Search

A patent has been issued for a financial autopilot system by the United Services Automobile Association, involving machine learning and personal financial management to automate bill payments, savings, goal setting, and debt management based on historic transactions. The system uses a neural network trained with supervised learning techniques to predict expenses and create a financial plan for users. This innovative technology aims to provide a holistic financial experience for individuals. [Extracted from the article]

Researchers have submitted a patent application for an "Imaging Device Of Eliminating Electromagnetic Interference Of Magnetic Resonance And Imaging Method Thereof" to the USPTO. The technology focuses on ultra-low-field magnetic resonance imaging (ULF MRI) technology, which offers benefits such as excellent openness, portability, and clinical value in diagnosing conditions like stroke and tumors. The patent application details an imaging device and method that aim to eliminate electromagnetic interference during the imaging process, ensuring high-quality images without the need for electromagnetic shielding. The technology involves acquiring magnetic resonance imaging signals, measuring electromagnetic interference, training an interference eliminating model, and performing image reconstruction to obtain a magnetic resonance image. [Extracted from the article]

Researchers at the University of Hong Kong have developed a novel technique called the Attention-Enhanced Deep Neural Network (AEDNN) for early diabetes screening and prediction. By utilizing large-scale datasets and advanced neural network layers, the AEDNN model achieved a high prediction accuracy of 98.4% in the Pima Indians Diabetes dataset and 99.82% in the Diabetes-NHANES dataset. The model outperformed traditional methods and showed strong generalization ability, making it a valuable tool for clinicians in screening high-risk individuals for diabetes. [Extracted from the article]

A patent has been issued for paired neural networks developed by Canary Speech LLC, which can diagnose health conditions through speech analysis. The technology aims to improve the accuracy and efficiency of diagnosing various health conditions, including mental health disorders, concussions, Alzheimer's disease, and congestive heart failure, by processing speech samples from multiple time periods. By utilizing mathematical models and neural networks, this innovative approach could potentially enhance the quality of life, increase life expectancy, and reduce medical costs associated with early diagnosis and treatment of health conditions. [Extracted from the article]

A recent study conducted at Hunan University of Technology in Zhuzhou, People's Republic of China, focused on developing a pH-sensitive dual-network hydrogel for the treatment of ulcerative colitis. The hydrogel, made from sodium alginate and methacrylic acid, demonstrated excellent pH-responsive swelling properties and controlled drug release mechanisms. The research highlighted the potential of drug-loaded gels in advancing controlled drug delivery systems, emphasizing their biocompatibility and stability. This study was supported by various funding sources and has been peer-reviewed, showcasing promising advancements in the field of drug delivery for digestive system diseases. [Extracted from the article]

Researchers have submitted a patent application for a medical device designed to promote bone growth in the oral cavity. The device utilizes a polymeric nanocomposite treatment portion that can be controlled electronically and includes pressure sensors for monitoring growth pressure. This innovative system aims to address various conditions such as micrognathia, palatal abnormalities, and fractures, offering a non-surgical approach to healing and promoting tissue regeneration. The patent application outlines the detailed structure and functionality of the medical device, emphasizing its potential for supervised directional growth of teeth, bones, and soft tissue. [Extracted from the article]

A patent application titled "Medical Device Having A Polymeric Nanocomposite Actively Controlled For Rapid Healing Of Fractures And Soft Tissue Injury" was filed by inventor GHUGE, Raghavendra Vitthalrao and made available online on October 10, 2024. The patent application describes medical devices with a polymeric nanocomposite treatment device that aims to accelerate the healing process of fractures and soft tissue injuries. The device includes a controller in communication with the treatment device, featuring advanced technology such as carbon nanostructures, quantum micro-chiplets, and sensor modules for effective treatment. The goal of this innovation is to reduce healing time, minimize complications, and enhance patient rehabilitation. [Extracted from the article]

A patent application has been submitted for a method of predicting seizure onset in electroencephalography (EEG) data recorded with a subcutaneous EEG device. The method involves recording the subcutaneous EEG data, accessing a trained machine learning algorithm, transmitting the data to a computer system, and applying the data to the algorithm to generate output indicating seizure onset. The algorithm is trained using a multi-stage process and can be adjusted based on user feedback. The invention aims to improve long-term monitoring of patients with epilepsy. [Extracted from the article]

Google LLC has been issued a patent for predicting the likelihood of conditions being satisfied using neural networks. The patent describes methods for processing a temporal sequence of health events using a recurrent neural network system. The neural network analyzes a sequence of health events, such as those derived from an electronic medical record, and predicts the likelihood of future health events occurring within a specified time period. This technology has the potential to effectively analyze health data and provide insights for patient care. [Extracted from the article]

