Abstract: One or more systems, devices, computer program products and/or computer-implemented methods of use provided herein relate to incorporating explainability constraints into machine learning model training. A computer-implemented system can comprise a memory that can store computer executable components. The computer-implemented system can further comprise a processor that can execute the computer executable components stored in the memory, wherein the computer executable components can comprise a training component that can train a machine learning model using an objective function that can be modified to incorporate an explainability metric for the machine learning model in addition to a model performance metric of the machine learning model.

Abstract: Providing a machine-generated explanation of an outcome of a process execution engine includes inputting into a complex event processing (CEP) engine at least one predetermined event pattern specification and a sequence of events, the sequence of events generated, at least in part, by a process execution engine executing a predetermined process in real time. Using the CEP engine, an enriched event log is generated based on the input. The enriched event log includes the sequence of events and additionally includes one or more situational events corresponding to one or more of the sequence of events. The one or more situational events are derived by the CEP engine based on the at least one predetermined event pattern specification. A hypothesis-oriented, situationally aware explanation for the outcome of the predetermined process is determined using an artificial intelligence (AI) explainability framework with input of the enriched event log.

Abstract: Providing a machine-generated explanation of an outcome of a process execution engine includes inputting into a complex event processing (CEP) engine at least one predetermined event pattern specification and a sequence of events, the sequence of events generated, at least in part, by a process execution engine executing a predetermined process in real time. Using the CEP engine, an enriched event log is generated based on the input. The enriched event log includes the sequence of events and additionally includes one or more situational events corresponding to one or more of the sequence of events. The one or more situational events are derived by the CEP engine based on the at least one predetermined event pattern specification. A hypothesis-oriented, situationally aware explanation for the outcome of the predetermined process is determined using an artificial intelligence (AI) explainability framework with input of the enriched event log.

Abstract: A system for analyzing image data can include a processor to detect a set of medical images and partition each medical image into a plurality of non-overlapping or overlapping patches. The processor can also generate an ordered set of patch clusters derived from the medical images, wherein the patch clusters are sorted based on a type of tissue identified in each of the medical images. Additionally, the processor can generate a digital representation of a percentage of the type of tissue calculated for each of the medical images and transmit an alert in response to detecting that the percentage of the type of tissue for an image exceeds a predetermined threshold.

Abstract: A method comprising receiving a digital mammogram of a human breast; automatically extracting a contour boundary of said breast in said digital mammogram; automatically calculating a convex hull for said contour boundary; automatically comparing said contour boundary to said convex hull, to detect a plurality of gap segments, wherein each of said plurality of gap segments has an associated inflection point; and automatically determining a nipple location along said contour boundary as between a pair of said inflection points.

Abstract: A system for analyzing image data can include a processor to detect a set of medical images and partition each medical image into a plurality of non-overlapping or overlapping patches. The processor can also generate an ordered set of patch clusters derived from the medical images, wherein the patch clusters are sorted based on a type of tissue identified in each of the medical images. Additionally, the processor can generate a digital representation of a percentage of the type of tissue calculated for each of the medical images and transmit an alert in response to detecting that the percentage of the type of tissue for an image exceeds a predetermined threshold.

Abstract: A method comprising receiving a digital mammogram of a human breast; automatically extracting a contour boundary of said breast in said digital mammogram; automatically calculating a convex hull for said contour boundary; automatically comparing said contour boundary to said convex hull, to detect a plurality of gap segments, wherein each of said plurality of gap segments has an associated inflection point; and automatically determining a nipple location along said contour boundary as between a pair of said inflection points.

