Method and apparatus for adaptively detecting acute hypotensive episode events in a patient - Patents.com

Abstract

The present disclosure provides a method and apparatus for adaptively detecting an acute hypotensive episode (AHE) event in a patient, the method including processing, by a processor, a signal related to the patient to determine (101) at least one mean arterial pressure (MAP) threshold for the patient, and detecting, by the processor, an AHE event in the patient based on the at least one MAP threshold (102).

Description

[Technical field]

[0001] The present invention relates broadly, but not exclusively, to methods and apparatus for adaptively detecting acute hypotensive episode (AHE) events in a patient. [Background technology]

[0002] Hypotension is a sign of cardiovascular problems. Severe hypotension is common in patients undergoing surgery and in critically ill patients. In patients undergoing major surgical procedures, intraoperative hypotension is commonly associated with general anesthesia. It is also associated with adverse postoperative complications.

[0003] The duration and severity of an acute hypotensive episode (AHE) event is a critical event for the patient related to organ hypoperfusion and dysfunction, which can lead to irreversible shock, which represents the beginning of a pathway that may lead to death.

[0004] AHE events require effective and rapid intervention. In other words, AHE events must be treated reactively after hypotension values ​​have already occurred. If detected in time, appropriate intervention can significantly reduce the risk to the patient. Summary of the Invention [Problem to be solved by the invention]

[0005] To date, monitoring systems in operating rooms or intensive care units allow for processing and continuous display of numerical and waveform blood pressure data in real time. However, management of AHE events mainly consists of reactive treatment with vasopressors and fluids after hypotensive values ​​have already occurred. In particular, clinicians react to absolute blood pressure (e.g., mean arterial pressure (MAP)) thresholds or short-term blood pressure changes to detect AHE events in patients and treat patients reactively to avoid severe or sustained hypotensive events. However, hypotensive values ​​typically occur late in the process of hemodynamic instability, i.e., when the overall cardiovascular dynamics have already significantly changed and compensatory mechanisms have been exhausted.

[0006] Furthermore, predicting AHE events before they become clinically evident allows clinicians to preemptively treat the event, thereby reducing the duration and severity of hypotension. Here, AHE events can be predicted on the scale of minutes before they occur based on blood pressure waveforms using machine learning algorithms and strict thresholds of blood pressure (e.g., MAP). However, because cardiovascular physiology is highly interdependent, functions within a complex network, and is influenced by compensation mechanisms within each individual patient, it is difficult to analyze the complex cardiovascular systems of different patients using a single strict blood pressure threshold.

[0007] Therefore, there is a need to provide a personalized MAP threshold for each patient for early identification of AHE events or prediction of AHE events in patients before they become clinically evident, which would allow clinicians to apply preemptive treatment strategies to reduce the occurrence, duration, and severity of AHE events in patients.

[0008] Disclosed herein are embodiments of devices and methods for adaptively detecting AHE events in a patient that address one or more of the above problems.

[0009] Furthermore, other desirable features and characteristics will become apparent from the following detailed description and the appended claims, taken in conjunction with the accompanying drawings and this background of the disclosure. [Means for solving the problem]

[0010] In a first aspect, the present disclosure provides a method for adaptively detecting an acute hypotensive episode (AHE) event in a patient, the method including: processing, by a processor, a signal related to the patient to determine at least one mean arterial pressure (MAP) threshold for the patient; and detecting, by the processor, an AHE event in the patient based on the at least one MAP threshold.

[0011] In a second aspect, the present disclosure provides an apparatus for adaptively detecting an AHE event in a patient, comprising at least one processor and at least one memory including computer program code, the at least one memory and the computer program code configured to cause the apparatus, using the at least one processor, to perform at least a method according to the first aspect.

[0012] Further benefits and advantages of the disclosed embodiments will become apparent from the specification and drawings, in which the benefits and / or advantages may be obtained individually through the various embodiments and features of the specification and drawings, and not all need be provided in order to obtain one or more of such benefits and / or advantages. Effect of the Invention

[0013] In accordance with the present disclosure, it is possible to provide devices and methods for adaptively detecting AHE events in a patient that address one or more of the above problems.

[0014] The accompanying drawings, in which like reference numbers refer to identical or functionally similar elements throughout the different views, and which, together with the following detailed description, are incorporated in and form a part of this specification, illustrate various embodiments by way of non-limiting examples only and serve to explain various principles and advantages according to the present embodiments.

[0015] Embodiments of the present invention will be better understood and readily apparent to those skilled in the art from the following description, given by way of example only, in conjunction with the drawings in which: [Brief description of the drawings]

[0016] [Figure 1] 1 shows a flowchart illustrating a method for adaptively detecting an acute hypotensive episode (AHE) event in a patient, according to various embodiments of the present disclosure.

[0017] [Diagram 2] FIG. 1 shows a block diagram illustrating an apparatus for adaptively detecting an AHE event in a patient, according to various embodiments of the present disclosure.

[0018] [Diagram 3] 3 shows a block diagram 300 illustrating a signal source 301 and a personalized AHE definition module 304 for determining a patient's mean arterial pressure (MAP) threshold, according to one embodiment of the disclosure.

[0019] [Figure 4] 1 is a flow chart illustrating a portion of a method for determining a MAP threshold for a patient according to one embodiment of the present disclosure.

[0020] [Diagram 5] 1 shows a graph of average θ versus time t illustrating two different states s of a two-state model used to determine a patient's MAP threshold in accordance with an embodiment of the present disclosure.

[0021] [Figure 6A] 1A and 1B show a MAP waveform and a processed MAP waveform, respectively, according to one embodiment of the present disclosure.

[0022] [Figure 6B] 1A and 1B show a MAP waveform and a processed MAP waveform, respectively, according to one embodiment of the present disclosure.

[0023] [Figure 7] 1 illustrates a processed MAP waveform and personalized AHE parameters according to one embodiment of the present disclosure.

[0024] [Figure 8] 11 is a flow chart illustrating another portion of a method for determining a MAP threshold for a patient according to an embodiment of the present disclosure.

[0025] [Figure 9A] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0026] [Figure 9B] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0027] [Figure 10A] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0028] [Figure 10B] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0029] [Figure 10C] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure. [Figure 11A] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0030] [Figure 11B] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0031] [Figure 11C]1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure. [Figure 12A] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0032] [Figure 12B] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0033] [Figure 12C] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure. [Figure 13A] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0034] [Figure 13B] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0035] [Figure 13C] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0036] [Figure 13D] 1 illustrates various MAP waveforms and AHE events detected therein, in accordance with various embodiments of the present disclosure.

