SA Technologies (United States)

This study presents a one-year microbiological monitoring campaign conducted at the Oceanographic Institute of Monaco to evaluate microbial dynamics within a large public marine aquarium. The approach combined automated near-real-time flow cytometry, culture-based analyses, species identification, and antibiotic susceptibility testing. Continuous bacterial enumeration using the BactoSense® cytometer enabled high-resolution tracking of intact bacterial counts (ICC) across five hydraulic sectors, from seawater intake to exhibition tanks. The results revealed pronounced spatial gradients, with mean ICC values decreasing from 1.4 × 10⁵ cells mL⁻¹ at intake to below 1.3 × 10⁴ cells mL⁻¹ in internal tanks, confirming the high efficiency of UV-C disinfection and filtration systems. Seasonal analysis showed clear fluctuations at the intake, with peaks in spring and early summer associated with increased biological activity in the surrounding Mediterranean environment. Sector-specific variations indicated that hydraulic configuration and residence time influenced microbial stabilization, with localized differences in UV efficiency detected through near-real-time monitoring. Complementary bacterial identification revealed the presence of several opportunistic taxa, including V. fortis , V. harveyi , and Shewanella spp., typical of recirculating seawater systems. Antibiotic susceptibility testing revealed heterogeneous resistance profiles, with β-lactam resistance observed in several Vibrio isolates and multidrug resistance detected in S. algae and V. jasicida . The integration of biosensing, microbial identification, and resistance profiling provides an operational framework for evidence-based microbial surveillance in marine aquaria. These findings highlight the potential of automated near-real-time cytometry to enhance preventive management, support targeted interventions, and ensure microbiological stability in closed aquatic ecosystems. • First long-term marine use of the BactoSense® automated near-real-time flow cytometry system. • One-year monitoring revealed spatial and seasonal bacterial dynamics in a public aquarium. • UV-C disinfection and filtration reduced bacterial loads by more than one order of magnitude. • Hydraulic design and residence time shaped microbial stability across aquarium sectors. • Detection of opportunistic and resistant bacteria supports evidence-based aquarium management.

Abstract Purpose Several clinical studies have shown correlations between certain physiological measurements and an autism spectrum disorder (ASD) diagnosis. Such findings, however, have generally not resulted in tangible progress towards practical translation. In fact, most studies have been retrospective in nature and compare biomarker profiles in children with an ASD diagnosis to those who are typically developing. A clinically meaningful test, however, requires an ASD diagnostic center to distinguish children from those with a developmental disorder pre-diagnosis. Methods This paper presents, for the first time, a double-blind case/control trial design in which metabolic profiles, collected at two developmental pediatric clinics, were collected from children on a diagnostic waitlist for the purpose of developing a blood-based test for ASD. Besides obtaining blood samples, the children underwent gold-standard clinical evaluations, including the Autism Diagnostic Observation Schedule (ADOS), Mullens Scale of Early Learning (MSEL), and Vineland Adaptive Behavior Scale (VABS). The analysis, together with a complete medical history and physical exam, allowed to confirm or rule-out suspected ASD using DSM-V criteria. The study was based on a cohort of 140 children between the ages 18-60 months, that were referred to a developmental pediatrician because of concerns in their development. Results 114 of these children received an ASD diagnosis, while 26 were diagnosed with non-ASD related developmental delays. Based on the measured metabolites, artificial intelligence-based classification algorithms allowed for an over 80% accuracy in predicting whether a sample came from a child diagnosed with ASD or not. Conclusion While these results need to be replicated in a larger study, especially involving more children with non-ASD related developmental delays, this is the first work using physiological measurements, coupled with AI, to support ASD diagnoses in a clinically relevant setting.

In recent years, advancements in deep neural network research have enabled the application of deep neural network models in various fields, particularly in computer vision. However, guaranteeing a zero percent error rate remains challenging with the current technology and hinders their use in mission-critical applications. Furthermore, when deploying these models on edge devices, quantization is required to compress the model sizes and accelerate the inference speed, which can further increase the risk caused by error. To mitigate risk in stock market prediction, a loss function called the cost harmonization function has emerged, which penalizes based on the risk associated with misrecognition. In this paper, we extend the cost harmonization function as a temperature-scaled risk penalty (RP) function and realize risk- and quantization-aware training. Our experiments using a self-driving RC Car dataset and the CIFAR-10 dataset demonstrate that combining the temperature-scaled RP with cross-entropy loss effectively mitigates the risk and enables the development of more reliable quantized models.

With advent of Quantum Computing, traditional cyber security mechanisms are failing. This paper discusses the effectiveness of certain machine learning models in detecting such quantum attacks on quantum systems. The dataset used is custom generated using qiskit and AerSimulator and it focuses on three states of system, normal state, side-channel attack state and intercept-resend attack state. Machine learning models like logistic regression, neural networks, SVM and XGBoost were trained on the dataset. The highest accuracy observed was 0.81 obtained using XGBoost. These findings highlight the challenges of quantum attack detection and suggest future research directions for improving classification accuracy.

The following contribution presents a novel wind tunnel design at Colorado State University (CSU) for the study of Solid Fuel Ramjet (SFRJ) combustion. This development represents an upgrade of the existing indraft supersonic wind tunnel at CSU’s Aerospace Propulsion and Diagnostics Laboratory, creating an optically accessible testbed for SFRJ research. The design, manufacturing, and validation of flow conditions are detailed. The setup is fitted with high-speed schlieren diagnostics for flow visualization and fast response pressure transducers for the study of combustion performance. Finally, a high-power CW laser (Pmax=330 Watts) is implemented within the tunnel for remote ignition of fuel grains. The results presented herein focused on cold flow testing and fuel grain ignition. Using static pressure measurements, we were able to confirm a Mach 0.54 flow inside the combustion chamber, in line with what was predicted by our 1-D isentropic flow model. Preliminary combustion testing of HTPB fuel grains demonstrated successful laser ignition and revealed a strong recirculation zone downstream of the backward facing cavity holder. Schlieren imaging confirmed the presence of a reattaching shear layer within the cavity, critical for flame holding. Ultimately, this modular testing facility will provide a reliable cost-effective testbed for future investigations into fuel regression, flame stability, and energetic additive optimization prior to full-scale SFRJ implementation.

