Hamzeh Mohammadigheymasi

Oxytocin is a cyclic nonapeptide used for labor induction, labor augmentation, and the prevention and treatment of postpartum hemorrhage (PPH). It remains a first-line uterotonic for PPH prevention and treatment because of its low cost, rapid action, availability, and clinician familiarity. However, oxytocin has limited formulation stability and is highly sensitive to storage conditions. In aqueous solution, it is susceptible to heat-induced loss of potency and is therefore stored under refrigerated conditions at 2–8 °C. This requirement is difficult to maintain in lowresource supply chains, where inconsistent manufacturing, long transport routes, weak cold storage, and temperature fluctuations at the point of care can lead to poor-quality oxytocin. Heat accelerates deamidation, beta-elimination, disulfide exchange, oligosulfide formation, dimerization, oxidation, and aggregation. This review connects real-world evidence of oxytocin quality issues with the chemical processes behind its degradation, and discusses formulation approaches that can reduce or prevent specific breakdown pathways. The most defensible stabilization approaches are maintaining the formulation at approximately pH 4.5 , optimized buffer concentration, divalent metal-ion coordination, N-terminal amino-group modulation by 18-crown-6, chlorobutanol-assisted stabilization, and solid-state immobilization through inhalable dry powders or fast-dissolving tablets. Heat-stable carbetocin demonstrates that thermal stability of uterotonic peptides are feasible, but it does not eliminate the need for stable oxytocin as it is used across a wider range of obstetric indications. The review concludes with a framework and development plan that integrate supply-chain risk control, formulation design-space analysis, stability-indicating analytics, LC-MS degradation mapping, excipient screening, solid-state formulation, and predictive kinetic modeling.

This work presents the implementation of an advanced deep-learning (DL)-based earthquake detection workflow on the Bhutan Pilot Experiment dataset, a five station seismic network that was deployed from January 2002 to March 2003 in Bhutan’s Eastern Himalaya region. Previous studies reported 175 local earthquakes in this dataset, not fully understanding the seismicity potential and tectonic settings of the area. To enhance the detection and location accuracy, we reprocess the dataset using a hybrid-pair-input DL model combined with an expert analyst check, enabling a more comprehensive and precise scanning of seismic events. Central to this approach is the well-known EQTransformer (EQT) model, which serves as a core tool, followed by a Siamese EQTransformer (SEQT) to further reduce the false negative rate through a pair-wise model. To deeply scan all potential events, we set extremely low-threshold parameters for both EQT and SEQT models. This strategy led to the SEQT demonstrating an approximately detected 32% increase in the average for P phases and an average of 14% for S phases in comparison to the EQT model. The identified seismic phases were subsequently linked using the Rapid Earthquake Association and Location method via grid search, resulting in a list of 2457 detected events. A meticulous recheck of these events by an expert analyst to assess their validity and the accuracy of detected phases led to the final count of 887 events. These events, which are more than five times the size of the reported earlier (175) events for this dataset, mainly cluster in eastern and central Bhutan, in particular along the Goalpara lineament, a well-known strike-slip fault, and three concentrated clusters of events with harmonic depth and geographical extents. These findings will enhance seismic hazard assessments and advance geophysical research in the region.

