International Business Machines Corporation

Blue Spark Innovations, LLC

Adam Perer

Kenney Ng

Peter Bak

Avi Yaeli

David Haim Gotz

Josua Krause

Zhiyuan Zhang

Fei Wang

Jianying Hu

Yajuan Wang

Matt Ream

John Gannon

Swapnil Chhabra

Joern Jaskolowski

Ruth Phillips

Frank Feddrix

Shay Segal

Sashi Kolli

Continuous temperature monitors can be used to collect frequent temperature data over large periods of time. A machine learning system trained with collected temperature data can be used to predict future temperature data for patients. Such predictions can be clinically beneficial for disease states in which fever can be fatal, particularly for those fevers having fast onset. For example, patients undergoing chimeric antigen receptor T-cell (CAR-T) therapy may suffer fevers caused by cytokine release syndrome (CRS). Accordingly, a continuous temperature monitor is used to collect temperature data from CAR-T patients with high risk of CRS (or like patients with high fever risk), and the collected temperature data processed by a machine learning system to predict the patient's future temperature profiles. Results of analysis of the predicted temperature profiles may then be provided to the patient and/or clinician.

System and method of using body temperature logging patch

Embodiments are directed to methodologies, systems and computer program products for generating, for each of a plurality of risk factors in a patient database containing information of a plurality of patients, an index of input values for the risk factor. For each patient P of the plurality of patients, a series of local impact scores is computed for the patient. Computing the series of local impact scores for the patient includes calculating a risk score for the patient with respect to each of the indexed input values for each of the plurality of risk factors. For at least one of the plurality of patients, at least some of the plurality of risk factors are ranked based at least partly on the computed local impact scores for each of the at least some risk factors, and an indication of the ranked risk factors for the at least one patient is provided.

Identifying and ranking risk factors using trained predictive models

Aspects of the present invention include a method, system and computer program product. The method includes identifying, by a processor, a set of global risk factors for a target event using training patients, and providing, by the processor, a disease progression timeline with defined time stamps by aligning longitudinal data of the training patients based on the defined time stamp of risk targets. The method also includes positioning, by the processor, a target patient at one of the defined time stamps on the disease progression timeline, and identifying, by the processor, at least one of the training patients similar to the target patient with the same one of the defined time stamps on the disease progression timeline. The method further includes calculating, by the processor, a time-varying predictive pattern of at least a portion of the global set of risk factors for the target patient along the disease progression timeline.

Identifying personalized time-varying predictive patterns of risk factors

A pathway for each entity in a cohort may be extracted from raw data comprising a plurality of events. A common graph structure may be created based the extracted pathways. For each cohort, a relative signature trait may be computed for each edge based on a number of appearances of the edge in the extracted pathways that are associated with the cohort. A relationship expression may be obtained. The relationship expression may be with respect to at least two cohorts for which relative signature traits were computed. The common graph structure may be processed based on the relationship expression to determine an output label for each edge. A visualization of the common graph structure may be displayed.

Relative signature traits of cohorts

There is provided a method of creating a cohort clinical pathway graph based on knowledge-driven manual user input and automated data-driven mining comprising: receiving via a graphical user interface (GUI), manual selections including: knowledge-driven variable(s) denoting clinically significant values representing elements of a clinical decision making process, and an anchoring location of each knowledge-driven node denoting a respective knowledge-drive variable within a directed acyclic graph (DAG), computing individual clinical pathways for each of the sampled population of patients by automatically computing data-driven nodes denoting the data-driven discovery of event types relative to the manual selections, and aggregating the individual clinical pathways to compute a cohort clinical pathway DAG, wherein the cohort clinical pathway DAG includes nodes comprising the knowledge-driven nodes, the data-driven nodes, and links connecting the nodes, each link denoting an automatically discovered sequence between two respective nodes, and presenting the cohort clinical pathway DAG within the GUI.

Systems and methods for constructing clinical pathways within a GUI

Methods and apparatus are provided for iterative refinement of cohorts using visual exploration and data analytics. A cohort comprised of multiple data objects is defined by obtaining an initial cohort seeding; visualizing the initial cohort using a selected view to present a current cohort; reducing the current cohort using one or more visual filters; visualizing the current cohort using a selected view; expanding the current cohort using one or more selected analytics; and determining whether the current cohort should be further modified using one or more of additional reductions and additional expansions. Cohorts can be passed between views and analytics via drag-and-drop interactions as an analysis unfolds.

Iterative refinement of cohorts using visual exploration and data analytics

Growing community of inventors

Pittsburgh, PA, United States of America

Adam Perer