Systems and methods are provided for machine learning based monitoring. Image data from a camera is received. On the hardware accelerator, a person detection model based on the image data is invoked. The person detection model outputs first classification result. Based on the first classification result, a person is detected. Second image data is received from the camera. In response to detecting the person, a fall detection model is invoked on the hardware accelerator based on the second image data. The fall detection model outputs a second classification result. A potential fall based on the second classification result is detected. An alert is provided in response to detecting the potential fall.

A pressure sensing mat having first and second conductive layers and an insulative layer. The first conductive layer defines a first aperture and includes first spaced apart conductive regions and first non-conductive regions therebetween. The first spaced apart conductive regions and non-conductive regions extends in a first direction. The second conductive layer defines a second aperture and includes second spaced apart conductive regions and second non-conductive regions therebetween. The second spaced apart conductive regions and the second non-conductive regions extend in a second direction different than the first direction. The insulative layer is between the first and second conductive layers and defines a third aperture. The first, second, and third apertures are aligned with each other such that the first, second, and third apertures form a vent through the first and second conductive layers and the insulative layer.

A monitoring system a user activity sensor to determine patterns of activity based upon the user activity occurring over time.

A wearable defibrillator includes one or more environmental sensors including an accelerometer, one or more ECG sensors configured to acquire ECG data, and an alarm manager and at least one processor operatively coupled to the one or more ECG sensors and the accelerometer. The at least one processor is configured to detect a cardiac abnormality in the patient, identify a notification having one or more characteristics, and detect an environmental condition of the wearable defibrillator. The accelerometer is configured to detect a presence or lack of patient motion, and/or detect a body position of the patient as the detected environmental condition. The at least one processor is configured to determine whether one or more factors exist that inhibit the patient's ability to recognize the notification, on determining their existence, adapt the one or more characteristics of the notification and provide an adapted notification, and issue the adapted notification.

Automatically gathering and processing data relating to risk of patient falls. The data is analyzed to determine a level of risk each patient has for falling. Patient risk for falls is stratified and protocols are implemented to mitigate that risk. The system communicates instructions and alerts to caregivers to complete tasks or provide aid to patients. Updates to patient medical information and updates to best practices in fall risk management result in updates to patient risk stratifications. In turn, tasks and alerts are updated to reflect updates to patient risk stratifications.

Systems and methods are disclosed herein for switching an application executing on an ambulatory medical device to a new application without interrupting therapy provided by the ambulatory medical device to a subject. The ambulatory medical device may receive an indication that an update to an application executing on the ambulatory insulin pump is available, establish a communication connection to a host computing system, download and install the application update, while a prior version of the application continues to run. The disclosed systems and methods can confirm successful installation of the application update on the ambulatory medical device and switch control of the ambulatory medical device from the prior version to the new version of the application without interrupting therapy provided to the subject.

A millimeter wave radar apparatus determining a fall posture is applied to a human body. The millimeter wave radar apparatus includes a microprocessor and a millimeter wave radar. The millimeter wave radar is electrically connected to the microprocessor. The millimeter wave radar is configured to transmit a radar wave to the human body. The millimeter wave radar is configured to receive a reflected radar wave reflected from the human body based on the radar wave. The microprocessor is configured to obtain a point cloud information based on the reflected radar wave. The microprocessor is configured to utilize the point cloud information to determine whether the human body is in the fall posture.

A method and system for monitoring a user's intoxication including receiving a set of signals, derived from a set of samples collected from the user at a set of time points; providing a sobriety task to the user proximal to a time point of the set of time points; generating a performance dataset characterizing performance of the sobriety task by the user; receiving a supplementary dataset characterizing a demographic profile of the user and/or a physiological state of the user; determining a set of values of an intoxication metric, derived from the set of signals; generating a predicted temporal profile of the intoxication metric for the user based upon the set of values, the set of time points, and the supplementary dataset; generating an analysis of the user's sobriety based upon the performance dataset and the predicted temporal profile; and providing a notification to the user based upon the analysis.

A biological signal of a subject is acquired so as to calculate biological information from the acquired biological signal. When the biological information has been determined to be anomaly, whether the biological information is one that was calculated under a high-accuracy condition is determined. When the biological information is determined to be one that was calculated under the high-accuracy condition, a notice is given based on a first criterion. In the other cases, a notice is given based on a second criterion. Thereby, it is possible to provide an abnormality notification system that can give a necessary notification appropriately while suppressing unnecessary notification, by changing the criteria for notification in accordance with the accuracy of the determined biological information when the biological information of the subject was determined to be anomaly.

There is provided a method and apparatus for detecting a fall event of the user. In particular, the method includes collecting data associated with activities of the user from a plurality of sensors and distributing the collected data to data sub-windows using signal windowing and segmentation, the data sub-windows indicative of a pre-fall moment, a fall moment, and a post-fall moment. The method further includes extracting a plurality of features from one or more of the data sub-windows and determining whether the event is a fall event at least in part based on the extracted features. The determination of whether the event is a fall event can further be determined by applied support vector machine (SVM) technique. The developed machine learning based methods may be substantially optimum in terms of a trade-off between accuracy and complexity of the evaluation. The method further includes multiple rejection filters in order to aid with the prevention of false alarms due to fall-like activities of daily living (ADLs). The method further includes a personalization process to update the machine learning based methods associated with each user.