[0037] [Figure 14] FIG. 1 shows a block diagram of a hypotension prediction module for adaptively detecting and predicting AHE events in a patient, according to one embodiment of the present disclosure.

[0038] [Figure 15] 1 shows a flowchart illustrating a portion of a method for pre-processing a signal and extracting signal features to determine a prediction of an AHE event in a patient, according to one embodiment of the present disclosure.

[0039] [Figure 16] FIG. 16 shows a block diagram 1600 illustrating the inputs and outputs of a stacking ensemble model-based predictive modeling engine 1601 used to adaptively detect and predict AHE events in patients, according to one embodiment of the present disclosure.

[0040] [Figure 17A] 1 illustrates various input data used to adaptively detect and predict AHE events in a patient, according to various embodiments of the present disclosure.

[0041] [Figure 17B] 1 illustrates a MAP waveform used as an input for adaptively detecting and predicting AHE events in a patient, according to one embodiment of the present disclosure.

[0042] [Figure 18] 1 illustrates features extracted from signals for multiple patients used for model training according to one embodiment of the present disclosure.

[0043] [Figure 19] FIG. 1 shows a diagram illustrating a process for extracting at least one feature in a signal used to adaptively detect and predict AHE events in a patient, according to one embodiment of the present disclosure.

[0044] [Figure 20] FIG. 1 shows a block diagram illustrating an apparatus for detecting and predicting an AHE event in a patient, according to one embodiment of the present disclosure.

[0045] [Figure 21] FIG. 1 shows a flow diagram illustrating a process for detecting and predicting an AHE event in a patient, according to one embodiment of the present disclosure.

[0046] [Figure 22] 1 shows a schematic diagram of an exemplary computing device suitable for use in performing the method of FIG. 1 and implementing the apparatus of FIG. 2. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0047] (Terminology explanation) Patient - The patient may be a user or a person at home, who is being transported to a hospital, e.g., an accident and emergency department, via, e.g., an ambulance or medically equipped vehicle, or who is currently hospitalized, and who requires that his or her mean arterial pressure (MAP) threshold be determined, that AHE events be adaptively detected based on his or her MAP threshold, and that subsequent AHE events be predicted.

[0048] Signal - The signal may be one or a combination of hemodynamic data such as systolic blood pressure (BPSys or SBP), heart rate (HR), respiratory rate (RESP), MAP, diastolic blood pressure (BPDia or DBP), pulse pressure (PP), blood oxygen saturation (SPO2), and other clinical data stored or received, for example, from equipment or measurements made in the intensive care unit (ICU), emergency department (ED), input made by occupational therapy (OT), patient monitoring unit (PMU), and / or patient data from perioperative department, ambulance or medically equipped vehicle, data from hospital ward, and / or electronic medical record (EMR). Such data may be obtained directly or indirectly from the patient, invasively or non-invasively. In various embodiments below, MAP data or MAP value is used as a signal for detection and prediction of AHE in a patient. It is to be understood that the method and apparatus may be configured to achieve similar results using other hemodynamic or clinical data.

[0049] Mean Arterial Pressure (MAP) Threshold - The MAP threshold is used to compare the value of the MAP waveform for a patient to detect the patient's AHE event. Conventionally, a fixed MAP threshold of 60mmHg or 65mmHg is defined, and if any patient's MAP waveform has a MAP value lower than the fixed MAP threshold 90% of the time within a 30-minute period, an AHE event is detected. In contrast, the following various embodiments provide personalized AHE definitions for different patients based on personalized MAP thresholds. In other words, different MAP thresholds are defined for different patients for the detection of each personalized AHE event. The MAP thresholds are calculated using the respective baselines and the variability of their own MAP waveforms. Such MAP threshold(s) are MAP values ​​offset from the MAP waveform baseline, and the amount of offset depends on the variability of the patient's MAP waveform. Subject to other determining factors, a personalized AHE is detected when the patient's MAP waveform crosses a threshold(s), i.e., when the patient's MAP switches from one state to another, e.g., from an out-of-control (OOC) state to an in-control or baseline state, or vice versa.

[0050] Baseline - The baseline of a signal refers to a value, a derived value (e.g., an average or an average of multiple values), a point or portion of a signal that is used to compare against the remaining values, measurements or portions of a signal to determine the deviation of the signal and MAP threshold(s) for a patient. In one example, the baseline is taken directly or derived from a first portion of the signal, e.g., values ​​or measurements within the first seconds, minutes, and hours of the signal. In another example, the baseline of a signal can be set automatically upon successive detection of similar measurements at such values, points, or portions of the signal, or manually by an operator or analyst. In various processes and calculations below, the baseline may be referred to as the baseline measurement or θ0.

[0051] Variance - The variance of a signal refers to the variance of the signal's values ​​or measurements. In the various procedures and calculations below, the variance is defined as var(zt ) is called.

[0052] Features - Features relate to attributes of a signal, typically a sample or segmented portion of a signal. Such features can be calculated and represented by feature data and compared with other signals to determine if they exhibit similar features, attributes, or feature data. Examples of features include maximum power, average power, minimum power, maximum frequency, mean frequency, median frequency, minimum frequency, maximum of segmented data, average of segmented data, minimum of segmented data, variance of segmented data, skewness of segmented data, kurtosis of segmented data, RMS value of segmented data, etc. In one example, a prediction of the occurrence of an AHE event in a patient can be determined by comparing feature data about the patient with feature data of other patients with / without AHE.

[0053] Embodiments of the present invention will now be described, by way of example only, with reference to the drawings in which like reference numbers and letters indicate similar elements or equivalents.

[0054] Some portions of the description which follow are presented explicitly or implicitly in terms of algorithms and functional or symbolic representations of operations on data within a computer memory. These algorithmic descriptions and functional or symbolic representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. An algorithm is here, generally, conceived to be a self-consistent sequence of steps leading to a desired result. These steps require physical manipulations of physical quantities, such as electrical, magnetic or optical signals capable of being stored, transferred, combined, compared, and otherwise manipulated.

[0055] Unless otherwise indicated, and as will become apparent below, descriptions utilizing terms such as "receive," "calculate," "determine," "update," "generate," "initialize," "output," "receive," "retrieve," "identify," "distribute," "authenticate," and the like throughout this specification will be understood to refer to the actions and processes of a computer system or similar electronic device that manipulates and converts data represented as physical quantities within a computer system into other data similarly represented as physical quantities within a computer system or other information storage, transmission, or display device.

