Abstract: A method is provided. The method includes receiving, by an optimization engine, inputs from previous ablation procedures. The method also includes training, by the optimization engine, a machine learning algorithm of the optimization engine to learn scenarios for the previous ablation procedures. The method also includes generating, by the optimization engine, ablation indices for the scenarios.

Abstract: A system, computer-readable storage medium and method for prognosis of head and neck cancer, includes an input for receiving electronic health records (EHR) of a patient, an input for receiving multimodal images of a head and neck area of the patient, a feature extraction module for converting the electronic health records and multimodal images into at least one feature vector, a hybrid machine learning architecture that includes a multi-task logistic regression (MTLR) model and a multi-layer artificial neural network, the hybrid architecture takes as input the at least one feature vector and outputs a final risk score of prognosis for head and neck cancer for the patient.

Abstract: A computing device comprising a processor is configured to perform the techniques of this disclosure. The processor may duplicate a first node of a tree data structure to create a duplicate node, and create an inbound edge of the duplicate node to a parent node of the first node and an outbound edge to at least one child node of the first node. The processor may receive an update to the at least one child node of the first node. In response to determining that the at least one child node has multiple parent nodes, the processor may duplicate the at least one child node to create a duplicate child node, create an outbound edge of the duplicate node to the duplicate child node, delete the outbound edge of the duplicate node to the at least one child node, and perform the update to the at least one child node.

Abstract: We disclose computational models that alleviate the effects of human ascertainment biases in curated pathogenic non-coding variant databases by generating pathogenicity scores for variants occurring in the promoter regions (referred to herein as promoter single nucleotide variants (pSNVs)). We train deep learning networks (referred to herein as pathogenicity classifiers) using a semi-supervised approach to discriminate between a set of labeled benign variants and an unlabeled set of variants that were matched to remove biases.

Abstract: Provided herein is a method, apparatus, and system for identifying a health status of an individual, by using a health challenge of the individual to gather vital sign information to generate a health status of the individual. Method for identifying health status include: generating a non-persistent identifier for a user; receiving an indication of an initiation of a health challenge for the user associated with the non-persistent identifier; receiving, responsive to the health challenge, an indication of at least one vital sign of the user; determining, from the at least one vital sign, a health status of the user; and providing an indication of the health status of the user and the associated non-persistent identifier, where the health status includes a binary indication of health of the user, and where the health status is used to permit or deny a service to the user.

Abstract: Methods and systems to train and use an ensemble of artificial intelligence/machine learning (AI/ML) models to extract information from social determinants of health (SDoH), including training each of multiple dimensionality reduction models to reduce dimensionality of socio-demographic variables associated with a respective one of multiple SDoH categories, training a predictive model to predict a patient behavior for a geographic region (e.g., risk of non-adherence to treatment regimens) based on dimensionally reduced SDoH (alone or in combination with selected socio-demographic variables and/or other data), training a patient classification model to classify patients based on prescription transactions, and/or training a regional similarity model to determine a measure of similarity between geographic regions based on SDoH and/or dimensionally reduced SDoH. Also disclosed are techniques to visually represent outputs of the models on a user-interactive display.

Abstract: Methods, systems, and computer-readable media for generating a personalized action recommendation are provided. The method acquires a request for a service that is associated with a user and the user's condition. The method then identifies one or more features of the user based on stored user information. The method next assigns the user to a segment based on the identified one or more features, generates a set of one or more recommended actions for the user based on the segment, and determines an expected value of each of the one or more recommended actions. The method determines a rank of the one or more recommended actions based on the expected value of each of the one or more recommended actions, and outputs a recommended action with a highest expected value for the user in response to the request for the service.

Abstract: An approach is provided that trains an artificial intelligence (AI) system with a set of eating characteristics corresponding to a human subject. The eating characteristics include one or more eating patterns, health data, and activity data. The trained AI system generates a meal recommendation corresponding to the human subject. The meal recommendation includes a recommended meal time, and one or more food recommendations that are based upon a determined set of caloric needs pertaining to the human subject. The system automatically provides the generated meal recommendation at a time that is based on the recommended meal time using a voice-enabled virtual assistant that is accessible by the AI system.

Abstract: Methods, systems, apparatuses, devices, and computer program products are described. A system may use a rules engine and a reinforcement learning artificial intelligence (AI) model to recommend treatment options for a patient. In some examples, the AI model may be trained for a specific diagnosis. The system may receive patient information including the patient's diagnosis, genomic profile (e.g., partial or full genomic information), and treatment history. The system may input the genomic profile into the rules engine to determine any relevant treatment modifications for the user based on biomarkers in the genomic profile. The system may additionally input the patient information into the AI model to determine a set of treatment option recommendations and corresponding confidence metrics. The system may send the treatment option recommendations (e.g., which in some cases may be modified based on the output of the rules engine) to a user device for display.

Abstract: A system and method for automatically generating a care record are provided, which are suitable for a baby bed cover. The system includes an image capturing device and a processor. The image capturing device captures a real-time image of a baby in the baby bed. The processor is configured to receive the real-time image and detects whether an abnormal event occurs in the real-time image. The processor detects an opening time and a closing time of the bed cover to obtain a start and end time of a care operation. The processor recognizes an abnormal event type according to the abnormal event, selects a care record template according to the abnormal event type, and generates a care record according to an occurrence time of the abnormal event, the abnormal event type, the start and end time of the care operation, and the care record template.

