Abstract: Embodiments are disclosed for health assessment and diagnosis implemented in an artificial intelligence (AI) system. The AI system takes as input information from a multitude of sensors measuring different biomarkers in a continuous or intermittent fashion. The proposed techniques disclosed herein address the unique challenges encountered in implementing such an AI system.

Abstract: Embodiments are disclosed for health assessment and diagnosis implemented in an artificial intelligence (AI) system. In an embodiment, a method comprises: capturing, using one or more sensors of a device, signals including information about a user's symptoms; using one or more processors of the device to: collect other data correlative of symptoms experienced by the user; and implement pre-trained data driven methods to: determine one or more symptoms of the user; determine a disease or disease state of the user based on the determined one or more symptoms; determine a medication effectiveness in suppressing at least one determined symptom or improving the determined disease state of the user; and present, using an output device, one or more evidence for at least one of the determined symptoms, the disease, disease state, or an indication of the medication effectiveness for the user.

Abstract: Embodiments are disclosed for health assessment and diagnosis implemented in an artificial intelligence (AI) system. In an embodiment, a method comprises: obtaining, using one or more processors of a device, a speech sample from a user uttering a first sentence; processing the speech sample through a neural network to predict a first set of one or more disease-related symptoms of the user; and generating, using the one or more processors, a second sentence to predict a second set of one or more disease-related symptoms or confirm the first set of disease-related symptoms.

Abstract: Embodiments are disclosed for health assessment and diagnosis implemented in an artificial intelligence (AI) system. In an embodiment, a method comprises: obtaining one or more interpretations from an interpretable artificial intelligence (AI); sorting the AI interpretations based on one or more impact values; selecting one or more augmentations based on the sorted one or more AI interpretations; and applying the selected augmentations to a training dataset for a machine learning model. In another embodiment, a method comprises: obtaining one or more predicted symptoms from a symptom classifier for a plurality of users; feeding the one or more predicted symptoms into a speaker classifier; and predicting an owner of a symptom based on output of the speaker classifier.

Abstract: Embodiments are disclosed for health assessment and diagnosis implemented in an artificial intelligence (AI) system. In an embodiment, a method comprises: feeding a first set of input features to the AI model; obtaining a first set of raw output predictions from the model; determining a first set of impact scores for the input features fed into the model; training a neural network with the first set of impact scores as input to the network and pre-determined sentences describing the model's behavior as output; feeding a second set of input features to the AI model; obtaining a second set of raw output predictions from the model; determining a second set of impact scores based on the second set of output predictions; feeding the second set of impact scores to the neural network; and generating a sentence describing the AI model's behavior on the second set of input features.

Abstract: Embodiments are disclosed for health assessment and diagnosis implemented in an artificial intelligence (AI) system. In an embodiment, a method comprises: generating, using one or more processors, profiles for a plurality of users, each profile including information collected from monitoring users over time; categorizing, using the one or more processors, a set of users into one or more predetermined categories based on the generated profiles; and recommending, using the one or more processors, a set of actions for each category of users.

Abstract: Embodiments are disclosed for health assessment and diagnosis implemented in an artificial intelligence (AI) system. In an embodiment, a method comprises: feeding a first set of input features to the AI model; obtaining a first set of raw output predictions from the model; determining a first set of impact scores for the input features fed into the model; training a neural network with the first set of impact scores as input to the network and pre-determined sentences describing the model's behavior as output; feeding a second set of input features to the AI model; obtaining a second set of raw output predictions from the model; determining a second set of impact scores based on the second set of output predictions; feeding the second set of impact scores to the neural network; and generating a sentence describing the AI model's behavior on the second set of input features.

Abstract: Embodiments are disclosed for health assessment and diagnosis implemented in an artificial intelligence (AI) system. In an embodiment, a method comprises: capturing, using one or more sensors of a device, signals including information about a user's symptoms; using one or more processors of the device to: collect other data correlative of symptoms experienced by the user; and implement pre-trained data driven methods to: determine one or more symptoms of the user; determine a disease or disease state of the user based on the determined one or more symptoms; determine a medication effectiveness in suppressing at least one determined symptom or improving the determined disease state of the user; and present, using an output device, one or more evidence for at least one of the determined symptoms, the disease, disease state, or an indication of the medication effectiveness for the user.

Abstract: Embodiments are disclosed for health assessment and diagnosis implemented in an artificial intelligence (AI) system. The AI system takes as input information from a multitude of sensors measuring different biomarkers in a continuous or intermittent fashion. The proposed techniques disclosed herein address the unique challenges encountered in implementing such an AI system.

Abstract: Embodiments are disclosed for health assessment and diagnosis implemented in an artificial intelligence (AI) system. In an embodiment, a method comprises: obtaining, using one or more processors of a device, a speech sample from a user uttering a first sentence; processing the speech sample through a neural network to predict a first set of one or more disease-related symptoms of the user; and generating, using the one or more processors, a second sentence to predict a second set of one or more disease-related symptoms or confirm the first set of disease-related symptoms.

Abstract: Systems and methods for performing personalized operations of a mobile device are provided. An example method includes determining a sensor signature has been received. The sensor signature may be a combination of acoustic and non-acoustic inputs. A classification or recognition or matching score for the received sensor signature may be calculated. A context associated with one or more of the operations of the mobile device may be determined. A confidence level or weight may be determined for the acoustic and non-acoustic inputs based on the context. Based on the confidence level or weight, a determination may be made whether to perform the one or more operations. The determination whether to perform the one or more operations may be further based on a combination of the confidence level or weight and the score. Determining the context can include identifying the type and level of noise associated with the mobile device's environment.

