Koninklijke Philips Corporation N.V.

Taiyao Wang

Cornelis Van Zon

Ali Akbar Ahmad Samadani

Francesco Vicario

Ikaro Garcia Araujo Da Silva

A CDS system including a controller extracting EHRs of an IHG and selecting SMs to form a target patient database, capturing a time interval between each subsequent record for each of the measurements defined by the SMs, computing hospital level features including statistics on the distribution of the captured time intervals for each of the measurements, clustering the plurality of hospitals of the IHG in accordance with the computed hospital level features, determining whether an other hospital is within a threshold distance to a closest one of the plurality of clusters; placing the other hospital in the closest cluster when it is determined that the other hospital is within the threshold distance, and placing the other hospital in a new cluster when it is determined that the other hospital is not within the threshold distance.

Incremental clustering and hierarchy formation system for clinical decision support (CDS) system development and method of operation thereof

Described herein are methods, devices and systems for predicting ventilator associated events (VAEs). In an example, data of a patient including one or more features useful in predicting VAEs is accessed and the features are extracted. The extracted features are evaluated automatically to determine a probability that a VAE will occur. If it is determined, based on the evaluating, that a VAE has a probability of occurring that exceeds a threshold, an indication that the VAE is predicted to occur is provided. Other examples are disclosed and claimed.

Prediction of ventilator-associated events (VAEs)

Described herein are methods, devices and systems for predicting ventilator associated pneumonia (VAP). In an example, data of a patient including one or more features useful in predicting VAPs are accessed and the features are extracted. A model is activated based on a time parameter associated with mechanical ventilation or length of stay in ICU. The extracted features are evaluated automatically by the activated model to determine a probability that a VAP will occur. If it is determined, based on the evaluating, that a VAP has a probability of occurring that exceeds a threshold, an indication that the VAP is predicted to occur may be provided. Other examples are disclosed and claimed.

Growing community of inventors

Cambridge, MA, United States of America

Taiyao Wang