WO2019241444A1 - Prédiction du risque d'arrêts cardiaques en hôpital à l'aide de données de surveillance de télésurveillance électrocardiographique en continu - Google Patents

Prédiction du risque d'arrêts cardiaques en hôpital à l'aide de données de surveillance de télésurveillance électrocardiographique en continu Download PDF

Info

Publication number
WO2019241444A1
WO2019241444A1 PCT/US2019/036850 US2019036850W WO2019241444A1 WO 2019241444 A1 WO2019241444 A1 WO 2019241444A1 US 2019036850 W US2019036850 W US 2019036850W WO 2019241444 A1 WO2019241444 A1 WO 2019241444A1
Authority
WO
WIPO (PCT)
Prior art keywords
block
negative
ecg
positive
trend
Prior art date
Application number
PCT/US2019/036850
Other languages
English (en)
Inventor
Duc Hong HO
Noel Gerard BOYLE
Alan Kuo
Xiao Hu
Original Assignee
The Regents Of The University Of California
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by The Regents Of The University Of California filed Critical The Regents Of The University Of California
Publication of WO2019241444A1 publication Critical patent/WO2019241444A1/fr

Links

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
    • A61B5/346—Analysis of electrocardiograms
    • A61B5/349—Detecting specific parameters of the electrocardiograph cycle
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/0002—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network
    • A61B5/0004—Remote monitoring of patients using telemetry, e.g. transmission of vital signals via a communication network characterised by the type of physiological signal transmitted
    • A61B5/0006—ECG or EEG signals
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7271—Specific aspects of physiological measurement analysis
    • A61B5/7275—Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
    • G16H40/60—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices
    • G16H40/63—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices for local operation
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Definitions