[0056] This specification also discloses an apparatus for performing the operations of the methods. Such an apparatus may be specially constructed for the required purposes, or may comprise a computer or other device selectively activated or reconfigured by a computer program stored in the computer. The algorithms and displays presented herein are not inherently related to any particular computer or other apparatus. Various machines can be used with programs in accordance with the teachings of this specification. Alternatively, the construction of more specialized apparatus to perform the required method steps may be appropriate. The structure of the computer will be apparent from the following description.

[0057] In addition, this specification also implicitly discloses a computer program in that it is clear to a person skilled in the art that the individual steps of the methods described herein may be implemented by computer code. The computer program is not intended to be limited to any particular programming language and its implementation. It will be understood that a variety of programming languages ​​and their coding may be used to implement the teachings of the present disclosure contained herein. Furthermore, the computer program is not intended to be limited to any particular control flow. There are many other variations of the computer program that can use different control flows without departing from the spirit or scope of the present invention.

[0058] Furthermore, one or more of the steps of the computer program may be performed in parallel rather than sequentially. Such a computer program may be stored on any computer-readable medium. The computer-readable medium may include storage devices such as magnetic or optical disks, memory chips, or other storage devices suitable for interfacing with a computer. The computer-readable medium may also include hardwired media, as exemplified in the Internet system, or wireless media, as exemplified in the GSM mobile telephone system. The computer program, when loaded and executed on such a computer, effectively produces an apparatus that implements the steps of the preferred method.

[0059] EXEMPLARY EMBODIMENTS

[0060] Various embodiments of the present disclosure relate to methods and apparatus for adaptively detecting AHE events in a patient.

[0061] Traditionally, prediction of AHE events is based on the definition of AHE as a MAP value of 60 mmHg or less 90% of the time within a 30-minute time frame. This fixed and strict threshold of 60 mmHg is applied to all patients. Furthermore, prediction models are based on binary class models such as binary classification or logistic regression, supervised learning using neural networks.

[0062] Recently, this algorithm has been developed as a feature of the EV1000 and Hemosphere clinical monitoring platforms (Edwards Lifesciences). In particular, a hypotension prediction index (GPI) value ranging from 0 to 100 is calculated every 20 seconds and displayed on the monitor. The HPI value represents the probability that the patient will experience a hypotensive event, defined as a MAP value below 65 mmHg for at least 1 minute. An audible alarm is activated when the HPI value exceeds 85.

[0063] There are two main problems that have been identified with traditional and existing AHE detection solutions: 1) the use of traditional definitions of hypotension based on a single MAP threshold, e.g., 60mmHg or 65mmHg, which does not take into account the individual patient; 2) the use of binary classification or logistic regression in predictive systems to provide only the probability of occurrence of AHE at a given onset time and yes / no information.

[0064] Specifically, with regard to the first problem statement, some people do well with a systolic of 90 mmHg, others are hypotensive at 100 mmHg or less, while elderly people with atherosclerosis and hypertension may be hypotensive at, for example, 120 mmHg systolic. For patients with a history of hypertension, a reduction in MAP of more than 40 mmHg is considered hypotensive, even if the pressure is still above 60 mmHg. However, hypertensive patients or elderly people require higher MAP levels.

[0065] Furthermore, MAP is calculated based on the following formula (1):

number

[0066] Autoregulatory thresholds vary between organs and individuals. Fluids (in shock conditions with reduced venous return) and vasopressors are required to increase MAP and obtain the target MAP. High target MAP values ​​require higher loads of vasopressors and may induce excessive arterial systemic vasoconstriction, which may induce organ ischemia. Importantly, individual patients have a specific baseline of MAP and they should not be considered under the same group. This is further supported by the European Society of Intensive Care Medicine, who provide a new description of their latest consensus on individualizing blood pressure targets when treating shock.

[0067] Regarding the second problem statement, existing prediction models only provide yes / no information of the occurrence of AHE events in patients, but do not mention the severity level of AHE events or other important parameters for determining severe or persistent hypotension based on MAP values, and clinicians can only respond to absolute blood pressure thresholds for short-term blood pressure changes.

[0068] Therefore, there is a need to adaptively detect AHE events in patients by providing a personalized definition of AHE relative to MAP thresholds and integration with continuous hemodynamic waveform data to improve current models and develop new hypotension prediction models.

[0069] In various embodiments below, the present disclosure describes a method, or an apparatus having a detection module configured to implement the method, for defining a patient-specific or personalized AHE on a patient's MAP waveform and detecting a patient's AHE event based on a patient-specific or personalized MAP threshold.

[0070] Advantageously, the present disclosure provides a personalized definition of hypotension based on variability in MAP values ​​to obtain an optimal target MAP threshold that can prevent higher hypertensive loads. The present disclosure also provides methods and devices including a multi-level prediction module configured to implement the method for determining the potential occurrence of a subsequent AHE event and the level of severity of the AHE event or hypotension.

[0071] The present disclosure also describes in various embodiments below a method for determining short- to long-term predictions of a patient's subsequent AHE events based on a patient-specific or personalized AHE on the patient's MAP waveform, and an apparatus with a multi-level prediction module configured to implement the method, which can provide predictions of hypotension from 5 minutes to 2 hours in advance, covering short-, mid- to long-term predictions that may be beneficial in preemptive treatment for different care settings.

[0072] The present disclosure further provides a method and apparatus for adaptively detecting AHE events in a patient that can be applied in intraoperative care, intensive care, and postoperative care settings (recovery wards) to monitor the patient's condition. Because such methods and apparatus utilize hemodynamic data from routine patient monitor units, they can be retrofitted to existing healthcare units and settings, and thus do not require any special and additional hardware devices to acquire hemodynamic or other clinical data. Overall, the method and apparatus for adaptively detecting AHE events in a patient described in the present disclosure can solve the above-mentioned problems in conventional AHE event detection systems.

[0073] 1 shows a flow chart 100 illustrating a method for adaptively detecting an acute hypotensive episode (AHE) event in a patient, according to various embodiments of the present disclosure. In step 101, a signal related to the patient is processed by a processor to determine at least one MAP threshold value related to the patient. In step 102, the processor detects an AHE event in the patient based on the at least one MAP threshold value.

