Abstract: The invention relates to a system (200) for detecting one or more abnormalities in an x-ray image using an image classifier and one or more feature extractors. An abnormality is indicative of a pathology, a disease or a clinical finding present in the x-ray image. The feature extractors extract respective image quality features from the x-ray image indicative of a suitability of the x-ray image for detection of the abnormalities. The one or more feature extractors are applied to the x-ray image to determine the respective image quality features for the x-ray image. The image classifier is applied to the x-ray image to determine the classification scores for the one or more abnormalities. The image classifier has been trained to use the determined image quality features to determine said classification scores. A classification result is output based on the determined classification scores.

Abstract: Some embodiments are directed to a federated learning system. A federated model is trained on respective local training datasets of respective multiple edge devices. In an iteration, an edge device obtains a current federated model, determines a model update for the current federated model based on the local training dataset, and sends out the model update. The edge device determines the model update by applying the current federated model to a training input to obtain at least a model output for the training input; if the model output does not match a training output corresponding to the training input, include the training input in a subset of filtered training inputs to be used in the iteration; and determining the model update by training the current federated model on only the subset of filtered training inputs.

Abstract: Some embodiments are directed to training a model, e.g., a medical model. The training uses multiple model updates received from multiple client systems. At least some of the multiple client train on training sets that indicate values for different features. The model updates are aggregated in an aggregated model, for which feature weights are obtained. The feature weights provide information on the relative importance of the multiple features for the aggregated model's output.

Abstract: Some embodiments are directed to training method a classifier. The classifier receives sensor data as input and produces a label as output. A quality estimator is applied to meta-data of a training sample, obtaining a quality estimation of a ground-truth label of the training sample. The classifier may be trained on the training sample taking into account the quality of the ground-truth label.

Abstract: Techniques are described herein for prioritizing tasks of domain experts for machine learning model training. In various embodiments, prediction candidates, such as physiological abnormalities of interest, may be assigned priorities for use in labeling training examples that are to be used to train machine learning model(s). The plurality of training examples may be analyzed to generate a corresponding plurality of preliminary predictions. Each preliminary prediction may include a probability that the respective training example of the plurality of training examples includes at least one of the plurality of prediction candidates. A computing device operated by a domain expert may provide output that presents at least some of the plurality of training examples to the domain expert in a manner selected based on the priorities assigned to the plurality of prediction candidates and the plurality of preliminary predictions.

Abstract: A method provides fully autonomous patient matching by entities of a federated healthcare system. The method includes receiving an electronically formatted query for a patient from an autonomous healthcare entity in a federation of healthcare entities. The query includes at least an identifier of the entity, a unique patient identifier of the patient generated by the entity, and demographics of the patient. The method further includes searching a federated master patient index stored in master patient index storage for an entry likely to correspond to the patient. The method further includes identifying an entry for the patient. The method further includes updating the identified entry to include at least the identifier of the entity, the unique patient identifier of the patient generated by the entity and the demographics.

Abstract: A system for matching a pair of patient records as belonging to the same individual or patient is disclosed. A set of matching rules (4) represents persistent characteristics of clinical properties of patients. A clinical property extractor (5) is used for extracting at least one first clinical property from clinical information contained in the first patient record (1) and at least one second clinical property from clinical information contained in the second patient record (2). A match detector (3) is used for detecting whether the first patient record (1) and the second patient record (2) relate to the same patient, based on the first clinical property, the second clinical property, and the set of matching rules 4. A demographic property extractor (6) is used for extracting demographic data.

Abstract: The execution of a clinical trial is supported in the following manner. Patient model data is provided for modeling a patient-level workflow in the clinical trial in accordance with a trial protocol. Clinical data of patients participating in the clinical trial is accessed. Using a workflow engine, an instance of the patient-level workflow is instantiated for each of the patients to obtain a plurality of patient-level workflow instantiations. Each of the patient-level workflow instantiations is executed independently based on the clinical data of each respective one of the patients. The patient model data may include or be linked to a rule associated with a step of the patient-level workflow. The rule may be executed when the workflow engine reaches the step in one of the patient-level workflow instantiations. Accordingly, a virtual representation of the clinical trial on a patient-level is obtained which represents the actual state of the clinical trial yet which is computer-accessible.

Abstract: A system for linking corresponding patient information records is disclosed. A plurality of entities (1,1a) have respective patient databases comprising patient information records (3, 3a). Each entity (1, 1a) has associated therewith a patient identification algorithm (4,4a) for matching corresponding patient information records (3, 3a) of the same patient at different entities (1,1a). A linking subsystem (6) maintains a set of links (7) of a first entity (1) of the plurality of entities (1,1a). The linking subsystem (6) is arranged for linking patient information records (3) of the first entity (1) with corresponding patient information records (3a) of the other entities (1a). A link (ID, RID, RLoc) is established when a given patient information record (3) of the first entity (1) matches a corresponding patient information record (3a) of another entity (1a) based on the patient identification algorithm (4) of the first entity (1).

Abstract: An assertion acceptance value matrix (300) indicates the reliability of assertions, particularly assertions or decisions whether records match or do not match, made by other medical institutions in a federation of medical institutions with different patient record systems and some common patients. Records from different institutions with a high likelihood of matching or not matching are automatically matched or not matched. Those that are ambiguous are manually reviewed. The assertion acceptance value matrix is used to reduce or expedite the manual review.