  • IHCA In-hospital cardiac arrests
  • PDA/asystole arrests account for 70-80% of arrests, and fewer than 12% of patients suffering these arrests survive to discharge.
  • ECG data is continuously acquired for many hospitalized patients.
  • An aspect of the present description is a system and method for predicting the occurrence of cardiac arrests in a patient being monitored continuously by electrocardiogram by evaluation and comparison of trends in electrocardiographic parameters, which is incorporated into predictive models.
  • Use of continuous ECG data, while more demanding, overcomes weaknesses of existing systems by continuously providing information reflecting patients’ physiologic state, particularly in those not requiring invasive monitors.
  • continuous ECG monitoring is used for monitoring
  • One aspect is a method for processing continuous
  • electrocardiographic parameters derived from a signal averaged beat The following steps are then taken, which are continuously updated over time as data continues to be collected: filtering is performed to reduce noise and remove non-physiologic information obtained from artifact measurements; identifying aberrant jumps in the averaged beat time series that arise from abrupt changes in electrocardiographic morphology or changes in electrode placement; breaking up the time series into segments at the identified “jumps”; and then merging like segments (segments deemed continuous) back together. Each resultant segment is then evaluated separately in further steps. Varying time windows (minutes to hours) may be selected from the resultant segments. For each window, the dominant positive and negative trends are determined for each of the measured parameters.
  • the following method is used to derive the dominant trends: (a) at each point, the slope of change (positive or negative) is determined by performing robust linear regression using a number of points centered on this point (the number of points utilized varies by window size); (b) performing merging of short segments into longer, more dominant trends; and (c) defining a dominant trend for a particular window length as the trend in a direction (positive or negative) with the longest duration. For each dominant trend, the trend duration, maximal change in the ECG parameter value, and slope of trend are calculated. Comparisons between the above features may then be compared to those in the prior time window of the same length
  • the continuous electrocardiographic waveform data is furthermore processed into a time series of all RR intervals. This data is processed to evaluate for the presence of arrhythmias (atrial fibrillation, second degree heart block, pause) within a variety of time windows. The presence of a new arrhythmia is determined by comparing the presence of an arrhythmia compared to prior time windows.
  • arrhythmias atrial fibrillation, second degree heart block, pause
  • all variables measured as detailed above are combined in a random forest predictive model which will provide a risk of cardiac arrest.
  • a threshold for a risk score is selected, above which a patient is predicted to be high risk for impending cardiac arrest.
  • FIG. 1 shows a schematic flow diagram of a method for processing of data and model development for continuous electrocardiographic waveform monitoring in accordance with the present description.
  • FIG. 2A through FIG. 2H show plots illustrating various filtering
  • FIG. 3A and FIG 3B illustrate block selection for case PR data and control PR data, respectively.
  • FIG. 4A illustrates the linear fitting process for merging segments for case PR data.
  • FIG. 4B shows the resultant dominant trends for case PR data after a short negative trend segment is merged with the longer positive trend segment.
  • FIG. 5 shows an exemplary random forest importance plot.
  • FIG. 6 shows a schematic diagram of an IHCA monitoring system for monitoring a patient in accordance with the present description.
  • FIG. 7A through FIG. 7G show plots illustrating patterns of
  • FIG. 8 is a series of plots showing a comparison of
  • electrocardiographic parameter changes over time.
  • FIG. 9A through FIG. 9C show receiver operating characteristics curves for the model development set (logistic regression with backward stepwise regression, logistic regression with forward stepwise regression, and random forest, respectively).
  • FIG. 10A through FIG. 10C show receiver operating characteristics curves for the model testing on validation case block 1 vs. control block 2.
  • FIG. 11 A through FIG. 11 C show receiver operating characteristics curves for the model testing on case block 2 vs. validation set control block 1 .
  • FIG. 12A through FIG. 12C show receiver operating characteristics curves for the model testing on case block 2 vs. control block 2.
  • FIG. 1 shows a schematic flow diagram of a method 10 of
  • patient data 12a is acquired, including case patient data from patients with a known cardiac condition or event (e.g. code blue event) such as pulse-less electrical activity (PEA) or asystole, and control patient data without code blue events.
  • a known cardiac condition or event e.g. code blue event
  • PEA pulse-less electrical activity
  • asystole e.g., asystole
  • raw telemetry data from the acquired control and case patient data is processed to obtain: 1 ) a minute-by-minute time series 14a of ECG parameters, including but not limited to PR interval (the beginning of the P wave to the beginning of the QRS complex), QRS duration, averaged QRS amplitude, ST segment height in lead II and V2, QTc, and RR) derived from a 5-minute signal-averaged beat obtained in a rolling window fashion, and 2) a time series 14b of all consecutive RR intervals.
  • the averaged beat-derived time series 14a is processed with a series of filters 16a to reduce noise and remove non-physiologic data from supraventricular arrhythmias or pacing.
  • the RR time series 14b was processed with artifact filter 16b for removal of data points affected by excessive signal artifact.
  • FIG. 2A through FIG. 2H show plots illustrating various filtering
  • step 16 for filtering averaged-beat ECG parameters.
  • the examples shown are for PR interval data from different patients. Dashed zones in the images detail points that are retained or removed following application of the filter.
  • FIG. 2A and FIG. 2B illustrate case PR data prior to and after
  • a median filter for reducing signal noise.
  • the median filter reduces artifacts and outliers while preserving signal. The larger the k value, the greater the decrease in artifact but this comes at the cost of signal loss and increase in time delay.
  • FIG. 2C and FIG. 2D illustrate a tachyarrhythmia filter applied to control FIR data and control PR data, respectively, to remove segments of noncomparable data due to tachyarrhythmias.
  • a tachyarrhythmia filter applied to control FIR data and control PR data, respectively, to remove segments of noncomparable data due to tachyarrhythmias.
  • the post-median filter RR interval time series was processed minute by minute to look for data points that exceeded 115% of the maximum of the preceding 20 points or was 85% less than the minimum of the preceding 15 points. Each such point was flagged; N flags resulted in N+1 segments. If the mean RR for each segment was less than 80% of the mean of the entire RR time series (i.e. segment FIR exceeded 20% of the mean FIR), the particular segment was flagged as a tachyarrhythmia block. PR interval, P wave duration, QTc, and RR data were removed for tachyarrhythmia blocks given the lack of comparability to non- tachyarrhythmia data for the purposes of trend analysis.