The form of feature extraction of conventional 3D parametric modeling of thermal power plant fan is mostly set as target extraction, with a small coverage, resulting in a decline in the final conversion efficiency ratio. Therefore, a 3D parametric modeling of thermal power plant fan based on JSOA algorithm and geometric characteristics is proposed. According to the current measurement, basic data acquisition and pre-processing shall be carried out first, and multi-step method shall be adopted to expand the coverage and realize multi-step extraction of fan geometric characteristics. Based on this, the three-dimensional parametric modeling of thermal power plant fan with JSOA algorithm and geometric characteristics is realized by means of output and delivery verification. The test results show that, compared with the traditional fan geometry 3D parametric modeling of thermal power plant fans and the traditional sparse processing 3D parametric modeling of thermal power plant fans, the designed JSOA algorithm and geometric characteristics 3D parametric modeling of thermal power plant fans finally achieve a relatively high conversion efficiency ratio, which indicates that with the help of JSOA algorithm and geometric characteristics. The designed three-dimensional parametric modeling of thermal power plant fan has higher practical application effect, stronger pertinence, and the degree of implementation of modeling meets the standards.

Clinically significant pulmonary abnormalities, like tuberculosis, nodules, and chronic obstructive pulmonary disease (COPD), hinder lung function. As a non-invasive diagnostic tool, radiology can identify various pulmonary conditions. Artificial intelligence (AI) advancements, particularly deep learning-based applications, have improved the efficiency and accuracy of radiological examinations. However, creating accurate AI models is still challenging due to the requirement for extensive manual annotation. To address this, we propose Smart Imagery Framing and Truthing (SIFT) to assist annotators in labeling and delineating pulmonary abnormalities and corresponding boundaries on CT images. Integrating Mask-RCNN, the system is developed on a training dataset of 9,078 CT images with 166,724 slices and 18,367 regions of interest (ROIs) corresponding to 47 different abnormalities/diseases. The proposed system processes an independent testing set with 2,199 CT images containing 40,493 slices and 4,280 ROIs by predicting both abnormality/disease types and corresponding ROI boundary locations. Two senior radiologists (≥ 30 years of radiological experience) delineated each image of training and testing sets in a double-blinded way, and their intersection area functioned as the ground truth. Evaluation metrics include intersection over union (IOU) value calculated based on the ROI area and the ground truth, the area under the ROC curve (AUC), and sensitivity on the slice-level of abnormality label category. For ROI segmentation, the results showed that 91.5% and 36.2% abnormalities had an IOU over 0.6 and 0.7, respectively. Regarding the metric of AUC of classification, there were 97.9%, 80.9%, and 42.6% of abnormalities whose AUC values were above 0.7, 0.8 and 0.9, respectively. For the sensitivity metric, all 47 label categories exceeded 0.8, and 68.1% (32/47) exceeded 0.9. SIFT demonstrates high performance in determining abnormality/disease types with corresponding boundary locations for the ROIs, which can be used to predict training and testing sets to develop AI technologies.

This paper presents a hybrid CNN-LSTM model integrated with GMAC for intrusion detection and secure communication in IoT networks. The system uses Yule-Simon Distribution-Based Lyrebird Optimization Algorithm for feature selection that can achieve the highest anomaly detection accuracy at minimal computational cost in real-time applications. The proposed model uses GMAC with simultaneous encryption and authentication and the CNN-LSTM hybrid for the detection of intrusion in IoT traffic data. Hence, the detection accuracy and the processing time is much better compared to the traditional methods. System detection accuracy at 99.10% was achieved with a latency of 980ms; thus, there is robust performance with minimal overhead of computation. This advanced approach improves the security of IoT and offers an effective and scalable solution for smart city and industrial IoT networks, providing high accuracy, low latency, and strong encryption.

Data security is now a major problem with the growth of cloud storage. Putting strong security measures in place has become crucial as more private data is kept on cloud servers. In order to improve the overall security of cloud storage solutions, this investigation explores a number of data protection techniques, with a particular emphasis on encryption, access control, and data integrity. In this investigation the main goal is to find and evaluate efficient security techniques for use in cloud storage systems, that utilize minimum resources whilst providing guaranteed confidentiality of sensitive data. The investigation would examine how user authentication systems, access control devices and encryption function. It also seeks to provide recommendations for a ‘defence-in-depth' security approach that protects data and helps instil trust in the user base. In this research, address the two major parts of cloud storage security; problems and their potential solutions in our the investigation through detailed literature review. It covers a range of encryption practices, access management strategies and data integrity measures. It brings expert comments as well some relevant case studies which back the dissertation's comparative study of current approaches and frameworks, specifying their pros and cons. The findings revealed that layered security measures improve the data durability of cloud storage. Encryption is one of the critical factors to safeguard sensitive information, which could be encrypted and here you go for access control process efficiently halt illicit accesses. It also clearly shows that user authentication is essential to secure data exchange and may help increase confidence of users in the cloud storage options. Protecting data in cloud-based storage via meticulous security protocols was emphasized by this study. Combining user authentication, access control and encryption is the best way to mitigate it for organizations. It offers useful tips that you may use to improve your cloud-based security protocols, something which would up user trust overall (so long as the management of private data is secured).