Noblis

Fall prevention has always been a crucial topic for injury prevention. Research shows that real-time posture monitoring and subsequent fall prevention are important for the prevention of fall-related injuries. In this research, we determine a real-time posture classifier by comparing classical and deep machine learning classifiers in terms of their accuracy and robustness for posture classification. For this, multiple classical classifiers, including classical machine learning, support vector machine, random forest, neural network, and Adaboost methods, were used. Deep learning methods, including LSTM and transformer, were used for posture classification. In the experiment, joint data were obtained using an RGBD camera. The results show that classical machine learning posture classifier accuracy was between 75% and 99%, demonstrating that the use of classical machine learning classification alone is sufficient for real-time posture classification even with missing joints or added noise. The deep learning method LSTM was also effective in classifying the postures with high accuracy, despite incurring a significant computational overhead cost, thus compromising the real-time posture classification performance. The research thus shows that classical machine learning methods are worthy of our attention, at least, to consider for reuse or reinvention, especially for real-time posture classification tasks. The insight of using a classical posture classifier for large-scale human posture classification is also given through this research.

The Environmental Consulting Industry in the United States has historically prioritized engineering approaches over geologic science in addressing groundwater contamination. This engineering-centric bias has often resulted in oversimplified conceptual site models (CSMs) that fail to capture subsurface heterogeneity, limiting the effectiveness of groundwater remediation strategies. Recognizing the critical role of geology, the industry is increasingly adopting a Remediation Geology approach, which emphasizes the development of robust geologic models as the foundation for remediation programs. Geologic models optimize site lithologic data to define subsurface permeability architecture. The geologic model primarily serves as the structure to develop a Process-Based CSM, which is a holistic model that supports the entire remediation life cycle. A Process-Based CSM addresses the physical, chemical, and biological processes governing contaminant occurrence with the goal of modeling and predicting subsurface conditions for improved decision making with respect to monitoring programs and remediation design. Case studies highlight the transformative impact of Remediation Geology and Process-Based CSMs, demonstrating significant improvements in cleanup efficiency and resource optimization across diverse hydrogeologic settings. By addressing site complexities such as fine-grained units and fracture networks, Remediation Geology and Process-Based CSMs have proven effective for contaminants ranging from chlorinated solvents to per- and polyfluoroalkyl substances (PFASs) and radionuclides.

Optical fibers are between the most common implantable devices for delivering light in the nervous system for optogenetics and infrared neural stimulation applications. Tapered optical fibers, in particular, can offer homogeneous light delivery to a large volume and spatially resolved illumination compared to standard flat-cleaved fibers while being minimally invasive. However, the use of tapers for neural applications has up to now been limited to silica optical fibers, whose large Young's modulus can cause detrimental foreign body response in chronic settings. Here, we present the fabrication and optimization of tapered fiber implants based on polymer optical fibers (POFs). After numerically determining the optimal materials and taper geometry, we fabricated two types of POFs by thermal fiber drawing. The fabrication of the taper was achieved by chemical etching of the fibers, for which several solvents previously reported in literature have been tested. The influence of different parameters on the etching process and on the quality of the obtained tapers was also investigated. The large illumination volume of the produced high-quality taper-based implants was finally tested in vitro in a brain phantom.

A network meta-analysis (NMA) was conducted to explore associations between fluorotelomer alcohol (FTOH) transformation and factors affecting perfluoroalkyl carboxylate (PFCA) evolution in synthetic and environmental media. Data were extracted from a total of 14 primary research articles aligned with bench-scale investigations into 6:2 and/or 8:2 FTOH biotransformation; these data were subjected to a random effect NMA to simultaneously evaluate influences of various experimental conditions on biotransformation. Effectiveness rankings were used to summarize experimental conditions promoting FTOH to PFCA transformation from the greatest to least effect on conversion. Results related to atmospheric conditions and/or electron accepting processes (from greatest to least effect on PFCA formation) are as follows: aerobic, nitrate reducing conditions, microoxic, anaerobic (unamended electron acceptor [EA]), sulfate reducing conditions, and iron reducing conditions. A greater association with PFCA formation was observed under mesophilic conditions compared to psychrophilic conditions. Multiple conditions that promote FTOH to PFCA bioconversion in environmental matrices were identified in this study. Several conditions under which transformation is unlikely have also been identified, suggesting the possibility of long-term FTOH retention in impacted media with limited short- and long-chain PFCA mobilization. Our findings further suggest the need for advancing our understanding of FTOH transformation in media receiving excessive nitrogen from anthropogenic sources.