[0074] 2 illustrates a block diagram 200 illustrating an apparatus 202 for adaptively detecting an AHE event in a patient, according to various embodiments of the present disclosure. In one implementation, the apparatus 202 can be generally described as a physical device comprising at least one processor and at least one memory containing computer program code. The at least one memory and the computer program code, together with the at least one processor, are configured to cause the physical device to perform the operations described in FIG.

[0075] In one implementation, the device 202 can receive signals such as MAP waveforms and other clinical data from a source 201, for example, from an instrument or measurement performed in an intensive care unit (ICU), an emergency department (ED), data entry by an occupational therapy (OT), data from a patient monitoring unit (PMU) and / or a perioperative department, data from an ambulance or medically equipped vehicle, a hospital ward, and / or an electronic medical record (EMR). The device 202 comprises a personalized AHE definition module 204 configured to determine at least one MAP threshold of the patient and define a personalized AHE based on the signals, and a hypotensive episode prediction module 206 configured to determine a prediction of a subsequent AHE event of the patient based on the signals.

[0076] In one implementation, the personalized AHE definition module 204 is configured to receive as input a MAP waveform of the patient and output a detection of an AHE event for the patient including the onset of the AHE event, the duration of the AHE event, and a personalized MAP threshold used for such detection based on the input. In one embodiment, the hypotensive episode prediction module 206 is configured to receive the MAP waveform and other clinical data and output a prediction of the occurrence of an AHE event as well as a severity level of the AHE event based on the input.

[0077] The device 202 may further comprise a monitoring module 210 configured to display signals from the source 201 and / or inputs and outputs of the personalized AHE definition module 204 and the hypotensive episode prediction module 206 on a user interface 211 to an operator of the device 202, such as a physician, medical staff, or analyst. The operator may also interact with the user interface 211 and input data into the device 202 via the monitoring module 210. Although the monitoring module 210 is shown in FIG. 2 as being part of the device 202, it should be understood that the monitoring module 210 may be part of a standalone device or another device and communicates with the device 202 through a connection. Such a connection may be wired, wireless (e.g., via NFC communication, Bluetooth, etc.), or over a network (e.g., the Internet).

[0078] The device may include a data storage 208 accessible by the device 202 for storing the signals from the source 201 and the inputs and outputs of the personalized AHE definition module 204 and the hypotensive episode prediction module 206. Although FIG. 2 shows the data storage as part of the device 202, it should be understood that the data storage 208 may not form part of the device 202 and communicates with the device via a connection. Such a connection may be wired, wireless (e.g., via NFC communication, Bluetooth, etc.), or via a network (e.g., the Internet) or via a cloud server connection. In another implementation, each of the modules 204, 206, 210 and the signal source 201 may include its own data storage for storing its input / output data and signals.

[0079] FIG. 3 illustrates a block diagram 300 showing a signal source 301 and a personalized AHE definition module 304 for determining a patient's mean arterial pressure (MAP) threshold, according to one embodiment of the disclosure.

[0080] The signal source 301 is a source of a patient's continuous MAP waveform measured from or conducted to the patient using devices including one or a combination of a patient monitoring unit 313 (e.g., a patient bedside monitor), an invasive blood pressure monitoring device 314 (e.g., an arterial blood pressure monitoring device, a central venous pressure monitoring device), and a non-invasive or minimally invasive blood pressure monitoring device 315 (e.g., a continuous blood pressure monitoring system by CNSystems). The source 301 may further comprise a data storage configured to store the device signals and measurements. In one implementation, the data storage 311 may be included in the personalized AHE definition module 304.

[0081] In one implementation, the personalized AHE definition module 304 may be generally described as a part of a physical device comprising at least one processor and at least one memory including computer program code. The at least one memory and computer program code, together with the at least one processor, are configured to cause the physical device or part to receive a signal from the source 301 and perform the operations in steps 321-324. In particular, the personalized AHE definition module 304 may be configured to perform the steps of receiving MAP waveform data for the patient directly from the signal source 301 and / or via the data storage 311 in step 321. The module 304 may be further configured to process the signal, such as pre-processing the signal and filling missing values ​​in the signal in step 322, initiating associated activity detection using change point theory in step 323, and determining the patient's MAP threshold and providing detection of the patient's AHE event, including onset and duration of the AHE event, in step 324. Data regarding the patient's MAP threshold and the detected AHE onset and duration are received from the source 301 and / or via the data storage 311. 301 The data can be transmitted to the data storage 311.

[0082] FIG. 4 shows a flow chart illustrating a portion of a method for determining a patient's MAP threshold according to an embodiment of the disclosure. In step 401, a step of obtaining pre-processed MAP waveform or time series data is performed. In step 402, a step of setting a baseline state time point is performed. In one example, the baseline state time point is set by assuming that the data is a two-state model with two distributions in the combination of baseline as the in-control state and physiological change as the out-of-control state. In step 403, a step of calculating the mean and standard deviation of the in-control state and the out-of-control state is performed. In step 404, a step of calculating an exponentially weighted moving average (EWMA) statistic and an EWMA statistic variance of the MAP time series data is performed. In step 405, a step of calculating a lower control limit and an upper control limit of the MAP time series data is performed. In step 406, a step of determining whether there is a state change from the baseline using the calculated control limits is performed. The process can then proceed to another portion of the method shown in FIG. 8.

[0083] To set the baseline state time point in the preprocessed MAP waveform or time series data, a two-state model is first considered, and the data is modeled as a combination of two normal distributions: one with mean θ0 and (co)variance Σ, and a second normal distribution with mean θ1 and (co)variance Σ.

[0084] 5 illustrates a graph of average θ versus time t showing two different states s of a two-state model used to determine a patient's MAP threshold, according to one embodiment of the disclosure. In this exemplary embodiment, two states are shown in FIG. 5, an in-control or baseline state and an out-of-control (OOC) state, using measurements θ0 and θ1 at times t (t=1,...,τ) before τ and t (t=t+1,...,n) after τ, respectively, and the current state value s t can be summarized using equation (2).

number

[0085] Mean Observed Value TIFF0007647995000003.tif1510 and current observation value x t can be calculated using the following formula:

number

number

[0086] The EWMA statistic z, which is a smoothed version of the MAP waveform or time series data t can be calculated using equation (5).

number

[0087] z t Each value of is the current observation x t and the previous value of the EWMA statistic. For simplicity, the starting value z0 is set equal to the baseline measurement θ0 (at t=0). Thus, equation (5) can also be written as:

number

[0088] 6A and 6B respectively show a MAP waveform 601 and a smoothed MAP waveform 602 according to one embodiment of the present disclosure. The smoothed MAP waveform 602 may result from processing the MAP waveform 601 according to the EWMA statistics and equations (5) and (6).