Seismic networks monitor seismic activities across the globe, recording distinctive events within specific geographical and temporal frames. Whether old or new, each seismic record preserves valuable information, with its extraction relying mainly on the sophistication of the method. This study presents the implementation of an advanced earthquake detection workflow on a relatively old dataset, the Bhutan Pilot Experiment. This temporary five-station seismic network in Eastern Himalaya comprised a set of Broadband sensors deployed for 14 months from January 2002 to March 2003. However, outdated methodologies have limited the analysis of the recorded data, resulting in the reporting of only 175 local microearthquakes in this area. In this study, we reprocess the data using the recently introduced deep-scan Integrated Pair-Input deep learning and Migration Location workflow [1] to detect and locate local earthquakes. The IPIML employs the well-known Earthquake Transformer (EqT) model as its core function for initial phase picking, followed by a pair-input Siamese EQTransformer (S-EqT) to further mitigate the false negative rate using a pair-wise model. The S-EqT step demonstrated an approximately 40% increase in average detected phases compared to the standard EqT model. The detected phases are associated using the Rapid Earthquake Association and Location (REAL) method through grid searching, providing a preliminary list of detected events. This list encompasses 2458 detected events, several times larger than the previously reported catalog for this dataset. These events primarily cluster in central and eastern Bhutan, particularly along the Golpara lineament, a recognized strike-slip fault. The subsequent phase of this study involves precisely locating these events through the implementation of the Migration Location (MIL) method.References[1] H. Mohammadigheymasi et al., "IPIML: A Deep-Scan Earthquake Detection and Location Workflow Integrating Pair-Input Deep Learning Model and Migration Location Method," in IEEE Transactions on Geoscience and Remote Sensing, vol. 61, pp. 1-9, 2023, Art no. 5914109, doi: 10.1109/TGRS.2023.3293914.

The efficacy of feed-forward multi-layer neural networks relies heavily on their training procedure, where identifying appropriate weights and biases plays a pivotal role. Nonetheless, conventional training algorithms such as backpropagation encounter limitations, including getting trapped in sub-optimal solutions. To rectify these inadequacies, metaheuristic population algorithms are advocated as a dependable alternative. In this paper, we introduce a novel training methodology termed, DDE-OP, which leverages the principles of differential evolution enriched with a division-based scheme and an opposite-direction strategy. Our approach integrates two effective concepts with differential evolution. Initially, the proposed algorithm identifies partitions within the search space through a clustering algorithm and designates the obtained cluster centres to serve as representatives. Subsequently, an updating scheme incorporates these clusters into the current population. Lastly, a quasi-opposite-direction strategy is used to augment search space exploration. Extensive evaluation on diverse classification and approximation tasks demonstrate that DDE-OP surpasses conventional and population-based methodologies.

Laplacian learning method is a well-established technique in classical graph-based semi-supervised learning, but its potential in the quantum domain remains largely unexplored. This study investigates the performance of the Laplacian-based Quantum Semi-Supervised Learning (QSSL) method across four benchmark datasets -- Iris, Wine, Breast Cancer Wisconsin, and Heart Disease. Further analysis explores the impact of increasing Qubit counts, revealing that adding more Qubits to a quantum system doesn't always improve performance. The effectiveness of additional Qubits depends on the quantum algorithm and how well it matches the dataset. Additionally, we examine the effects of varying entangling layers on entanglement entropy and test accuracy. The performance of Laplacian learning is highly dependent on the number of entangling layers, with optimal configurations varying across different datasets. Typically, moderate levels of entanglement offer the best balance between model complexity and generalization capabilities. These observations highlight the crucial need for precise hyperparameter tuning tailored to each dataset to achieve optimal performance in Laplacian learning methods.

