Dsilo, Inc.

Koninklijke Philips Corporation N.v.

Lalit Gupta

Jaya Prakash Narayana Gutta

Sharad Malhautra

John Petruzzello

Ajay Anand

Rajendra Singh Sisodia

Pallavi Vajinepalli

Celine Firtion

Ganesan Ramachandran

Shiwei Zhou

Some embodiments may perform operations of a process that includes obtaining a natural language text document and use a machine learning model to generate a set of attributes based on a set of machine-learning-model-generated classifications in the document. The process may include performing hierarchical data extraction operations to populate the attributes, where different machine learning models may be used in sequence. The process may include using a pre-trained Bidirectional Encoder Representations from Transformers (BERT) model augmented with a pooling operation to determine a BERT output via a multi-channel transformer model to generate vectors on a per-sentence level or other per-text-section level. The process may include using a finer-grain model to extract quantitative or categorical values of interest, where the context of the per-sentence level may be retained for the finer-grain model.

Database generation from natural language text documents

A process includes obtaining a document, determining a set of vectors based on a count of n-grams of the document, and determining a first set of information based on the document using a first set of neural networks. The process includes selecting a text section of the natural language document using a second set of neural networks and a code template of a plurality of code templates based on the text section based on the first set of information and the text section. The process includes determining an entity identifier, a value of a conditional statement, a second set of information, and a third set of information based on the text section, the first set of information, and the code template. The process includes generating a first set of program code based on the entity identifier, the value, the second set of information, and the third set of information.

Self-executing protocol generation from natural language text

Some embodiments may obtain a natural language question, determine a context of the natural language question, and generate a first vector based on the natural language question using encoder neural network layers. Some embodiments may access a data table comprising column names, generate vectors based on the column names, and determine attention scores based on the vectors. Some embodiments may update the vectors based on the attention scores, generating a second vector based on the natural language question, determine a set of strings comprising a name of the column names and a database language operator based on the vectors. Some embodiments may determine a values based on the determined database language operator, the name, using a transformer neural network model. Some embodiments may generate a query based on the set of strings and the values.

Database query generation using natural language text

Disclosed is an ultrasound device for measuring blood flow velocity in a blood vessel of a subject without imaging functionality in the device. The measurement depends upon reflections of a collimated beam of ultrasound from a subject's body part. Received electrical signals representative of the reflected ultrasound energy is used for generating a representation of blood flow at a plurality of predetermined locations in the volume and calculating a first blood flow velocity at each of the locations. The representation of flow is used for delineating the blood flow in the blood vessel in the volume. An angle calculating unit calculates the Doppler angle between the direction of the radiated collimated beam in the delineated blood flow at each point. A velocity calculator calculates a second blood flow velocity at the plurality of points based on the calculated first velocities and the calculated angle at the point.

Method and apparatus for automated ultrasonic doppler angle and flow velocity estimation

A device is configured for interrogating a blood vessel to derive flow characteristics (S) and for, responsive to the deriving and based on the derived characteristics, anatomically identifying the vessel. A spatial map of the vessels may be generated based on the interrogating, and specifically the Doppler power computed from data acquired in the interrogating. Subsequent interrogating (S) may occur, based on the map and on a user-selected set of vessels and/or vessel categories, to derive clinical Doppler indices. The device can be designed to automatically set a sample volume () for the subsequent interrogating, and to operate automatically from the user selection to display of the indices. The display may further include an image () of the vessels summoned by the set, annotated by their individual anatomical names, and optionally a diagnosis relating to blood flow. The displayed image may be enlarged to zoom in on the user's onscreen selection. The device may feature a two-dimensional ultrasound non-phased array of transducer elements.

Automatic blood vessel identification by name

A device images time-wise in parallel using transducer elements of a group. The elements are of a current group and imaging is time-wise sequential by group. The groups may be spatially disposed with respect to each other so as to mutually intermesh element-wise. The imaging may include volumetric imaging. The device may be configured for not collectively using any of the elements to focus, nor to steer, a beam used in the imaging. The device may further be operable to transition between spacing states at least one of which is characterized by a respective minimum, nonzero, degree of intra-group, element-to-element non-adjacency, or may be fixed at a selected spacing state. The transitioning may be automatic, in response to input indicative of blood vessel size and/or depth.

Excitation schemes for low-cost transducer arrays

An automatic, stand-alone, hand-held ultrasound blood-vessel examination device includes a reduced number of transducer elements and presents a simplified user interface, without the need for displaying an image of any of the vessels. The probe, in one embodiment, acquires and examines a volume of interest, searches for a target vessel, tests the vessel for normality of blood flow, and reports the diagnosis, all automatically and without need for user intervention. In another embodiment, the probe finds a body vessel in a volume, and extracts a clinical Doppler parameter, all automatically and without need for user intervention.

Growing community of inventors

Noida, India

Lalit Gupta