[0089] EWMA statistics variance var(z t ) is the EQMA statistic z t is the variance of the vector, and in the presence of white noise, it can be calculated based on the following equation (7):

number

[0090] The upper and lower control limits (UCL / LCL) are calculated by subtracting the EWMA statistic z from the baseline using the following formula: t It is defined based on the deviation.

number

[0091] Using a two-state model, z at any time t If V exceeds / below the control limit calculated from equation (8), then it is determined that there is a deviation from the baseline and the process is considered to have changed state from its baseline region.

[0092] When a change state or variation is detected, a step is performed to calculate exactly when the change occurred in order to estimate the unknown parameter τ, i.e., the change point (time point) at which the state change occurred. The zero-crossing method is one of several methods that can be utilized to estimate τ. Based on the "zero-crossing" method, the last time the process crosses θ0 is the estimate of τ.

[0093] 7 illustrates a processed MAP waveform 700 and personalized AHE parameters according to one embodiment of the present disclosure. In one example, the processed MAP waveform is derived using the two-state model and equations (2)-(8) shown above, and includes the baseline area, baseline measurement θ, EWMA statistic z, as shown in FIG. t (i.e., the processed MAP waveform), EWMA statistic variance var(z t Derivatives such as τ (i.e., the variance of the MAP waveform), UCL, LCL, out-of-control (OOC) area, and estimated τ are obtained in the processing.

[0094] In this embodiment, the first 60 minutes of the processed MAP waveform 700 is determined as the baseline region of the MAP waveform 700, and the baseline, i.e., the baseline measurement θ in this case, is the average of the MAP values ​​in the baseline region of 87.53 mmHg.

[0095] The UCL and LCL are calculated based on the baseline and variance of the MAP waveform 700. In this case, the variance is the EWMA statistic variance var(z t The OOC regions are set to regions where the MAP value is higher than the UCL or lower than the LCL, such as OOC region 1 and region 2.

[0096] Each estimated time point of a state shift from the baseline state to the OCC state can be determined using the "zero crossing" method by referencing the last time the processed MAP waveform crossed θ. In this embodiment, the estimated change points at which the change to OOC region 1 and OOC region 2 occurred are τ and τ, respectively, i.e., approximately 340 minutes and 480 minutes.

[0097] FIG. 8 shows a flow chart illustrating another part of a method for determining a patient's MAP threshold according to an embodiment of the present disclosure. In one example, after a state change from an in-control state to an out-of-control state is detected in step 406, step 801 of detecting a change point or change points to determine the start of the change(s) is performed precisely using a zero-crossing method. Step 802 is a step of determining whether the detected OOC region(s) of the processed signal (EWMA statistics) has a value lower than the LCL. If the value is lower than the LCL, step 804 is performed, otherwise step 803 is performed. Step 803 is a step of determining whether the detected OOC region(s) of the processed signal (EWMA statistics) has a value higher than the UCL, whether the UCL is lower than a pre-configured threshold, in this case the pre-configured threshold is 75 mmHg, and whether the mean value of the baseline is lower than a pre-configured threshold, in this case the pre-configured threshold is 65 mmHg. If all three conditions are met, step 804 is executed, otherwise step 805 is executed.

[0098] In step 804, a step of identifying each duration of the processed signal (EWMA statistic) that exceeds the control limit, i.e., the OOC region, is longer than a preconfigured duration, in this case the preconfigured duration is 15 minutes (step 806). If the duration is longer than the preconfigured duration, step 807 is performed, otherwise step 805 is performed.

[0099] In step 805, it is determined that a personalized AHE is not detected. Step 807 is to declare a personalized AHE parameter based on the detected change in the EWMA statistic, the onset of AHE, the duration of AHE, and the personalized AHE value, i.e., the MAP threshold defined for the patient. Detection of the patient's AHE event is then performed based on the personalized AHE / MAP threshold.

[0100] The following various embodiments illustrate MAP waveform evaluation, personalized AHE definition, and AHE event detection using the steps of the flowchart 800 of FIG. 8 and its accompanying description above. However, it will be understood by one of skill in the art that this is only one of many implementations. Different types of signals (e.g., HR and SPO2) and steps with different criteria can be used to determine personalized AHE definitions for different patients.

[0101] FIG. 9A shows a processed MAP waveform 901 with AHE detected using a conventional single threshold method, and FIG. 9B shows personalized AHE parameters of the processed MAP waveform 901 according to one embodiment of the present disclosure. In FIG. 9A, AHE is detected when the MAP data falls below a fixed MAP threshold of 60 mmHg. FIG. 9B shows the baseline region in the first 60 minutes of the same processed MAP waveform 901 and the UCL / LCL calculated based on the variability of the baseline and the processed MAP waveform 901. Note that the UCL is lower than the pre-configured threshold of 75 mmHg and the mean of the baseline is higher than the pre-configured threshold of 65 mmHg.

[0102] A small number of state change points, such as points 903-907, are detected. In particular, point 905 meets the criteria of having a MAP value lower than the LCL, and the duration of the state lasted longer than the preconfigured threshold duration of 15 minutes. Thus, a personalized AHE for the patient is defined at point 905 at 67.41 mmHg (LCL) in MAP. Two AHE events are detected in the area between points 905-906 and points 906-907, respectively, based on the personalized AHE, and their AHE parameters, such as the onset (points 905, 907) and duration of the AHE events, are declared and recorded. In this embodiment, the personalized AHE is detected earlier than the traditional single threshold. This can provide more time for clinicians to manage the patient for AHE.

[0103] FIG. 10A shows the patient's processed MAP waveform 1001 and personalized AHE parameters, which are identical to FIG. 9B. FIG. 10B shows a subsample 1002 of the processed MAP waveform 1001. A new baseline region in the first 15 minutes of the portion is determined, and the UCL / LCL is calculated based on the new baseline and the MAP value variation within the subsample. There is no state change from in-control to OOC beyond the LCL. A few state change points between in-control and OOC beyond the UCL are detected in the waveform 1002, such as points 1008-1011. Note, however, that the UCL is higher than the preconfigured threshold of 75 mmHg and the mean value of the baseline is higher than the preconfigured threshold of 65 mmHg. None of the state change points meet the criteria set in FIG. 8. Therefore, personalized AHE was not detected.