  • FIG. 2E and FIG. 2F illustrate a pacer dependence filter applied to case FIR data and case PR data, respectively, to remove segments of paced data from analysis.
  • the pacer-dependence filter identifies extended periods of extremely constant (invariable) FIR, which is not physiologic.
  • An extended constant FIR was defined as lasting at least 120 consecutive FIR data points. Thus, all 120 consecutive data point windows were processed to check for invariability. The median value of these 120 data points was subtracted from each of the 120 FIR values. If the maximal absolute difference of these 120 differences were less than 0.12 heart beats, then the 120-data point window was extended downstream one data point at a time, undergoing the same iterative checks as described above, until the largest window of invariability was identified.
  • Corresponding metric data was dropped as pacing would artificially change the ECG metric of interest. Multiple windows of complete pacer dependence could be identified in a single patient. The minimum of 120 data points and the 0.12 maximal allowable heart rate difference were chosen purposefully to be extremely stringent to avoid wrongfully eliminating non-paced data.
  • FIG. 2H shows a continuity filter applied to case PR data to remove significant outlier points which are likely artefactual.
  • Data points for cases more than 24 hours from arrest are shown here only for illustration purposes.
  • the data points representing calculations made for filtering, as well as excluded points are shown in dashed lines.
  • all metrics of interest, except ST were processed through the continuity filter based on the assumption that changes to ECG metrics should be gradual and that large jumps are most likely secondary to artifacts and/or noise.
  • FIG. 2G shows histogram created with 10 equal sized bins. Data points belonging to the highest frequency bin were defined as "modes". A continuity check was conducted in both a forwards and backwards fashion starting at each "mode".
  • FIG. 3A and FIG. 3B illustrate block selection for case PR data and control PR data, respectively.
  • Three consecutive 3-hour blocks (blocks 3, 2, 1 in that order) were selected for further analysis: in case patients, block 1 immediately preceded IHCA whereas in control patients block 1 was selected at random. Blocks 3 and 2 were then selected as the two 3-hour blocks that immediately preceded block 1 in either case or control patients.
  • electrocardiographic morphology or changes in electrode placement is performed first.
  • the time series is broken up into segments at these “jumps” prior to final block selection.
  • step 20 the selected blocks are processed.
  • step 20a is performed to determine the dominant positive and negative trend for each ECG parameter.
  • a block of data In each block of data, a
  • FIG. 4A illustrates the linear fitting process for merging segments for case PR data. Zones illustrated with dashed lines denotes negative trends. Small/short segments flanked by larger/longer segments of the opposite slope sign/directionality are merged into the larger segment. Change in dominant positive and negative trends in block n (Ay n + , Ay n ) is calculated by subtracting the maximal and minimal value for the trend. The slope of the dominant positive and negative trends (A y n+ /A x n+ , A y n 7A x n ), calculated by dividing the change over the duration.
  • FIG. 4B shows the resultant dominant trends for case PR data after a short negative trend segment is merged with the longer positive trend segment.
  • the dominant trend in either direction is then determined by the trend in that direction with the longest duration, i.e. the dominant positive and negative trends are defined as the trend with the longest consecutive number of data points with positive and negative slopes respectively, were then determined for each block.
  • step 20b is performed to identify periods of atrial fibrillation (AF), second degree heart block (2° HB), and pauses greater than 3 seconds.
  • AF atrial fibrillation
  • 2° HB second degree heart block
  • pauses greater than 3 seconds a modified method of the non-empty-cells approach was employed. It is appreciated, however, that other methods available in the art may be used to determine atrial fibrillation at a time point.
  • a sliding window size of 128 beats was chosen with the window sliding/shifting by a single beat across time. As such, each beat was covered by 128 consecutive sliding windows containing 128 consecutive beats. For each beat the RR interval in milliseconds (ms) was noted.
  • the (RR, dRR) pair of each beat is then graphed on a 2-D plot with RR on the x-axis and dRR on the y-axis.
  • the plot was then segmented into squares or "cells" that are 25 by 25 ms in size.
  • the number of cells that contain at least 1 pair of (RR, dRR) is defined as a non-empty cell (NEC).
  • the number of NECs is the surrogate for the degree of irregularity with higher NECs correlating with higher degree of irregularity and higher probability of atrial fibrillation.
  • Atrial fibrillation episodes were annotated in 117 cases and 117 controls. 700 iterations of random sampling were then performed, with replacement, from these cases/controls and the F1 statistic (which balances sensitivity and positive predictive value) was calculated using every combination of NEC and number of windows. This data was then compiled to determine the optimal threshold over the 700 iterations.
  • the optimal F1 statistic was obtained with a threshold of 52 NEC and 100% of windows, with an F1 value of 0.87 (Sensitivity 92.4%, Specificity 97.5%, PPV 81.6%, NPV 99.1 %).
  • step 22 block comparison is performed. The difference in
  • dominant positive and negative trend change and slope between a patient’s block n and immediately preceding block (n-1 ) is determined at step 22a.
  • Four values are calculated: difference in dominant: 1 ) positive change (Ay n - (n-i) + ), 2) negative change (Dg h -(h- ⁇ ) ), 3) positive slope (Dg h + /Dc h + - Dg( P - ⁇ )7Dc (h - ⁇ ) + ), and 4) negative slope (Dg h /Dc h - Dg(h-1) /Dc(h-1) ) ⁇
  • other parameters e.g. trend duration, maximal change in the ECG parameter value, may also be calculated and used in comparison step 24 detailed below.
  • continuous variables are assessed for normality using the Shapiro-Wilks test.
  • the Wilcoxon signed-rank test is used for within-group comparisons and Wilcoxon rank-sum test for between-group comparisons given non-normality of many variables.
  • an indicator variable is calculated for the presence of those arrhythmias in block n, and a second for the presence of those arrhythmias in block n but not block (n-1 ).
  • step 24 the case and control block 1 are divided into an 80% model development and 20% validation set by stratified random sampling based on case/control status.
  • Using the development set a univariate logistic regression analyses is performed, and any variable with p ⁇ 0.05 is retained for use in multivariable model development. Missing values are imputed using multiple imputations.
  • 3 models were created: logistic regression with backward stepwise regression, forward stepwise regression, and random forest with 10,000 trees.