Abstract: Systems and methods for voice authentication are provided. An example method starts with dynamically generating authentication information and a prompt associated therefor. The authentication information is generated to emphasize differences between an authorized user and others. The authentication information may be generated based on at least one of a user profile, an acoustic environment in which a mobile device is located, and a history of interactions with the mobile device. The authentication information may be an answer to a question which the authorized user would uniquely be able to provide, or it may be a distinctive password. The prompt, for the authentication information, is provided to a user attempting to use the mobile device. The method proceeds with capturing an acoustic sound of a speech of the user and detecting the authentication information in the speech. Based on the detection, a confidence score is determined.

Abstract: Systems and methods for keyword detection using keyword repetitions are provided. An example method includes receiving an acoustic signal representing at least one captured sound. Using a keyword model, a first confidence score for the first acoustic signal may be acquired. The method also includes determining the first confidence score is less than a detection threshold within a first value. In response, lowering the threshold by a second value for a pre-determined time interval. The method also includes receiving a second acoustic signal captured during the pre-determined time interval and acquiring a second confidence score for the second acoustic signal. The method also includes determining the second confidence score equals or exceeds the lowered threshold, and then confirming keyword detection. The threshold may be restored after the pre-determined time interval. The keyword model may be temporarily replaced by a tuned keyword model to facilitate keyword detection in low SNR conditions.

Abstract: Systems and methods for multi-sourced noise suppression are provided. An example system may receive streams of audio data including a voice signal and noise, the voice signal including a spoken word. The streams of audio data are provided by distributed audio devices. The system can assign weights to the audio streams based at least partially on quality of the audio streams. The weights of audio streams can be determined based on signal-to-noise ratios (SNRs). The system may further process, based on the weights, the audio stream to generate cleaned speech. Each audio device comprises microphone(s) and can be associated with the Internet of Things (IoT), such that the audio devices are Internet of Things devices. The processing can include noise suppression and reduction and echo cancellation. The cleaned speech can be provided to a remote device for further processing which may include Automatic Speech Recognition (ASR).

Abstract: Systems and methods for voice authentication are provided. An example method starts with dynamically generating authentication information and a prompt associated therefor. The authentication information is generated to emphasize differences between an authorized user and others. The authentication information may be generated based on at least one of a user profile, an acoustic environment in which a mobile device is located, and a history of interactions with the mobile device. The authentication information may be an answer to a question which the authorized user would uniquely be able to provide, or it may be a distinctive password. The prompt, for the authentication information, is provided to a user attempting to use the mobile device. The method proceeds with capturing an acoustic sound of a speech of the user and detecting the authentication information in the speech. Based on the detection, a confidence score is determined.

Abstract: Described are noise suppression techniques applicable to various systems including automatic speech processing systems in digital audio pre-processing. The noise suppression techniques utilize a machine-learning framework trained on cues pertaining to reference clean and noisy speech signals, and a corresponding synthetic noisy speech signal combining the clean and noisy speech signals. The machine-learning technique is further used to process audio signals in real time by extracting and analyzing cues pertaining to noisy speech to dynamically generate an appropriate gain mask, which may eliminate the noise components from the input audio signal. The audio signal pre-processed in such a manner may be applied to an automatic speech processing engine for corresponding interpretation or processing.

Abstract: Methods for voice sensing and keyword analysis are provided. An example method allows for causing a mobile device to transition to a second power mode, from a first power mode, in response to a first acoustic signal. The method includes authenticating a user based at least in part on a second acoustic signal. While authenticating the user, the second acoustic signal is compared to a spoken keyword. The spoken keyword is analyzed for authentication strength based on the length of the spoken keyword, quality of a series of phonemes used to represent the spoken keyword, and likelihood of the series of phonemes to be detected by a voice sensing. While receiving the first and second acoustic signals, a signal to noise ratio (SNR) is determined. The SNR is used to adjust sensitivity of a detection threshold of a voice sensing.

Abstract: Systems and methods for performing personalized operations of a mobile device are provided. An example method includes determining a sensor signature has been received. The sensor signature may be a combination of acoustic and non-acoustic inputs. A classification or recognition or matching score for the received sensor signature may be calculated. A context associated with one or more of the operations of the mobile device may be determined. A confidence level or weight may be determined for the acoustic and non-acoustic inputs based on the context. Based on the confidence level or weight, a determination may be made whether to perform the one or more operations. The determination whether to perform the one or more operations may be further based on a combination of the confidence level or weight and the score. Determining the context can include identifying the type and level of noise associated with the mobile device's environment.

Abstract: Systems and methods for multi-sourced noise suppression are provided. An example system may receive streams of audio data including a voice signal and noise, the voice signal including a spoken word. The streams of audio data are provided by distributed audio devices. The system can assign weights to the audio streams based at least partially on quality of the audio streams. The weights of audio streams can be determined based on signal-to-noise ratios (SNRs). The system may further process, based on the weights, the audio stream to generate cleaned speech. Each audio device comprises microphone(s) and can be associated with the Internet of Things (IoT), such that the audio devices are Internet of Things devices. The processing can include noise suppression and reduction and echo cancellation. The cleaned speech can be provided to a remote device for further processing which may include Automatic Speech Recognition (ASR).