[0104] 10C shows personalized AHE parameters of a processed MAP waveform 1003 for a patient who has already lost a large amount of blood due to an accident and has had a hypotensive episode, e.g., AHE1, before arriving at the Accident and Emergency Department of a hospital. FIG. 10C shows the baseline area in the first 60 minutes of the processed MAP waveform 1003 and the UCL / LCL calculated based on the variability of the baseline and the processed MAP waveform 1003. Note that the UCL is below the pre-configured threshold of 75 mmHg and the mean of the baseline is below the pre-configured threshold of 65 mmHg.

[0105] In the MAP waveform 1003, three state change points are detected, such as points 1005-1007. In particular, point 1005 meets the criteria of having a MAP value higher than the UCL, a baseline average MAP less than 65 mmHg, and a state duration sustained longer than a preconfigured threshold duration of 15 minutes. Thus, a personalized AHE for the patient is defined at point 1005 at 64.9 mmHg (LCL) in the MAP. AHE events are detected in the area between points 1006-1007 based on the personalized AHE, and their AHE parameters, such as the onset (point 1006) and duration of the AHE event, are declared and recorded. In this embodiment, the personalized AHE is detected earlier than another hypotensive episode, e.g., AHE2.

[0106] 11A shows personalized AHE parameters of a patient's processed MAP waveform 1101 according to one embodiment. FIG. 11A shows the baseline area in the first 60 minutes of the processed MAP waveform 1101 and the UCL / LCL calculated based on the variability of the baseline and the processed MAP waveform 1101. Note that the UCL is higher than the pre-configured threshold of 75 mmHg and the mean of the baseline is higher than the pre-configured threshold of 65 mmHg.

[0107] In the MAP waveform 1101, three state change points are detected, such as points 1104-1106. In particular, point 1106 meets the criteria of having a MAP value lower than the LCL and a duration longer than a preconfigured threshold duration of 15 minutes. Thus, a personalized AHE for the patient is defined at point 1106 at 71.63 mmHg (LCL) in the MAP.

[0108] FIG. 11B shows a subsample 1102 of a processed MAP waveform 1101. A new baseline region in the first 15 minutes of the portion is determined, and a UCL / LCL is calculated based on the new baseline and the variance of the MAP values ​​within that subsample. Note that the UCL is higher than a preconfigured threshold of 75 mmHg, and the mean value of the baseline is higher than a preconfigured threshold of 65 mmHg. A state change point from an in-control state to an OOC state that exceeds the UCL is detected at point 1107 where the criterion is not met. A state change from an in-control state to an OOC state that exceeds the LCL is not detected. Thus, a personalized AHE was not detected.

[0109] 11C shows personalized AHE parameters of a processed MAP waveform 1103 for a patient who has already had a hypotensive episode due to a large amount of blood loss due to an accident before arriving at the Accident and Emergency Department of a hospital. FIG 11C shows the baseline area in the first 60 minutes of the processed MAP waveform 1103 and the UCL / LCL calculated based on the variability of the baseline and the processed MAP waveform 1103. Note that the UCL is below the pre-configured threshold of 75 mmHg and the mean of the baseline is below the pre-configured threshold of 65 mmHg.

[0110] The state change point from an in-control state to an OOC state above the UCL is detected at point 1108 on the MAP waveform 1003. Specifically, point 1108 meets the criteria of having a MAP value higher than the UCL, a baseline average MAP less than 65 mmHg, and a sustained duration of the state greater than a pre-configured threshold duration of 15 minutes. Thus, a personalized AHE for the patient is defined at point 1108 on the MAP at 64.9 mmHg (UCL). In this embodiment, the personalized AHE is detected earlier than with conventional methods.

[0111] 12A shows personalized AHE parameters of a patient's processed MAP waveform 1201 according to one embodiment. FIG. 12A shows the baseline area in the first 60 minutes of the processed MAP waveform 1201 and the UCL / LCL calculated based on the variability of the baseline and the processed MAP waveform 1201. Note that the UCL is higher than the pre-configured threshold of 75 mmHg and the mean of the baseline is higher than the pre-configured threshold of 65 mmHg.

[0112] A few state change points, such as points 1204-1209, are detected in the MAP waveform 1201. In particular, point 1209 meets the criteria of having a MAP value lower than the LCL and a duration longer than a preconfigured threshold duration of 15 minutes. Thus, a personalized AHE for the patient is defined at point 1106 at 65.65 mmHg (LCL) in the MAP.

[0113] FIG. 12B shows a subsample 1202 of the processed MAP waveform 1201. A new baseline region in the first 15 minutes of the portion is determined, and the UCL / LCL is calculated based on the new baseline and the variance of the MAP value within that subsample. Note that the UCL is higher than the preconfigured threshold of 75 mmHg, and the mean value of the baseline is higher than the preconfigured threshold of 65 mmHg. Six state change points 1210-1215 are detected, but none of them meet the criteria set in FIG. 8. For example, point 1210 has a MAP value lower than the LCL, but does not last longer than the preconfigured duration of 15 minutes. Therefore, no personalized AHE was detected from FIG. 12B.

[0114] 12C shows personalized AHE parameters of a processed MAP waveform 1203 for a patient who has already had a hypotensive episode, having lost a large amount of blood due to an accident, before arriving at the Accident and Emergency Department of a hospital. FIG. 12C shows the baseline area in the first 60 minutes of the processed MAP waveform 1203, and the UCL / LCL calculated based on the variability of the baseline and the processed MAP waveform 1203. Note that the UCL is below the pre-configured threshold of 75 mmHg, and the mean of the baseline is below the pre-configured threshold of 65 mmHg.

[0115] A significant state change point from an in-control state above the UCL to an OOC state is detected at point 1216 on the MAP waveform 1203. In particular, point 1216 meets the criteria of having a MAP value higher than the UCL and a state duration that lasts longer than a preconfigured threshold duration of 15 minutes. Thus, a personalized AHE for the patient is defined at point 1216 on the MAP at 65.65 mmHg (UCL). In this embodiment, the personalized AHE is detected at an earlier time point than conventional methods.

[0116] 13A-13D show four exemplary MAP waveforms of a patient without an AHE event. AHE was not detected in all four exemplary MAP waveforms based on the steps of flowchart 800 of FIG. 8 and its accompanying description.