  • FIG. 5 shows an exemplary random forest importance plot, which shows the 15 variables with the highest importance in the random forest, based on mean decrease in the Gini coefficient (measure of how much each variable contributes to homogeneity of the nodes and leaves in the random forest). Higher values signify higher importance of the variable.
  • each model of the validation set was evaluated by using the area under the curve (AUC) and sensitivity for classifying a block as IHCA while maintaining a low false positive rate (FPR).
  • AUC area under the curve
  • FPR sensitivity for classifying a block as IHCA while maintaining a low false positive rate
  • further testing of model robustness was performed by setting a classification threshold where FPR on the validation set was approximately 5% and evaluated the sensitivity and specificity on case block 1 vs. case block 2 (temporal differentiation of IHCA detection), validation case block 1 vs. all control block 2, all case block 2 vs. validation control block 1 , all case block 2 vs. all control block 2.
  • model development steps 12 through 24 and evaluation/testing steps 26 and 28 in FIG. 1 may be implemented as programming or software for building a predictive model capable of continuous ECG trend analysis and risk calculation for code blue events/cardiac arrests. It is appreciated that the above analysis and risk calculation may also be performed in combination with other non-ECG data, including vital signs, monitor alarms, laboratory results, etc.
  • FIG. 6 shows a schematic diagram of an IHCA monitoring system 50 for monitoring a patient 58 in accordance with the present description.
  • Continuous ECG monitoring 62 is performed via ECG leads 60 placed on the patient.
  • system 50 incorporates an afferent limb 52 where the ECG signal is continually processed and updated via computer or remote server 60, and an efferent limb 56 where IHCA risk is continually updated to alert a physician if a threshold is reached.
  • Computer or remote server 60 comprises a processor 62 and instructions/application programming 66 stored in memory 64 executable on the processor 62 for performing continuous ECG trend analysis and risk calculation 68.
  • Application software 66 preferably incorporates one or more processes detailed in FIG.
  • Application programming 66 may comprise one or more predictive models developed using method 10, the predictive model configured for continuous ECG trend analysis and risk calculation for code blue events/cardiac arrests, and providing a warning to the physician wherein a certain threshold for risk of a code blue event.
  • a two-sided p ⁇ 0.05 was considered statistically significant.
  • FIG. 7A through FIG. 7G show plots illustrating patterns of
  • electrocardiographic parameter changes by cause of cardiac arrest The different figures show block 1 (final 3 hours before cardiac arrest) in 7 case patients, with the determined cause of the cardiac arrest. Graphs from top to bottom show trends in RR, QRS duration, ST lead II, and ST lead V2.
  • FIG. 7A through FIG. 7C all show shortening of the RR followed by terminal prolongation of RR, with similar patterns of ST changes also seen in FIG. 7A and FIG. 7B. Different patterns can be noted despite similar determined cause of cardiac arrest (e.g. FIG. 7D vs. FIG. 7A through FIG. 7C).
  • Unknown causes of cardiac arrest can potentially be matched to cases with known causes to determine the possible cause of arrest (e.g. FIG. 7F and FIG. 7G).
  • FIG. 8 is a series of plots showing a comparison of
  • IHCA in-hospital cardiac arrest
  • FIG. 9A through FIG. 9C show receiver operating characteristics curves for the model development set (logistic regression with backward stepwise regression, logistic regression with forward stepwise regression, and random forest, respectively).
  • FIG. 10A through FIG. 10C show receiver operating characteristics curves for the model testing on validation case block 1 vs. control block 2.
  • Block 1 is the 3-hour block immediately preceding in-hospital cardiac arrest, while Block 2 is the 3-hour block preceding block 1.
  • Validation set case block 1 and all of control block 2 are included in this analysis.
  • the (X) marks the sensitivity and specificity at the threshold chosen based on the validation set.
  • AUC area under the curve.
  • FIG. 11 A through FIG. 11 C show receiver operating characteristics curves for the model testing on case block 2 vs. validation set control block 1.
  • Block 1 is the 3-hour block immediately preceding in-hospital cardiac arrest, while Block 2 is the 3-hour block preceding block 1. All case block 2 and validation set control block 1 are included in this analysis.
  • FIG. 12A through FIG. 12C show receiver operating characteristics curves for the model testing on case block 2 vs. control block 2.
  • Block 1 is the 3-hour block immediately preceding in-hospital cardiac arrest, while Block 2 is the 3-hour block preceding block 1. All case block 2 and all control block 2 are included in this analysis.
  • 0.1 mV ST depression in a patient with baseline left ventricular hypertrophy confers a significantly lower risk compared to a patient with a baseline normal ST segment.
  • QRS prolongation by intraventricular conduction delay can reflect progressive ischemia
  • a decrease in measured QRS duration is likely artefactual as a result of decreasing QRS amplitude, reported in septic shock states.
  • ECG parameters particularly RR interval and ST segment
  • RR interval and ST segment can fluctuate significantly even in healthy states with exertion/rest and other stressors.
  • Flence differentiation of physiologic versus pathologic changes (e.g. physiologic slowing of heart rate versus slowing heart rate preceding many types of PEA), can be better deduced by comparing magnitudes and rates of change to those in earlier time periods when the patient was known to be stable (see FIG. 8).
  • atrial fibrillation in patients with a prior history is not.
  • window alone can provide predictive information for imminent IHCA independent of acquisition of other patient data. It is appreciated that selection of a 3-hour window is exemplary, and that other timeframes may be selected as appropriate and risk models combining multiple window lengths may also be used as appropriate.
  • risk factors ECG metrics in our study
  • ECG metrics ECG metrics in our study
  • logistic regression models may perform poorly due to model instability.
  • non-linear classifiers such as the random forest model perform better with such classification problems, at the cost of becoming a“black box”. While the random forest was used in this study, it is appreciated that other types of classifier models available in the art may be implemented for risk prediction in accordance with the present description.
  • the AUC achieved by the logistic regression models were slightly superior to that of the random forest model, since relatively few patients suffer IHCA and therefore models with low false positive rate for similar sensitivity achieved may be preferred to prevent excessive false alarms.
  • the random forest model performed best, attaining 63.2% sensitivity with 94.6% specificity at the selected threshold.
  • the random forest model furthermore, showed excellent temporal discriminatory ability with a 91.2% sensitivity and 83.5% specificity at distinguishing case block 1 vs. block 2.
  • ECG metrics may also complement predictive algorithms that