Abstract: Methods for constructing an ADR prediction model for elderly patients, a prediction system, and/or a storage medium are disclosed. The method includes constructing ADR trigger entries for elderly patients; based on the ADR trigger entries for elderly patients, combining evidence-based evaluation results to construct a dataset of risk factors related to ADRs in elderly patients; completing the annotation of risk factors and ADR discrimination for the dataset related to ADRs in elderly patients; using the annotated dataset information to train a machine learning model, so as to obtain an ADR risk prediction model for elderly patients. A prediction system utilizing the ADR prediction model for elderly patients is constructed.

Abstract: A machine learning educational content recommender system that provides educational clinical content is disclosed. An educational content recommender system may receive input about a specific case and about a specific clinician. Using this data, the educational content recommender system may recommend educational clinical content for the specific clinician to use for a specific case. Feedback from the specific clinician may be used to train the educational content recommender system. This feedback may include either or both explicit and implicit feedback items.

Abstract: An apparatus and method for determining a code as a function of subject data. The apparatus includes at least a processor and a memory communicatively connected to the at least a processor. The memory instructs the processor to collect subject data, receive user input associated with the subject data, determine, using a first encoder, portions of the user input, wherein the portions comprise tokens and noise, identify, using a second encoder, the tokens of the portions, assign, using the second encoder, a code to the tokens, and identifying the code as a function of the subject data, and display the code.

Abstract: The invention relates to a method for controlling at least one actuator (4) of an orthopedic device (2) with an electronic control device (E), which is coupled to the actuator (4) and at least one sensor (8) and which has an electronic processor (?C) for processing sensor data (s), wherein at least one state machine (SM) in which states (z) of the orthopedic device (2) and state transitions of the actuator (4) are determined is stored in the control device (E), wherein a classifier (K) in which sensor data (s) and/or states (z) are automatically classified within the scope of a classification method is stored in the control device (E), wherein the state machine (SM) and the classification method are used in combination and, on the basis of the classification and the states (z), a decision is made about the manner of activating or deactivating the actuator (4) as a control signal.

Abstract: A deep neural network pre-training method for classifying electrocardiogram (ECG) data and a device for the same are disclosed. A method for training an ECG feature extraction model may include receiving a ECG signal, extracting one or more first features related to the ECG signal by inputting the ECG signal to a rule-based feature extractor or a neural network model, extracting at least one second feature corresponding to the at least one first feature by inputting the ECG signal to an encoder, and pre-training the ECG feature extraction model by inputting the at least one second feature into at least one of a regression function and a classification function to calculate at least one output value. The pre-training of the ECG feature extraction model may include training the encoder to minimize a loss function that is determined based on the at least one output value and the at least one first feature.

Abstract: The disclosure belongs to the field of genetic testing and biomedicine, relating to a method for constructing a prognostic model of hepatoma and an application thereof, comprising 1) obtaining and identifying fibroblasts with high FAP expression; 2) obtaining and identifying TAMs; 3) analyzing co-localization between fibroblasts with high FAP expression obtained and the TAMs obtained previously; 4) communicating and analyzing the fibroblasts with high FAP expression after the localization in the Step 3) with TAMs to obtain CCC ligand-receptor genes; 5) screening the CCC ligand-receptor genes obtained previously based on machine learning to obtain key CCC ligand-receptor genes; and 6) constructing a prognostic model of hepatoma according to the key CCC ligand-receptor genes obtained in the Step 5). The present disclosure provides a method for constructing a prognostic model of hepatoma that can be applied to auxiliary judgment of the prognosis of hepatoma patients and an application thereof.

Abstract: Methods for applying machine learning algorithms to nucleic acid sequencing-based diagnostics tests for detection of copy number variation and other genomic abnormalities are described.

Abstract: A model-assisted system for processing data to extract a patient event date may include a processor. The processor may be programmed to access a database storing a medical record associated with a patient, the medical record comprising unstructured data; analyze the unstructured data to identify a plurality of dates represented in at least one document included in the medical record; identify a plurality of snippets of information included in the at least one document, each snippet of the plurality of snippets being associated with a date of the plurality of dates; inputting the plurality of snippets into a machine learning model, the machine learning model having been trained to determine associations between dates and patient events based on a training set of snippet data; and determine whether each date of the plurality of dates is associated with a patient event based on an output of the machine learning model.

Abstract: A system and method for patient diagnosis and implementation of orthodontic apparatuses and treatments. The patient can be located remotely from a remote treatment system, thereby allowing for automatic diagnosis and treatment of the patient without interaction with a medical professional or associated staff. A neural network of an artificial intelligence engine (AI engine) of the remote treatment system can utilize images and other data captured by a patient, as well as patient responses to questionnaires, to derive information regarding the patient's symptoms, and the severity of those symptoms. The AI engine can further be adapted, including via use of historical information regarding a plurality of patients and information from the patient, to obtain a diagnosis and treatment recommendation. The AI engine can further be configured to determine, based on analysis of patient provided images, an appropriately sized orthopedic apparatus, which can be sent to the patient for implementation.

Abstract: The present disclosure discloses a cross-session brainprint recognition method based on a tensorized spatial-frequency attention network (TSFAN) with domain adaptation. For most of existing multi-source domain adaptation methods, domain gaps between multiple source domains and target domains are individually bridged, but a relationship between domain-invariant features in distribution alignment is ignored. The present disclosure assists performance of a target domain by modeling an important relationship of the domain-invariant features without being affected by a distribution difference between source domains. A new TSFAN is used to combine pairwise source and target and an appropriate common spatial-frequency feature across source domains. Considering of a dimension, the TSFAN is further approximated as a low-rank Tucker format, to enable the TSFAN to adapt to scale linearly in a quantity of domains, and apply the TSFAN to a case of any quantity of sessions.