[0117] 14 illustrates a block diagram of a hypotension prediction module 1404 for adaptively detecting and predicting AHE events in a patient according to an embodiment of the present disclosure. In one implementation, the hypotension prediction module 1404 can be generally described as a part of a physical device comprising at least one processor and at least one memory including computer program code. The at least one memory and the computer program code, together with the at least one processor, are configured to cause the physical device or part to perform operational steps 1405-1407 for model training and 1408-1409 for management of a new patient based on the trained model.

[0118] The data storage 1401 is configured to receive signals relating to new patients, such as waveform data and other clinical data 1403, Pressure Measurement Module 1404 In addition, such data storage 1401 may also include Pressure Measurement Module 1404 and the personalized AHE definition module 1402. The data stored therein may be shared between the personalized AHE definition module 1402 to determine a personalized AHE and detect AHE events in the data. The personalized AHE and other parameters may be recorded in data storage and optionally used by the hypotension prediction module 1404 to generate its output.

[0119] For model training, the hypotension prediction module 1404 is configured to perform the steps of acquiring signals such as blood pressure waveforms and other clinical data for a number of patients (e.g., existing patients) in step 1405, executing software algorithms to perform pre-processing and feature extraction of the waveform data and clinical data for the existing patients in step 1406, and generating a predictive model based on the extracted features of the waveform data and clinical data and a corresponding feature dataset (hereinafter referred to as a "training dataset" or a "test dataset") in step 1407.

[0120] For example, to manage a new patient after a personalized AHE has just been determined, the hypotension prediction module is configured to execute, in step 1409, a software algorithm that performs pre-processing and feature extraction of the received waveform data and other clinical data 1403 relating to the new patient, and then execute an AHE prediction classifier in which the extracted features and feature data from the new patient's signal 1403 (hereinafter referred to as the "new data set") are classified using the predictive model generated in step 1407 to determine a prediction of a subsequent AHE event for the new patient.

[0121] As shown in block 1410, a prediction of the occurrence of a subsequent AHE event for a new patient, as well as its severity level of the AHE event, can be determined.

[0122] The hypotension prediction module 1404 can be configured to display the new patient signal 1403, the inputs and outputs of the hypotension prediction module 1404 on a user interface 1411 to an operator of the device 202, such as a doctor, medical staff, or analyst. The operator can also interact with the user interface 1411 to input data into the hypotension prediction module 1404 or other modules associated with the module 1404. In one implementation, the user interface 1411 is configured to be displayed on a monitoring module (not shown) integrated with or separate from the hypotension prediction module 1404, but connected to the hypotension prediction module 1404 via a connection. Module 1404. Such a connection may be wired, wireless (e.g., via NFC communication, Bluetooth, etc.), over a network (e.g., the Internet), or via another device with which both modules communicate.

[0123] In various embodiments of the present disclosure, the term hypotension prediction module may be described as a predictive modeling engine and is used interchangeably.

[0124] FIG. 15 shows a flow chart illustrating a portion of a method for pre-processing signals and extracting signal features to determine a prediction of an AHE event in a patient, according to one embodiment of the present disclosure.

[0125] In one example, high fidelity hemodynamic waveform data and other clinical data such as HR, SBP, DBP, MAP, PP, SPO2% are received. For example, at least MAP waveforms of existing patients are stored in a database or cloud. These stored patient data may or may not have experienced AHE during their stay in the ICU or during surgery for surgery. Other waveform data and clinical data may also be received. The hemodynamic waveform data is sampled at a 125 Hz frequency or other sampling rate such as once per minute or once per second.

[0126] After receiving and managing input data, such as hemodynamic and other clinical data of multiple test patients used for model training or new patients for prediction of AHE events, pre-processing of the input data (e.g., continuous waveform data) is performed using filtering, smoothing, and interpolation in step 1501. In one example, noisy input data was pre-processed by removing baseline, eliminating extreme data values, smoothing, and filling missing values ​​by linear interpolation.

[0127] The pre-processed data then undergoes a segmentation process in step 1502. In one example, a sliding window method with a non-overlapping window size of 10 seconds is applied.

[0128] Next, non-linear and linear feature extraction processes are performed on the segmented data in step 1503. In one example, time dependent features are extracted from the segmented waveform data. These features are maximum power, average power, minimum power, maximum frequency, mean frequency, median frequency, minimum frequency, maximum of segmented data, mean of segmented data, minimum of segmented data, variance of segmented data, skewness of segmented data, kurtosis of segmented data, RMS value of segmented data, 1st autoregression of segmented data, slope sign change of segmented data, zero crossing of segmented data, autocorrelation from lag 1 to lag 5, 0% quantile, 25% quantile, 50% quantile, 75% quantile, 100% quantile.

[0129] The extracted features and corresponding feature data are then sent to a predictive modeling engine, such as predictive modeling engine 1601 of FIG. 16, for further processing.

[0130] FIG. 16 illustrates a block diagram 1600 showing the inputs and outputs of a stacking ensemble model-based predictive modeling engine 1601 used to adaptively detect and predict AHE events in a patient, according to one embodiment of the present disclosure.

[0131] The stacking ensemble model-based predictive modeling engine 1601 may receive the extracted features from the segment window W1, as shown in 1602, and other clinical data for multiple patients (e.g., test patients or existing patients). The predictive modeling engine 1601 may also receive input from a user interface, such as a time frame or a determined time window within which the occurrence of an AHE event of a new patient within the time frame is predicted. The time frame may refer to a short period or time window, such as 3-30 minutes, or a long period or time window, such as 3 hours to 3 days, and the predictive modeling engine 1601 accordingly provides a prediction of an AHE event occurring within the time frame or determined time window. Using the inputs 1602, 1603, the predictive modeling engine 1601 determines a prediction of the occurrence of an AHE event of a new patient as well as a severity level of the AHE event, and outputs the prediction to the user interface.

[0132] 17A illustrates various input data used to adaptively detect and predict AHE events in a patient according to various embodiments of the present disclosure. FIG. 17B illustrates a MAP waveform used as an input to adaptively detect and predict AHE events in a patient according to one embodiment of the present disclosure. In this embodiment, the MAP waveform data is for a patient with a history of AHE.

[0133] 18 illustrates features extracted from signals for multiple patients used for model training, according to one embodiment of the present disclosure. Two signals 1802, 1804 of the multiple signals are shown, one associated with patient P1 with a history of AHE and the other associated with patient P2 without a history of AHE. N Each signal is associated with a number of segment windows W1 to W N The segments are preprocessed and segmented according to the segment windows W1 to W N The signal features and corresponding feature data are extracted for each of P1 and P NData corresponding to features extracted from signals relating to are shown in tables 1803, 1805.