This abstract presents the findings of a comprehensive seismicity analysis conducted in Ghana using the advanced Deep-Scan workflow [1]. The study compiled data from 461 earthquakes, revealing new insights into active seismogenic sources in this relatively under-explored region. A more accurate crustal velocity model is computed, and the detailed Green functions for focal mechanisms of detected events are estimated. Data integration from a single Ivory Coast station improved the azimuthal coverage of hypocenters, facilitating more precise velocity model estimations for deeper layers. The study utilized 5213 accurately picked arrival phases from 284 earthquakes, applying joint inversion techniques to optimize crustal velocity and hypocentral parameters. This resulted in a detailed 6-layer model, with P-wave velocity values ranging from 5.6 to 6.3 km/s and a high-velocity layer at 7.8 km/s, maintaining a consistent Vp/Vs ratio of 1.70. Analysis of the earthquakes revealed six compact clusters, predominantly aligned with the Akwapim fault zone and linked to the active section of the Romanche fracture zone in Axim. The distribution of hypocentral depths consistently falls within the range of 9-18 km, confined to the upper crust of the region. Utilizing the revised velocity model notably enhanced the identification of the seismogenic zone in southern Ghana. Moreover, it facilitated the generation of detailed Green functions for the next step for moment tensor (MT) inversion of small-magnitude earthquakes, employing a bootstrap-based probabilistic inversion schema. The results enhance our understanding of Ghana’s seismicity and represent key points for hazard and risk assessment in the region. This research is supported by Funda¸c˜ao para a Ciˆencia e Tecnologia (UIDB/00645/2020 and https://doi.org/10.54499/UIDB/00645/2020, and UIDB/50008/2020).References[1] H. Mohammadigheymasi et al., "IPIML: A Deep-Scan Earthquake Detection and Location Workflow Integrating Pair-Input Deep Learning Model and Migration Location Method," in IEEE Transactions on Geoscience and Remote Sensing, vol. 61, pp. 1-9, 2023, Art no. 5914109, doi: 10.1109/TGRS.2023.3293914.

Optimized Deep Learning (DL)-based workflows can improve the efficiency and accuracy of earthquake detection and location processes. This paper introduces a six-step automated event detection, phase association, and earthquake location workflow, which integrates the state-of-the-art Pair-Input DL model and waveform Migration Location methods (IPIML). Applying IPIML on an 18-months dataset of Ghana Digital Seismic Network (GHSDN) recorded from 2012-2014, a catalog with 461 events is automatically obtained. Compared to other DL catalogs obtained using EQTransformer (EQT) and Siamese EQTransformer (S-EQT), the seismic event clusters in the IPIML catalog focus more on tectonically active regions or known seismogenic source areas and show a consistent depth distribution. The compiled catalog is 6.3 times larger than the reported catalog obtained by applying EQT with the default settings, indicating the importance of optimization and hyper-parameter tuning when applying DL models. As a result, a previously unknown seismogenic fault with a clear spatial trend has been identified using the new IPIML catalog, which provides more insights into the fault activities and seismic hazards in the region. The IPIML codes and data sets are available at the GitHub repository https://github.com/SigProSeismology/IPIML.git, contributing to the geoscience community.

A temporary seismic network consisting of 32 broadband seismic sensors was installed in Cameroon between March 2005 and December 2006 to study the seismic structure of the crust and upper mantle beneath the Cameroon Volcanic Line (CVL). This study aims to re-evaluate the seismicity in this period by processing this database and calculating an updated crustal velocity model for the region incorporating the acquired earthquake bulletin. The earthquake detection and location procedure applies hybrid deep learning (DL) and phase validation methods. We use an integrated workflow composed of Earthquake Transformer (EqT) and Siamese Earthquake Transformer (S-EqT) for initial earthquake detection and phase picking. Then, PickNet is used as a phase refinement step, and REAL for earthquake association and rough location. A set of thresholding parameters for earthquake detection and P- and S-picking equal to 0.2 and 0.07 are adjusted, respectively. By combining a set of 33282 P and 29251 S-picked phases associated with 743 earthquakes with 1.3 ≤ ML ≤ 4.6, we implement a joint inversion for estimating an updated 1D crustal velocity model. The obtained mode comprises thicknesses of 8, 12, 14, 20, and 30km, from the surface to a depth of 45km, with Vp = 6.1, 6.4, 6.6, 7.6, 8.25, and 8.5km/s, respectively. The newly detected events are primarily concentrated in three main clusters, 1) the east flank of Mount Cameroon, 2) an area between Mount Cameroon and Bioko Island, and 3) southern Bioko Island. The compiled catalog for this time interval is 1.7 times larger than the already reported catalog for this data set. Finally, we present a 3D time-lapsed animation of the detected earthquake sequences.Acknowledgements: The authors would like to thank the support of the Instituto de Telecomunicações. This work is funded by FCT/MCTES through national funds and, when applicable co-funded EU funds under the project UIDB/50008/2020.