Widespread adoption of artificial intelligence (AI) in civilian and defense government agencies requires the stakeholders to have trust in AI solutions. One of the five principles of ethical AI, identified by the Department of Defense, emphasizes that AI solutions be equitable. The AI system involves a series of choices from data selection to model definition, each of which is subject to human and algorithmic biases and can lead to unintended consequences. This paper focuses on allowing AI bias mitigation with the use of synthetic data. The proposed technique, named Fair-GAN, builds upon the recently developed Fair-SMOTE approach, which used synthesized data to fix class and other imbalances caused by protected attributes such as race and gender. Fair-GAN uses Generative Adversarial Networks (GAN) instead of the Synthetic Minority Oversampling Technique (SMOTE). While SMOTE can only synthesize tabular and numerical data, GAN can synthesize tabular data with numerical, binary, and categorical variables. GAN can also synthesize other data forms such as images, audio and text. In our experiments, we use the Synthetic Data Vault (SDV), which implements approaches such as conditional tabular GAN (CTGAN) and tabular variational autoencoders (TVAE). We show the applicability of Fair-GAN to several benchmark problems, which are used to evaluate the efficacy of AI bias mitigation algorithms. It is shown that Fair-GAN leads to significant improvements in metrics used for evaluating AI fairness such as the statistical parity difference, disparate impact, average odds difference, and equal opportunities difference.

Neuroblastoma is the second most common solid tumor in the first decade of life. A retrospective study reviewed (39) children treated for N.B. at Al-Mansour Teaching Hospital for children from Jan. 1st 2001 to Dec. 31st 2006, the clinical data were analyzed. Thirty nine Patients were included (23) males (59%) and (16) females (41%) , the median age at diagnosis was (48) months .The abdomen was the site of initial tumor in 31(79.4%) of patients, hypertension is found in 6 out of eleven cases (54.5%) and hepatomegaly is found in 53.8% of cases . Anemia was recorded in 26(68.4%) of patients Stage IV was the predominant clinico-pathological stage .The median survival was 13.5 months (range 1 month — 45 months). There was significant association between the stage of the disease and event free survival. There was no statistically significant association between the age at diagnosis and the outcome because about 15 cases (38%) of the patients were lost to follow-up .The study recommends the use of the new method for early diagnosis and treatment of the disease like the use of autologous bone marrow transplant and the use of antibodies directed against the tumor cells.

Cholera remains a significant public health threat for many countries, and the severity largely varies by the population and local conditions that drive disease spread, especially in endemic areas prone to natural disasters and flooding. Epidemiological models can provide useful information to military planners for understanding disease spread within populations and the effectiveness of response options for preventing the transmission among deployed and stationed personnel. This study demonstrates the use of epidemiological modeling to understand the dynamics of cholera transmission to inform emergency planning and military preparedness in areas with highly communicable diseases. Areas with higher probability for a potential cholera outbreak in Haiti followed by a natural disaster were identified. The hotspots were then used to seed an extended compartmental model, EpiGrid, to simulate notional spread scenarios of cholera originating in three distinct areas in Haiti. Disease parameters were derived from the 2010 cholera outbreak in Haiti, and disease spread was simulated over a 12-week period under uncontrolled and controlled spread. For each model location, scenarios of mitigated (intervention with 30% transmission reduction via international aid) and unmitigated (without intervention) are simulated. The results depict the geographical spread and estimate the cumulative cholera infection for each notional scenario over the course of 3 months. Disease transmission differs considerably across origin site with an outbreak originating in the department of Nippes spanning the largest geographic area and resulting in the largest number of cumulative cases after 12 weeks under unmitigated (79,518 cases) and mitigated (35,667 cases) spread scenarios. We modeled the notional re-emergence and spread of cholera following the August 2021 earthquake in Haiti while in the midst of the global COVID-19 pandemic. This information can help guide military and emergency response decision-making during an infectious disease outbreak and considerations for protecting military personnel in the midst of a humanitarian response. Military planners should consider the use of epidemiological models to assess the health risk posed to deployed and stationed personnel in high-risk areas.