  • ECG changes which predominantly occur in the several hours immediately preceding IHCA, can help pinpoint the patient at“impending risk” of IHCA or who is rapidly deteriorating clinically.
  • Feedback to clinicians can be provided with an IHCA“impending risk” score and updated in real-time, allowing for earlier detection of patient deterioration, earlier clinical intervention, and potentially improved patient outcomes.
  • ECG parameters such as heart rate variability, the use of different detection window durations, and concurrent changes in 2 or more ECG parameters may also be considered in the system and methods of the present description.
  • Embodiments of the present technology may be described herein with reference to flowchart illustrations of methods and systems according to embodiments of the technology, and/or procedures, algorithms, steps, operations, formulae, or other computational depictions, which may also be implemented as computer program products.
  • each block or step of a flowchart, and combinations of blocks (and/or steps) in a flowchart, as well as any procedure, algorithm, step, operation, formula, or computational depiction can be implemented by various means, such as hardware, firmware, and/or software including one or more computer program instructions embodied in computer-readable program code.
  • any such computer program instructions may be executed by one or more computer processors, including without limitation a general-purpose computer or special purpose computer, or other programmable processing apparatus to produce a machine, such that the computer program instructions which execute on the computer processor(s) or other programmable processing apparatus create means for
  • blocks of the flowcharts, and procedures, algorithms, steps, operations, formulae, or computational depictions described herein support combinations of means for performing the specified function(s), combinations of steps for performing the specified function(s), and computer program instructions, such as embodied in computer-readable program code logic means, for performing the specified function(s).
  • each block of the flowchart illustrations, as well as any procedures, algorithms, steps, operations, formulae, or computational depictions and combinations thereof described herein can be implemented by special purpose hardware-based computer systems which perform the specified function(s) or step(s), or combinations of special purpose hardware and computer-readable program code.
  • embodied in computer-readable program code may also be stored in one or more computer-readable memory or memory devices that can direct a computer processor or other programmable processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory or memory devices produce an article of manufacture including instruction means which implement the function specified in the block(s) of the flowchart(s).
  • the computer program instructions may also be executed by a computer processor or other programmable processing apparatus to cause a series of operational steps to be performed on the computer processor or other programmable processing apparatus to produce a computer-implemented process such that the instructions which execute on the computer processor or other programmable processing apparatus provide steps for implementing the functions specified in the block(s) of the flowchart(s), procedure (s) algorithm(s), step(s), operation(s), formula(e), or computational
  • program executable refer to one or more instructions that can be executed by one or more computer processors to perform one or more functions as described herein.
  • the instructions can be embodied in software, in firmware, or in a combination of software and firmware.
  • the instructions can be stored local to the device in non-transitory media or can be stored remotely such as on a server, or all or a portion of the instructions can be stored locally and remotely. Instructions stored remotely can be downloaded (pushed) to the device by user initiation, or automatically based on one or more factors.
  • processors, hardware processor, computer processor, central processing unit (CPU), and computer are used synonymously to denote a device capable of executing the instructions and communicating with input/output interfaces and/or peripheral devices, and that the terms processor, hardware processor, computer processor, CPU, and computer are intended to encompass single or multiple devices, single core and multicore devices, and variations thereof.
  • a system for monitoring electrocardiographic (ECG) waveform data from a patient and predicting risk of a cardiac arrest comprising: a processor configured for receiving input from a continuous ECG waveform data stream acquired from one or more electrodes; and a memory storing programming executable on the processor; wherein when executed the programming performs one or more steps comprising: (a) acquiring an averaged beat time series of one or more ECG parameters from a signal averaged beat of the data stream; (b) filtering the data stream to reduce noise and remove non-physiologic information obtained from artefactual measurements; (c) selecting a plurality of consecutive blocks within the filtered data stream, each block having a specified time interval; (d) determining dominant positive and negative trends within each block for the one or more ECG parameters; (e) comparing the dominant positive and negative trends between a first block and a preceding block; and (f) developing a predictive model based on the comparison of the dominant positive and negative trends, wherein the predictive model is configured for prediction or identification of risk of
  • ECG parameters comprise one or more of: QRSd, QTc, RR, ST lead II and V2.
  • filtering the data stream comprises using one or more of the following filters: a median 2k+1 filter, tachyarrhythmia filter, pacer-dependence filter, and continuity filter.
  • arrhythmias comprise one of: periods of atrial fibrillation (AF), second degree heart block (2° HB), and pauses greater than 3 seconds wherein the inter-beat (RR) intervals are filtered with an artifact filter.
  • AF atrial fibrillation
  • 2° HB second degree heart block
  • RR inter-beat
  • determining dominant positive and negative trends comprises: (a) determining whether a slope at each point, is positive or negative; (b) joining consecutive points with the same slope directionality to form segments; (c) merging short segments into longer, more dominant trends; and (d) defining dominant positive and negative trends as a trend with a longest consecutive number of data points with positive and negative slopes respectively.
  • a system for monitoring electrocardiographic (ECG) waveform data from a patient and predicting risk of a cardiac arrest comprising: a plurality of ECG leads configured to be disposed at respective locations on a patient; a processor configured for receiving input from a continuous ECG waveform data stream acquired from one or more ECG electrodes; and a memory storing programming executable on the processor; wherein when executed the programming performs one or more steps comprising: (a)acquiring an averaged beat time series of one or more ECG parameters from a signal averaged beat of the data stream; (b) filtering the data stream to reduce noise and remove non-physiologic information obtained from artefactual measurements; (c) selecting a plurality of consecutive blocks within the filtered data stream, each block having a specified time interval; (d)determining dominant positive and negative trends within each block for the one or more ECG parameters; (e) comparing the dominant positive and negative trends between a first block and a preceding block; and (f) developing a predictive model based on the comparison of
