times10^{11}$ Msun/h. The relatively small box size makes this set unfit for measuring very large scales. The EuclidLargeBox set consists of 1000 simulations of 3.38 Gpc/h, with the same mass resolution and a footprint that covers half of the sky, excluding the Milky Way zone of avoidance. From this we produced a set of 1000 EuclidLargeMocks on the 30 deg radius footprint, whose comoving volume is fully contained in the simulation box. We validated the two sets of catalogues by analysing number densities, power spectra, and 2-point correlation functions, showing that the Flagship spectroscopic catalogue is consistent with being one of the realisations of the simulated sets, although we noticed small deviations limited to the quadrupole at k>0.2 h/Mpc. We show cosmological parameter inference from these catalogues and demonstrate that using one realisation of EuclidLargeMocks in place of the Flagship mock produces the same posteriors, to within the expected shift given by sample variance. These simulated skies will be used for the galaxy clustering analysis of Euclid's Data Release 1 (DR1).

The paper describes the objectives, design and findings of the pre-launch ground characterisation campaigns of the Euclid infrared detectors. The pixel properties, including baseline, bad pixels, quantum efficiency, inter pixel capacitance, quantum efficiency, dark current, readout noise, conversion gain, response nonlinearity, and image persistence were measured and characterised for each pixel. We describe in detail the test flow definition that allows us to derive the pixel properties and we present the data acquisition and data quality check software implemented for this purpose. We also outline the measurement protocols of all the pixel properties presented and we provide a comprehensive overview of the performance of the Euclid infrared detectors as derived after tuning the operating parameters of the detectors. The main conclusion of this work is that the performance of the infrared detectors Euclid meets the requirements. Pixels classified as non-functioning accounted for less than 0.2% of all science pixels. IPC coupling is minimal and crosstalk between adjacent pixels is less than 1% between adjacent pixels. 95% of the pixels show a QE greater than 80% across the entire spectral range of the Euclid mission. The conversion gain is approximately 0.52 ADU/e-, with a variation less than 1% between channels of the same detector. The reset noise is approximately equal to 23 ADU after reference pixels correction. The readout noise of a single frame is approximately 13 $e^-$ while the signal estimator noise is measured at 7 $e^-$ in photometric mode and 9 $e^-$ in spectroscopic acquisition mode. The deviation from linear response at signal levels up to 80 k$e^-$ is less than 5% for 95% of the pixels. Median persistence amplitudes are less than 0.3% of the signal, though persistence exhibits significant spatial variation and differences between detectors.

With an increasing focus on sustainable technologies in the pharmaceutical industry, milling provides a solvent-free approach to improve drug dissolution. Milling of drugs with an excipient offers additional opportunities to achieve supersaturation kinetics. Therefore, this work aims to present insights of co-milling fenofibrate and apremilast, two good glass formers with low and high glass transition temperatures (Tgs) respectively. Drugs were co-milled with croscarmellose sodium for various process durations followed by thermal analysis, investigation of crystallinity, surface area and dissolution. The dissolution enhancement of the low-Tg glass former fenofibrate highly correlated with the process-induced increase in surface area of co-milled systems (R2 = 0.96). In contrast, the high-Tg glass former apremilast lost its crystalline order gradually after ≥ 10 min of co-milling, and favourable supersaturation kinetics during biorelevant dissolution testing were observed. Interestingly, the melting point of co-milled apremilast decreased and linearly correlated with the highest measured drug concentration (cmax) during in vitro dissolution (onset temperature R2 = 0.98; peak temperature R2 = 0.96). The melting point depression remained stable after 90 days for apremilast, whereas fenofibrate co-milled for 20 min or more showed an increase in melting point upon storage. This study demonstrated that co-milling with croscarmellose sodium is ideally suited to good glass formers with a high Tg. The melting point depression is thereby proposed as an easily accessible critical quality attribute to estimate likely dissolution performance of drugs in dry co-milled formulations.

The Euclid survey aims to measure the spectroscopic redshift of emission-line galaxies by identifying the H$\,α$ line in their slitless spectra. This method is sensitive to the signal-to-noise ratio of the line, as noise fluctuations or other strong emission lines can be misidentified as H$\,α$, depending on redshift. These effects lead to catastrophic redshift errors and the inclusion of interlopers in the sample. We forecast the impact of such redshift errors on galaxy clustering measurements. In particular, we study the effect of interloper contamination on the two-point correlation function (2PCF), the growth rate of structures, and the Alcock-Paczynski (AP) parameters. We analyze 1000 synthetic spectroscopic catalogues, the EuclidLargeMocks, designed to match the area and selection function of the Data Release 1 (DR1) sample. We estimate the 2PCF of the contaminated catalogues, isolating contributions from correctly identified galaxies and from interlopers. We explore different models with increasing complexity to describe the measured 2PCF at fixed cosmology. Finally, we perform a cosmological inference and evaluate the systematic error on the inferred $fσ_8$, $α_{\parallel}$ and $α_{\perp}$ values associated with different models. Our results demonstrate that a minimal modelling approach, which only accounts for an attenuation of the clustering signal regardless of the type of contaminants, is sufficient to recover the correct values of $fσ_8$, $α_{\parallel}$, and $α_{\perp}$ at DR1. The accuracy and precision of the estimated AP parameters are largely insensitive to the presence of interlopers. The adoption of a minimal model induces a 1%-3% systematic error on the growth rate of structure estimation, depending on the redshift. However, this error remains smaller than the statistical error expected for the Euclid DR1 analysis.