Abstract: A method for detecting an indication of likelihood of malignancy, comprising: receiving a sequence of anatomical images of a breast of a target individual acquired over a time interval during which contrast is administered, analyzing the sequence of anatomical images to identify: a baseline pre-contrast image denoting lack of contrast, a peak contrast image denoting a peak contrast enhancement, an initial uptake image denoting initial contrast enhancement, and a delayed response image denoting final contrast enhancement, creating a multi-channel image representation comprising: intensity channel including the peak contrast enhanced image, contrast-update channel including the computed difference between the peak contrast enhanced image and the pre-contrast image, and contrast-washout channel including the computed difference between the initial uptake image and the delayed response image, and computing by a trained deep convolutional neural network, a classification category indicative of likelihood of malign

Abstract: Methods and systems for generating an electronic structured report associated with a displayed electronic medical image. One method includes receiving an annotation for the electronic medical image, automatically determining, with an electronic processor, an anatomical location within the electronic medical image associated with the annotation, and automatically determining, with the electronic processor, a location within the electronic structured report associated with the anatomical location based on a predetermined mapping. The method also includes automatically populating the location of the electronic structured report based on the annotation.

Abstract: A method for detecting an indication of likelihood of malignancy, comprising: receiving a sequence of anatomical images of a breast of a target individual acquired over a time interval during which contrast is administered, analyzing the sequence of anatomical images to identify: a baseline pre-contrast image denoting lack of contrast, a peak contrast image denoting a peak contrast enhancement, an initial uptake image denoting initial contrast enhancement, and a delayed response image denoting final contrast enhancement, creating a multi-channel image representation comprising: intensity channel including the peak contrast enhanced image, contrast-update channel including the computed difference between the peak contrast enhanced image and the pre-contrast image, and contrast-washout channel including the computed difference between the initial uptake image and the delayed response image, and computing by a trained deep convolutional neural network, a classification category indicative of likelihood of malign

Abstract: Methods and systems for generating an electronic structured report associated with a displayed electronic medical image. One method includes receiving an annotation for the electronic medical image, automatically determining, with an electronic processor, an anatomical location within the electronic medical image associated with the annotation, and automatically determining, with the electronic processor, a location within the electronic structured report associated with the anatomical location based on a predetermined mapping. The method also includes automatically populating the location of the electronic structured report based on the annotation.

Abstract: Methods and systems for generating an electronic structured report associated with a displayed electronic medical image. One method includes receiving an annotation for the electronic medical image, automatically determining, with an electronic processor, an anatomical location within the electronic medical image associated with the annotation, and automatically determining, with the electronic processor, a location within the electronic structured report associated with the anatomical location based on a predetermined mapping. The method also includes automatically populating the location of the electronic structured report based on the annotation.

Abstract: Methods and systems for reviewing medical images. One system includes an electronic processor configured to receive an annotation for a displayed electronic medical image, wherein the annotation includes a label of a lesion represented within the medical image. The electronic processor is further configured to automatically determine whether the lesion is labeled one or more times in other medical images acquired during an imaging exam associated with the displayed electronic medical image, and identify a stored rule based on the annotation, wherein the stored rule specifies whether the lesion should be labeled in the other medical images. The electronic processor is also configured to execute the stored rule based on whether the lesion is labeled one or more times in the other medical images, and automatically initiate at least one automatic action based on executing the stored rule.

Abstract: Methods and systems for generating an electronic structured report associated with a displayed electronic medical image. One method includes receiving an annotation for the electronic medical image, automatically determining, with an electronic processor, an anatomical location within the electronic medical image associated with the annotation, and automatically determining, with the electronic processor, a location within the electronic structured report associated with the anatomical location based on a predetermined mapping. The method also includes automatically populating the location of the electronic structured report based on the annotation.

Abstract: ECG apparatus comprises an ECG input for obtaining ECG signals; a high frequency analyzer configured for obtaining high frequency QRS components from a QRS complex within said ECG signals and identifying therein at least one reduced amplitude zone—RAZ—present within a given QRS complex; and a RAZ quantifier configured for obtaining a quantification of said at least one RAZ region within said QRS complex.

Abstract: ECG apparatus comprises an ECG input for obtaining ECG signals; a high frequency analyzer configured for obtaining high frequency QRS components from a QRS complex within said ECG signals and identifying therein at least one reduced amplitude zone—RAZ—present within a given QRS complex; and a RAZ quantifier configured for obtaining a quantification of said at least one RAZ region within said QRS complex.