  • ECG parameters comprise one or more of: QRSd, QTc, RR, ST lead II and V2.
  • filtering the data stream comprises using one or more of the following filters: a median 2k+1 filter, tachyarrhythmia filter, pacer-dependence filter, and continuity filter.
  • arrhythmias comprise one of: periods of atrial fibrillation (AF), second degree heart block (2° HB), and pauses greater than 3 seconds.
  • determining dominant positive and negative trends comprises: determining whether a slope at each point, is positive or negative; joining consecutive points with the same slope directionality to form segments; merging short segments into longer, more dominant trends; and defining dominant positive and negative trends as a trend with a longest consecutive number of data points with positive and negative slopes respectively.
  • a method for monitoring electrocardiographic (ECG) waveform data from a patient and predicting risk of a cardiac arrest comprising: receiving input from a continuous ECG waveform data stream acquired from one or more electrodes disposed at a patient; and acquiring an averaged beat time series of one or more ECG parameters from a signal averaged beat of the data stream; filtering the data stream to reduce noise and remove non-physiologic information obtained from artefactual measurements; selecting a plurality of consecutive blocks within the filtered data stream, each block having a specified time interval; determining dominant positive and negative trends within each block for the one or more ECG parameters; comparing the dominant positive and negative trends between a first block and a preceding block; and developing a predictive model based on the comparison of the dominant positive and negative trends, wherein the predictive model is configured for prediction or identification of risk of cardiac arrest in a patient being monitored via acquisition of continuous ECG waveform data; wherein said method is performed by a processor executing instructions stored on a non-transitory medium.
  • [00128] 34 The system or method of any of the preceding or subsequent embodiments: wherein for each dominant positive or negative trend, a difference in dominant positive and negative trend change and slope are calculated; and wherein the difference in dominant positive and negative trend change and slope between neighboring blocks are compared.
  • [00130] 36 The system or method of any of the preceding or subsequent embodiments, wherein the one or more ECG parameters comprise one or more of: QRSd, QTc, RR, ST lead II and V2.
  • filtering the data stream comprises using one or more of the following filters: a median 2k+1 filter, tachyarrhythmia filter, pacer-dependence filter, and continuity filter.
  • arrhythmias comprise one of: periods of atrial fibrillation (AF), second degree heart block (2° HB), and pauses greater than 3 seconds.
  • determining dominant positive and negative trends comprises: (a) determining whether a slope at each point, is positive or negative; (b) joining consecutive points with the same slope directionality to form segments; (c) merging short segments into longer, more dominant trends; and (d) defining dominant positive and negative trends as a trend with a longest consecutive number of data points with positive and negative slopes respectively.
  • a processor configured for receiving input from a continuous electrocardiographic waveform data stream acquired from one or more electrodes; and a memory storing programming executable on the processor; wherein when executed the programming performs one or more steps comprising:(a) filtering the data stream to reduce noise and remove non-physiologic information obtained from artefactual measurements; (b) deriving an averaged beat time series of electrocardiographic parameters from a signal averaged beat; (c) identifying aberrant jumps in the averaged beat time series, which arise from abrupt changes in electrocardiographic morphology or changes in electrode placement;(d) breaking up the time series into segments at said jumps and merging like segments deemed continuous; (e) selecting varying time windows from the resultant
  • determining dominant positive and negative trends for each of the measured parameters (g) for each dominant trend, the trend duration, calculating maximal change in the ECG parameter value, and slope of trend, and comparing said value and slope to the respective value and slip in the prior time window of the same length; (h) generating a time series of all inter-beat (RR) intervals from the continuous electrocardiographic waveform data, and evaluating the time series for the presence of arrhythmias (atrial fibrillation, second degree heart block, pause) within a variety of time windows, wherein presence of a new arrhythmia is determined by comparing presence of an arrhythmia compared to prior time windows; (i) combining variables measured as part of steps (g) and (h) a random forest predictive model configured to provide a risk of cardiac arrest; and (j) selecting a threshold for a risk score is selected, above which a patient is predicted to be high risk for impending cardiac arrest.
  • RR inter-beat
  • dominant trends are determined by performing steps comprising:(a)at each point, determining slope of change (positive or negative) is by performing robust linear regression using a number of points centered on this point, wherein the number of points utilized varies by window size;(b)merging short segments into longer, more dominant trends; and(c)defining a dominant trend for a particular window length as the trend in a direction (positive or negative) with the longest duration.
  • a set refers to a collection of one or more objects.
  • a set of objects can include a single object or multiple objects.
  • the terms “substantially” and “about” are used to describe and account for small variations.
  • the terms can refer to instances in which the event or circumstance occurs precisely as well as instances in which the event or circumstance occurs to a close approximation.
  • the terms can refer to a range of variation of less than or equal to ⁇ 10% of that numerical value, such as less than or equal to ⁇ 5%, less than or equal to ⁇ 4%, less than or equal to ⁇ 3%, less than or equal to ⁇ 2%, less than or equal to ⁇ 1 %, less than or equal to ⁇ 0.5%, less than or equal to ⁇ 0.1 %, or less than or equal to ⁇ 0.05%.
  • substantially aligned can refer to a range of angular variation of less than or equal to ⁇ 10°, such as less than or equal to ⁇ 5°, less than or equal to ⁇ 4°, less than or equal to ⁇ 3°, less than or equal to ⁇ 2°, less than or equal to ⁇ 1 °, less than or equal to ⁇ 0.5°, less than or equal to ⁇ 0.1 °, or less than or equal to ⁇ 0.05°.
  • range format is used for convenience and brevity and should be understood flexibly to include numerical values explicitly specified as limits of a range, but also to include all individual numerical values or sub-ranges encompassed within that range as if each numerical value and sub-range is explicitly specified.
  • a ratio in the range of about 1 to about 200 should be understood to include the explicitly recited limits of about 1 and about 200, but also to include individual ratios such as about 2, about

