Abstract: A system that calculates the probability that a patient will experience a seizure using a machine learning system. Inputs to the machine learning system may include ECG, EEG, vital signs, lab results, and patient history and demographics. Features may be extracted from the raw data and input into the machine learning system. Neurological features calculated from EEG waveforms may include characterizations of discharges, such as their frequency, duration, amplitude, and shape. Cardiovascular features calculated from ECG waveforms may include characterizations of P, Q, R, S, T waveforms, and heart rate variability metrics. Relationships between neurological and cardiovascular features may also be calculated and input into the system; these relationships may include for example correlations, cross-coherence, wavelet correlation, and cross-spectral entropy. The seizure risk probability may be updated periodically as new patient data arrives.

Abstract: A method that selects high-quality data within recorded signals from patient monitoring devices, where portions of the signals may be corrupted by noise and should therefore be excluded. Signals are compared to models of expected signal characteristics, and portions of the signals that do not match the models may be excluded. Some models may check for expected relationships between signals from different devices. One such model identifies feature points in two signals from two different devices and calculates the time difference between each feature point in one signal and the earliest subsequent feature point in the other signal; data is excluded if this time difference exceeds an expected range. For example, an expected relationship between electrocardiogram and blood pressure signals is that the R-wave peak should be followed by a blood pressure peak within an expected delay time (the pulse transit time); this check can exclude invalid ECG/BP data.

Abstract: A method for time-synchronizing waveforms from different patient monitors that does not require devices to have high-precision synchronized clocks or to be coupled to a triggering synchronization signal generator. Comparable signals may be obtained from different devices either by placing selected sensors from the devices in the same locations, or by filtering signals from one device to obtain a signal comparable to signals from another device. Filtering may for example transform waveforms into independent components and identify a component that matches a signal from another device. The comparable signals may then be transformed into frequency variation curves, such as time intervals between peak values, to facilitate detection of the time shift between the signals. Cross correlation of the frequency variation curves may be used to locate the precise time shift between the signals. Use of frequency variation curves may be more robust than directly comparing and correlating the original signals.

Abstract: A system that analyzes alarms from patient monitoring devices and calculates modified alarm thresholds that reduce the number of alarms to a desired level. This capability addresses the common problem of alarm overload and fatigue, where alarms occur so frequently that clinicians cannot respond effectively. The system supports highly efficient calculation of changes in alarm frequency, by storing summaries of alarm data that record maximum and minimum values during an alarm. New thresholds may be selected manually or automatically and may be transmitted directly to the patient monitoring devices as updates to their alarm thresholds. The system may also classify alarms as high (above an upper threshold) or low (below a lower threshold) when devices do not provide this data. The system may also estimate the number of additional alarms that would occur if an upper threshold were reduced, or a lower threshold were increased.

Abstract: A system that analyzes alarms from patient monitoring devices that trigger after a configurable delay time, and that calculates modified delay times that reduce the number of alarms to a desired level. This capability addresses the common problem of alarm overload and fatigue, where alarms occur so frequently that clinicians cannot respond effectively. The system supports highly efficient calculation of changes in alarm frequency, by storing summaries of alarm data that record alarm durations and first values during the alarm. New delays may be selected manually or automatically and may be transmitted directly to the patient monitoring devices as updates to their alarm delay times. The system may also estimate a current delay time when devices do not provide this data. The system may also estimate the number of additional alarms that would occur if an alarm delay were reduced.

Abstract: A system that synchronizes waveforms received over a network from one or more devices, such as medical devices. Because of network delays or losses, waveforms can arrive at varying rates and times. Precise post-synchronization of the received data, to within a few milliseconds, is needed for accurate analysis. Applications include automatic classification of waveforms, such as detection of myocardial infraction from heart monitor waveforms. Synchronization uses sequence numbers assigned by each device, but must also account for sequence number wraparounds. Waveforms may also be synchronized across devices, by calculating the bias between within-device synchronized times and a common time source or common disturbance. Waveform data may also be stored data in a database or data warehouse; embodiments may index the data using a key with a date-time prefix and a hash code suffix, to support distributed indexing while reducing the chance of hash collisions to a very small probability.

Abstract: A system that synchronizes waveforms received over a network from one or more devices, such as medical devices. Because of network delays or losses, waveforms can arrive at varying rates and times. Precise post-synchronization of the received data, to within a few milliseconds, is needed for accurate analysis. Applications include automatic classification of waveforms, such as detection of myocardial infraction from heart monitor waveforms. Synchronization uses sequence numbers assigned by each device, but must also account for sequence number wraparounds. Waveforms may also be synchronized across devices, by calculating the bias between within-device synchronized times and a common time source or common disturbance. Waveform data may also be stored data in a database or data warehouse; embodiments may index the data using a key with a date-time prefix and a hash code suffix, to support distributed indexing while reducing the chance of hash collisions to a very small probability.

Abstract: A method for time-synchronizing waveforms from different patient monitors that does not require devices to have high-precision synchronized clocks or to be coupled to a triggering synchronization signal generator. Comparable signals may be obtained from different devices either by placing selected sensors from the devices in the same locations, or by filtering signals from one device to obtain a signal comparable to signals from another device. Filtering may for example transform waveforms into independent components and identify a component that matches a signal from another device. The comparable signals may then be transformed into frequency variation curves, such as time intervals between peak values, to facilitate detection of the time shift between the signals. Cross correlation of the frequency variation curves may be used to locate the precise time shift between the signals. Use of frequency variation curves may be more robust than directly comparing and correlating the original signals.

Abstract: A system that synchronizes waveforms received over a network from one or more devices, such as medical devices. Because of network delays or losses, waveforms can arrive at varying rates and times. Precise post-synchronization of the received data, to within a few milliseconds, is needed for accurate analysis. Applications include automatic classification of waveforms, such as detection of myocardial infraction from heart monitor waveforms. Synchronization uses sequence numbers assigned by each device, but must also account for sequence number wraparounds. Waveforms may also be synchronized across devices, by calculating the bias between within-device synchronized times and a common time source or common disturbance. Waveform data may also be stored data in a database or data warehouse; embodiments may index the data using a key with a date-time prefix and a hash code suffix, to support distributed indexing while reducing the chance of hash collisions to a very small probability.

Abstract: A system that synchronizes waveforms received over a network from one or more devices, such as medical devices. Because of network delays or losses, waveforms can arrive at varying rates and times. Precise post-synchronization of the received data, to within a few milliseconds, is needed for accurate analysis. Applications include automatic classification of waveforms, such as detection of myocardial infraction from heart monitor waveforms. Synchronization uses sequence numbers assigned by each device, but must also account for sequence number wraparounds. Waveforms may also be synchronized across devices, by calculating the bias between within-device synchronized times and a common time source or common disturbance. Waveform data may also be stored data in a database or data warehouse; embodiments may index the data using a key with a date-time prefix and a hash code suffix, to support distributed indexing while reducing the chance of hash collisions to a very small probability.