International Business Machines Corporation

Intel Corporation

Oracle International Corporation

Optum Technology, Inc.

Atul S Kumar

Russell G Olsen

Sameep Mehta

Raymond Allen Bills

Karie L Kelly

Adam C McCoy

Guy B Mrnustik

Patrick L Walters

Justin David Weisz

James Thomas Rayfield

Vijay Arya

Siddharth Kumar Saraya

Omar Rogelio Alonso

Himanshu Gupta

Avraham Leff

Manish Kesarwani

Pranay Kumar Lohia

Ramnarayanan Ram Muthukaruppan

Yiping Hsiao

Ciaran Avram Fox

Yu Ma

Samiulla Zakir Hussain Shaikh

Katherine Vogt

Gregory F Boland

Rakesh Rameshrao Pimplikar

Rajmohan Chandrahasan

Amit Krishna

Kristina Marie Brimijoin

Rahul Dutta

Love Hasija

Srinivasulu Malepati

Deepa Saini

Santhosh Nen

Systems and methods are configured to enable guided interaction with query assistant software using brainwave data. In various embodiments, a client device presents a query assistant user interface to a monitored end-user that describes a query associated with a plurality of response options and an intended physiological action for each response option. Accordingly, one or more user monitoring data objects associated with the monitored end-user are received that include user brainwave monitoring data objects. These user monitoring data objects are processed using one or more response designation machine learning models to generate response designators based on the user monitoring data objects that includes a physiological response designator describing a selected intended physiological action that is deemed to be related to the user monitoring data objects. Accordingly, a user response is then determined based on the response designators and the user interface may be updated based on the user response.

Guided interaction with a query assistant software using brainwave data

One embodiment provides a method, including: receiving a target unstructured document for determining whether the target unstructured document comprises biased information; identifying an objective of the target unstructured document by extracting, from the target unstructured document, (i) entities and (ii) relationships between the entities; creating a structured knowledge base, wherein the creating comprises (i) creating an entry in the structured knowledge base corresponding to the target unstructured document, (ii) identifying other unstructured documents having a similarity to the target unstructured document, and (iii) generating an entry in the structured knowledge base corresponding to each of the other unstructured documents; applying a bias detection technique on the structured knowledge base; and providing an indication of whether the target unstructured document comprises bias.

Bias detection for unstructured text

Mechanisms are provided for administering health care assessments to a patient. The mechanisms analyze patient information stored in a patient registry and determine a plurality of health care assessments to be administered to the patient based on the patient information and one or more pre-defined health care assessment guidelines specifying conditions for which health care assessments are to be administered to patients and timing for administering the assessments to the patients. The mechanisms generate a sequence of health care assessments, in the plurality of health care assessments, based on the guidelines and the patient information. The sequence comprises an ordering of the health care assessments, and a timing interval between health care assessments, determined based on the guidelines and the patient information. The mechanisms administer at least one health care assessment to the patient in accordance with the determined sequence of health care assessments.

Dynamic selection and sequencing of healthcare assessments for patients

One embodiment provides a method for delaying malicious attacks on machine learning models that a trained using input captured from a plurality of users, including: deploying a model, said model designed to be used with an application, for responding to requests received from users, wherein the model comprises a machine learning model that has been previously trained using a data set; receiving input from one or more users; determining, using a malicious input detection technique, if the received input comprises malicious input; if the received input comprises malicious input, removing the malicious input from the input to be used to retrain the model; retraining the model using received input that is determined to not be malicious input; and providing, using the retrained model, a response to a received user query, the retrained model delaying the effect of malicious input on provided responses by removing malicious input from retraining input.

Detecting and delaying effect of machine learning model attacks

Mechanisms are provided for implementing a personalized health care management system. The mechanisms receive a personalized health care plan for a patient having at least one health goal of the patient, and dynamic patient monitoring data from one or more patient monitoring devices associated with the patient. The mechanisms analyze the dynamic patient monitoring data to determine at least one first communication to output to the patient containing content eliciting conformance of the patient with the personalized health care plan to achieve the at least one health goal. The mechanisms send, to a patient care manager computing device of a patient care manager associated with the patient, a second communication based on results of analyzing the dynamic patient monitoring data. The second communication initiates a new communication session, or continues an existing communication session, between the patient care manager computing device and a patient communication device associated with the patient.

Continuous health care plan coordination between patient and patient care team

A record of comments made in previous artifact versions is displayed in the latest artifact version. Comment and reply chains are maintained in the latest artifact version, linking them with the artifact versions where they were introduced. The comment/reply history of an artifact, is viewable in a single latest version of the artifact.

Comment linkage and display across the review cycle of an artifact

Described is a current sensor comprising: a main resistive branch having a target device coupled to a load at a first node; a replica resistive branch having a replica device which is a replica of the target device, the replica device coupled to a current digital to analog converter (DAC) at a second node; and an analog to digital converter (ADC) coupled directly or indirectly to the first and second nodes, the ADC to generate an output coupled to the current DAC.

Low power current sensor

A method, system, and computer program product for performing searching that generates improved queries, retrieves meaningful and relevant information, and presents the retrieved information to the user in a useful and comprehensive manner is described. The method of searching comprises the steps of: receiving from a user a search query requesting information, retrieving at least one recommendation relating to the search query, generating an expanded query based on the received query, performing a search using the expanded query to retrieve documents, and generating themes relating to the retrieved documents. The at least one recommendation relating to the search query is retrieved from a recommendation database. The recommendation database is generated by performing the steps of: performing data mining using users search query logs, user search patterns, and user profile information to generate a plurality of recommendations relating to search query strings, generating a data structure including the recommendations relating to search query strings, and generating a text index based on information in the data structure.

System and method for search and recommendation based on usage mining

A method and apparatus are provided for enhanced performance caching for path names. A vnode is built for a root directory and a path name is stored in a path cache for the vnode for the root directory. A second vnode is built for a second directory and a path name is stored in a path cache for the second vnode. When a next vnode is built for a next directory, then a path name is stored in a path cache for the next vnode. For a given vnode, the path name includes an entire path name that leads to the vnode. A given vnode is enabled to return its entire path name in one step, without reverse traversal.

Method and apparatus for enhanced performance caching for path names

A method of corrosion susceptibility testing of a magnetic recording head is disclosed. The method includes applying simulated disk corrosion products containing cobalt salts to the recording head. The recording head is then placed in an environmental chamber with elevated temperature and humidity. The resistance of the sensor on the recording head is measured after removal from the chamber and compared with the resistance before placement in the chamber. A significant change in resistance indicates a corrosion failure. This component level testing gives a more accurate indication of the corrosion performance of the recording head when placed in a disk drive.

Method for corrosion susceptibility testing of magnetic heads using simulated disk corrosion products

A method and computer program product are provided for implementing datalink path protection. A datalink designation is created in the computer system and a datalink indicator is turned on. A corresponding datalink path for the datalink designation is stored in a predefined prefix table. The datalink indicator is used for processing a directory rename or remove operation.

Method and computer program product for implementing datalink path protection

Growing community of inventors

Rochester, MN, United States of America

Atul S Kumar