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Pathology (AREA)
  • Animal Behavior & Ethology (AREA)
  • Data Mining & Analysis (AREA)
  • Primary Health Care (AREA)
  • Cardiology (AREA)
  • Veterinary Medicine (AREA)
  • Epidemiology (AREA)
  • Physics & Mathematics (AREA)
  • Biophysics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Physiology (AREA)
  • Databases & Information Systems (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Business, Economics & Management (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychiatry (AREA)
  • Signal Processing (AREA)
  • Business, Economics & Management (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

L'invention concerne un système et une méthode pour prédire l'apparition d'arrêts cardiaques chez un patient sous surveillance en continu au moyen d'un électrocardiogramme par évaluation et comparaison de tendances dans des paramètres électrocardiographiques. Le système et la méthode peuvent être incorporés dans des modèles prédictifs qui peuvent être utilisés pour la prédiction d'arrêts cardiaques parmi la dissociation électromécanique (PEA)/l'asystole.
PCT/US2019/036850 2018-06-12 2019-06-12 Prédiction du risque d'arrêts cardiaques en hôpital à l'aide de données de surveillance de télésurveillance électrocardiographique en continu WO2019241444A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US201862683815P 2018-06-12 2018-06-12
US62/683,815 2018-06-12

Publications (1)

Publication Number Publication Date
WO2019241444A1 true WO2019241444A1 (fr) 2019-12-19