A recent report from Northwest Minzu University in Lanzhou, China discusses the use of electroencephalography (EEG) in emotion recognition tasks. The researchers propose a multilevel multidomain feature fusion network model called MMF-Net, which aims to improve the accuracy of emotion classification by effectively utilizing the high temporal resolution provided by EEG. The model achieved high accuracy rates in subject-dependent and subject-independent experiments, demonstrating its potential in emotion recognition. This research provides valuable insights for engineers and researchers working in the field of EEG-based emotion recognition. [Extracted from the article]

The US Patent and Trademark Office has published a patent application titled "Server, Switch and Endpoints for Secure Communications," which is assigned to the Government of the United States of America. The application, filed by inventor Michael Karas, discusses the use of ports in computer networks and the role of firewalls in blocking or allowing network traffic based on security rules. It proposes different configurations and methods for establishing secure communications between servers and endpoints, including the use of encrypted VPNs and various security features such as ID verification and public-private key pairs. The system described in the patent application involves a firewall and switch that administer connections between the server and endpoints, each equipped with their own processors, memory, storage media, and network interfaces. [Extracted from the article]

This article discusses a patent application for a data management system for networked endpoints. The system involves the use of ports, which are software-based points where network connections start and end. Ports allow computers to differentiate between different types of traffic and target specific services or applications. The article also explains the concept of port numbers and provides examples of commonly used ports. Additionally, it mentions the use of firewalls to block or allow network traffic based on security rules. The patent application describes various configurations of the system, including a server that verifies the identity of endpoints and stores/processes information from approved endpoints. The system aims to provide secure and efficient data transfer between the server and endpoints. [Extracted from the article]

The US Patent and Trademark Office has published a patent application titled "Time-Based Server Management System for Networked Endpoints." The application, filed by Michael Karas and assigned to the US Government, describes a system that uses ports to manage network connections and improve efficiency. It also discusses the role of firewalls in blocking specific ports for security purposes. The application provides examples of commonly used port numbers and their associated protocols. The invention aims to establish secure communication between servers and endpoints on a network. The system includes features such as antivirus software, firewalls, VPNs, and encryption for secure transmissions. It can also be used for managing service records, scan records, statistical analysis, staffing, and diagnostic reports. [Extracted from the article]

A patent has been issued to QUALCOMM Incorporated for a configurable neural network for channel state feedback (CSF) learning. The patent describes methods for wireless communication using neural network training configurations for CSF. The invention aims to improve wireless communications systems by applying neural network processing to achieve greater efficiencies. The patent provides detailed information on the technical aspects and applications of the invention. [Extracted from the article]

A patent application has been submitted for a method and apparatus that aims to address the challenges of identifying harmful drug combinations. The invention involves generating multidimensional patient-drug tensors, prediction data objects, and contraindication data objects based on patient record data and drug input vectors. These data objects are used to determine drug combination indicators and predict multi-drug contraindications. The invention utilizes machine learning models and neural networks to make these predictions. The patent application was filed in November 2022. [Extracted from the article]

The patent application titled "Method And System For Predicting Touch Interaction Position On Large Display Based On Binocular Camera" outlines a method and system for predicting touch interaction positions on a large display using a binocular camera. The method involves capturing video frames of a user's arm and facial/eye movements, identifying key frames of tapping actions, associating finger coordinates with these key frames, training a predictive network model using a convolutional network, and using a video frame of the current operation to predict the touch interaction position. The system includes modules for video capture, identification, association, training, and prediction. This technology aims to improve user experience by reducing interaction delays and increasing the responsiveness of the display. It has potential applications in software and neural networks. [Extracted from the article]

A patent has been issued to Koninklijke Philips N.V. for a system that enables an adaptive user-facing dialog agent. The system generates multiple candidate responses to user input and selects a response based on predetermined global dialog objectives. The dialog agent can align the dialog with user input by considering various dialog features such as emotion, confusion, personality, and verbosity. The system uses machine learning models and feature monitoring processes to process user input and generate candidate responses. The selected response is then presented to the user, and the system updates the machine learning models based on the selected response. [Extracted from the article]

A patent application titled "Process For Transforming A Trained Artificial Neuron Network" has been published online. The inventors propose a method for transforming a trained artificial neural network by modifying the order of the layers. Specifically, they suggest displacing the batch normalization layer after the convolution layer, which can reduce the execution time and memory occupation of the neural network. The inventors also mention the possibility of converting the pooling layer into a binary pooling layer to further simplify the execution of the transformed neural network. The patent application provides detailed claims and technical information for implementing these transformations. [Extracted from the article]