Abstract: An assertion acceptance value matrix (300) indicates the reliability of assertions, particularly assertions or decisions whether records match or do not match, made by other medical institutions in a federation of medical institutions with different patient record systems and some common patients. Records from different institutions with a high likelihood of matching or not matching are automatically matched or not matched. Those that are ambiguous are manually reviewed. The assertion acceptance value matrix is used to reduce or expedite the manual review.

Abstract: Embodiments of the present invention relate to clinical knowledge driven healthcare scheduling. There is disclosed a method of healthcare scheduling, comprising receiving a request to generate an appointment for a target user, the request at least containing identification of the target user; obtaining clinical knowledge associated with the target user by accessing a knowledge base, the clinical knowledge at least indicating time constraints and at least one required resource for the appointment; generating a group of prioritized candidate appointments by processing the request at least partially based on the obtained clinical knowledge, each of the candidate appointments being represented at least by appointment time and a priority; and selecting at least one recommended appointment from the group of prioritized candidate appointments based on availability of the at least one required resource. Corresponding apparatus, system and computer program product are also disclosed.

Abstract: A system and method for managing an executable medical treatment plan. The system and method perform the steps of analyzing one of treatment patterns and prior treatment patterns in view of patient information to generate treatment plan recommendations using a processor, receiving a user input regarding the treatment plan recommendations via a user interface and generating a current defined treatment plan based on the treatment plan recommendations and the user input regarding the treatment plan recommendations using the processor.

Abstract: A text analysis system is described. A natural language input unit (1) is arranged for enabling a user to input a free text (10) in a natural language. A natural language processing unit (2) is arranged for processing at least a portion of the free text (10) while it is being inputted, to obtain an explicit representation (11) of semantics represented by the free text. An explicit information input unit (3) is arranged for enabling the user to input explicit information (12) relating to the explicit representation (11) of semantics. The system comprises a visualization unit (4) for visualizing at least part of the explicit representation (11) to the user while the user is still inputting the free text (10). A user interface (5) is arranged for providing a user with simultaneous access to both the natural language input unit (1) and the explicit information input unit (3).

Abstract: The execution of a clinical trial is supported in the following manner. Patient model data is provided for modeling a patient-level workflow in the clinical trial in accordance with a trial protocol. Clinical data of patients participating in the clinical trial is accessed. Using a workflow engine, an instance of the patient-level workflow is instantiated for each of the patients to obtain a plurality of patient-level workflow instantiations. Each of the patient-level workflow instantiations is executed independently based on the clinical data of each respective one of the patients. The patient model data may comprise or be linked to a rule associated with a step of the patient-level workflow. The rule may be executed when the workflow engine reaches the step in one of the patient-level workflow instantiations. Accordingly, a virtual representation of the clinical trial on a patient-level is obtained which represents the actual state of the clinical trial yet which is computer-accessible.

Abstract: A system is provided for visually rendering longitudinal patient data. The system makes use of a screen template defining at least one visual element for being rendered on a display, with an appearance of the visual element being defined by a visualization parameter. The screen template associates the visual element with a clinical decision rule. During operation, longitudinal patient data is accessed and analyzed, namely by evaluating the clinical decision rule using a time portion of the longitudinal patient data as input to obtain a rule output, and by determining the visualization parameter of the visual element based on the rule output. The visual element is then rendered on the display in accordance with the visualization parameter, namely as part of the screen template. This yields a rendering which enables a user to more efficiently process the large amount of data provided by longitudinal patient data.

Abstract: A system and method for providing quality review of healthcare reporting and supporting quality improvement. The system and method performing the steps of multiplying a select number of cases of a plurality of cases to be analyzed using a processor, assigning the plurality of cases and multiplied cases to members of a work pool using the processor, retrieving case reports created by members of the work pool for the multiplied cases and analyzing case reports for one of the multiplied cases in a first review using the processor.

Abstract: When scheduling a clinical procedure such as a surgery, a clinician enters a deadline (i.e., a date of surgery or other procedure) and desired clinical procedures to be performed in advance thereof (e.g., lab tests, pathology tests, medical record review, etc.). A feasibility algorithm is executed and analyzes median pathway step completion times for a clinical pathway or workflow that is executed to satisfy the clinician's work order. A probability of completion of the clinical pathway by the specified deadline is also provided to the clinician. If the probability of completion is low (e.g., if the requested test results are unlikely to be ready by the scheduled surgery), then the clinician is prompted to adjust one or more of the surgery date and the clinical pathway steps to be completed thereby.

Abstract: A text analysis system is described. A natural language input unit (1) is arranged for enabling a user to input a free text (10) in a natural language. A natural language processing unit (2) is arranged for processing at least a portion of the free text (10) while it is being inputted, to obtain an explicit representation (11) of semantics represented by the free text. An explicit information input unit (3) is arranged for enabling the user to input explicit information (12) relating to the explicit representation (11) of semantics. The system comprises a visualization unit (4) for visualizing at least part of the explicit representation (11) to the user while the user is still inputting the free text (10).

Abstract: A clinical decision support system comprises a clinical guideline (1) extended with expert recommendations. At least one of the nodes (3) is associated with a pair (4) of a clinical question (5) and a corresponding clinical answer (6), the pair forming an extension to the clinical guideline (1) for the purpose of the clinical decision support. A node unit (7) is arranged for determining a node (3) of the plurality of nodes, based on a condition of a specific patient and the set of clinical preconditions of the node. A presenting unit (8) is arranged for presenting at least a part of the pair (4) of the question (5) and/or the corresponding clinical answer (6) associated with the relevant node (3). A matching unit (10) is arranged for matching a question against a collection of existing questions previously answered, in dependence on the relevant node (3).