Family

ID=68843603

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2019/036850 WO2019241444A1 (fr) 2018-06-12 2019-06-12 Prédiction du risque d'arrêts cardiaques en hôpital à l'aide de données de surveillance de télésurveillance électrocardiographique en continu

Country Status (1)

Country Link
WO (1) WO2019241444A1 (fr)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113712564A (zh) * 2020-05-12 2021-11-30 深圳市科瑞康实业有限公司 一种心电信号分类设备和方法
WO2022184570A1 (fr) * 2021-03-05 2022-09-09 Koninklijke Philips N.V. Procédé et système d'intégration interprétable de tendances physiologiques et d'écarts de référence pour une aide à la décision clinique
CN115858632A (zh) * 2023-02-27 2023-03-28 青岛华芯晶电科技有限公司 一种氧化镓晶片检测装置的检测方法以及数据处理方法
EP4000516A4 (fr) * 2020-04-16 2023-08-02 Seknova Biotechnology Co., Ltd. Système de mesure d'informations biologiques et procédé de correction d'informations biologiques

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090076403A1 (en) * 2007-09-14 2009-03-19 Bruce Hopenfeld Waveform feature value averaging system and methods for the detection of cardiac events
US20100079292A1 (en) * 1997-01-27 2010-04-01 Lawrence A. Lynn Microprocessor system for the analysis of physiologic and financial datasets
US20100099995A1 (en) * 2008-10-16 2010-04-22 Jie Lian Method and apparatus for ectopic beat detection
US20160135706A1 (en) * 2014-11-14 2016-05-19 Zoll Medical Corporation Medical Premonitory Event Estimation
WO2016183683A1 (fr) * 2015-05-20 2016-11-24 University Health Network Procédé et système pour évaluer des composants qrs et le risque d'arythmies ventriculaires
US20170135577A1 (en) * 2014-04-25 2017-05-18 Texas State University Health Assessment via Eye Movement Biometrics

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100079292A1 (en) * 1997-01-27 2010-04-01 Lawrence A. Lynn Microprocessor system for the analysis of physiologic and financial datasets
US20090076403A1 (en) * 2007-09-14 2009-03-19 Bruce Hopenfeld Waveform feature value averaging system and methods for the detection of cardiac events
US20100099995A1 (en) * 2008-10-16 2010-04-22 Jie Lian Method and apparatus for ectopic beat detection
US20170135577A1 (en) * 2014-04-25 2017-05-18 Texas State University Health Assessment via Eye Movement Biometrics
US20160135706A1 (en) * 2014-11-14 2016-05-19 Zoll Medical Corporation Medical Premonitory Event Estimation
WO2016183683A1 (fr) * 2015-05-20 2016-11-24 University Health Network Procédé et système pour évaluer des composants qrs et le risque d'arythmies ventriculaires

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP4000516A4 (fr) * 2020-04-16 2023-08-02 Seknova Biotechnology Co., Ltd. Système de mesure d'informations biologiques et procédé de correction d'informations biologiques
CN113712564A (zh) * 2020-05-12 2021-11-30 深圳市科瑞康实业有限公司 一种心电信号分类设备和方法
CN113712564B (zh) * 2020-05-12 2023-09-01 深圳市科瑞康实业有限公司 一种心电信号分类设备和方法
WO2022184570A1 (fr) * 2021-03-05 2022-09-09 Koninklijke Philips N.V. Procédé et système d'intégration interprétable de tendances physiologiques et d'écarts de référence pour une aide à la décision clinique
CN115858632A (zh) * 2023-02-27 2023-03-28 青岛华芯晶电科技有限公司 一种氧化镓晶片检测装置的检测方法以及数据处理方法
CN115858632B (zh) * 2023-02-27 2023-05-12 青岛华芯晶电科技有限公司 一种氧化镓晶片检测装置的检测方法以及数据处理方法

Similar Documents

Publication Publication Date Title
US9456762B2 (en) Atrial fibrillation detection and associated methods
US20220015711A1 (en) System and method for automated analysis and detection of cardiac arrhythmias from electrocardiograms
WO2019241444A1 (fr) Prédiction du risque d'arrêts cardiaques en hôpital à l'aide de données de surveillance de télésurveillance électrocardiographique en continu
US9713431B2 (en) Method and apparatus for detection of heartbeat characteristics
Petrėnas et al. Low-complexity detection of atrial fibrillation in continuous long-term monitoring
EP2944251B1 (fr) Procédé, dispositif et support lisible par ordinateur de détermination de niveau de confiance hr ambulatoire basé sur un algorithme classificateur de rythme à trois voies
US8666483B2 (en) System for cardiac medical condition detection and characterization
US7537569B2 (en) Method and apparatus for detection of tachyarrhythmia using cycle lengths
EP1706031B1 (fr) Surveillance cardiaque
EP1219237B1 (fr) Procédé et dispositif de détection de la fibrillation auriculaire
US9724007B2 (en) System and method for detecting a characteristic in an ECG waveform
US8838224B2 (en) Method, apparatus and computer program product for predicting ventricular tachyarrhythmias
US20120302903A1 (en) System for Cardiac Impairment Detection Based on Signal Regularity
EP2259721B1 (fr) Système de détection de l'état du coeur
Do et al. Usefulness of trends in continuous electrocardiographic telemetry monitoring to predict in-hospital cardiac arrest
Lee et al. A real-time abnormal beat detection method using a template cluster for the ECG diagnosis of IoT devices
CN104755025A (zh) 用于识别高血糖的方法和设备
EP2846687A2 (fr) Système et méthode pour la stratification du risque sur la base d'une analyse non linéaire dynamique et d'une comparaison de repolarisation cardiaque avec d'autres signaux physiologiques
CN115770054A (zh) 心电信号处理方法、便携式心电采集设备和存储介质
Anisimov et al. Extrasystole Episodes Detection Algorithm in Real Time Mode
Do Development of a Risk Prediction Model for In-Hospital Cardiac Arrests Using Continuous Electrocardiographic Telemetry Monitoring Data
Voss et al. Monitoring in cardiovascular disease patients by nonlinear biomedical signal processing
De Melis The role of insertable loop recorders: Current perspectives and future possibilities
CARRARA et al. Cardiac rhythm classification and mortality risk assessment from RR intervals using non-linear dynamics indexes
Gopalan Analysis of ECG signals and its application in prediction of sudden cardiac death

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19820499

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19820499

Country of ref document: EP

Kind code of ref document: A1