[0134] FIG. 19 shows a diagram illustrating a process of extracting at least one feature in a signal used to adaptively detect and predict an AHE event in a patient according to one embodiment of the present disclosure. In this embodiment, an input of a signal related to a patient is accepted. The input may further include the patient's past recorded data 1911 as well as a target time frame 1912 to predict the patient's AHE event. The past recorded data may be segmented according to a number of segment windows W1-W9, and features from each of the segment windows W1-W9 are calculated to generate feature data as shown in table 1908. Such feature data is then used for model training and / or AHE event prediction.

[0135] 20 shows a block diagram illustrating an apparatus 2000 for detecting and predicting AHE events in a patient according to one embodiment of the present disclosure. The apparatus 2000 is configured to receive time-dependent features / feature data from a feature extraction module (not shown) or algorithm, such features including extracted non-linear and linear feature waveform data, and the feature data includes data from laboratory reports such as blood tests, urinalysis, clinical data such as demographic and comorbidity data, perioperative management data such as BP history, frailty score, etc. The input data may also include input of a time frame (short-term or long-term) for which a prediction of the occurrence of an AHE event in a new patient is determined.

[0136] The apparatus 2000 is further configured to generate a prediction model developed based on at least one of these features. In particular, the prediction model generates a prediction of AHE occurrence by applying a stacking ensemble method, and in generating the prediction, a multi-level sequential multi-class classification (or multi-class classification) of AHE is applied to obtain high performance and avoid overfitting in learning and model training. For example, in the first level, two or more predictions based on at least one extracted feature are determined using multiple basic classifiers 2022 (classifiers 1, 2, ... n), such as random forest, gradient boosting, naive Bayes, generalized linear model, etc. At the same time, this approach is not limited to our classification algorithm, and is adaptable to various designs of classifiers, such as SVM, KNN, neural network, etc.

[0137] At the second level, a comparison of two or more predictions is performed to determine the best prediction (based on any one of the predictions or a combination thereof) or the best prediction model (any one of the combinations thereof) among the multiple basic classifiers, which can at best accurately detect the most subsequent AHE events of the test patient. Such best prediction model is then applied to determine the prediction of subsequent AHE of new patients. In one example, a majority voting approach is used as a meta-classifier, which is not limited to our classification algorithm and can be applied to other designs of classifiers such as logistic regression, neural network, Xgboost, etc.

[0138] The device is further configured to generate a prediction of the occurrence of the patient's subsequent AHE event, e.g., a yes / no output or a probability (%) of occurrence, and the onset of AHE within a predefined time frame. Additionally, a MAP value or a MAP value range of the subsequent AHE event is also determined. The MAP value or value range corresponds to a severity level of the subsequent AHE event. For example, the device 2000 is further configured to generate a severity level of the patient's subsequent AHE event corresponding to the MAP value and, optionally, the patient's personalized AHE / MAP threshold(s). In one example, the device is configured to determine four severity levels: level 1 severity (L1) for a predicted MAP value of 50 mmHg or less, level 2 severity (L2) for a predicted MAP value of 50 mmHg to 65 mmHg, level 3 severity (L3) for a predicted MAP value of 65 mmHg to 75 mmHg, and level 4 severity (L4) for a predicted MAP value above 75 mmHg.

[0139] 21 shows a flow diagram illustrating a process for detecting and predicting an AHE event in a patient according to one embodiment of the present disclosure. In this embodiment, a prediction of the occurrence of an AHE event in a patient is first determined by a stage 1 or stage 1 model 2111, and then a prediction of the MAP value and severity level of the AHE event is determined by a stage 2 or stage 2 model 2112. In this embodiment, both the stage 1 model and the stage 2 model employ sequential multi-level classification, using multiple classifiers to first generate predictions in each of them, and then using a meta-classifier to compare the generated predictions and determine an optimal prediction model (hereinafter referred to as the "stage 1 model" and the "stage 2 model", respectively) for use in generating the final prediction result 2103.

[0140] A training dataset 2101 containing extracted features and other clinical data of multiple patients (Patient 1, 2, ... n) is input to each of multiple classifiers (Classifiers 1, 2, ... n) in the Stage 1 model 2111, and then input to a meta-classifier to determine a prediction outcome 2122 of the occurrence of an AHE event within a given decision time window. In this embodiment, the prediction outcome is simply "yes" or "no".

[0141] In step 2123, it is checked whether the predicted outcome of the occurrence of the AHE event is "no" and step 2121 is executed. In step 2121, a predicted occurrence of AHE in a predetermined decision time window is declared. If the predicted outcome of the occurrence of the AHE event is "yes", the stage 2 model 2112 is applied to determine a predicted outcome of the MAP value of the AHE event 2124 and a severity level of the AHE event based on the predicted MAP value. Then, in step 2125, a predicted severity of the AHE event in a predetermined decision time window is declared.

[0142] 22 illustrates an exemplary computing device 2200, hereinafter interchangeably referred to as a computer system 2200, in which one or more such computing devices 2200 may be used to perform the method of FIG. 1. The exemplary computing device 2200 may be the same as the device illustrated in FIG. 202 The following description of a computing device 2200 is provided by way of example only and is not intended to be limiting.

[0143] As shown in Figure 22, the exemplary computing device 2200 includes a processor 2204 for executing software routines. Although a single processor is shown for clarity, the computing device 2200 may also include a multi-processor system. The processor 2204 is connected to a communications infrastructure 2206 for communicating with other components of the computing device 2200. The communications infrastructure 2206 may include, for example, a communications bus, crossbar, or network.

[0144] The computing device 2200 further includes a main memory 2208, such as random access memory (RAM), and a secondary memory 2210. The secondary memory 2210 may include a storage drive 2212, which may be, for example, a hard disk drive, a solid state drive, or a hybrid drive, and / or a removable storage drive 2214, which may include a magnetic drive, an optical disk drive, a solid state storage drive (e.g., a USB flash drive, a flash memory device, a solid state drive, or a memory card), or the like. The removable storage drive 2214 reads from and / or writes to a removable storage medium 2218 in a well-known manner. The removable storage medium 2218 may include magnetic, optical disk, non-removable storage media, etc., that are read from and written to by the volatile memory drive 2214. As will be appreciated by one of skill in the art(s), the removable storage medium 2218 includes a computer-readable storage medium having stored thereon computer-executable program code instructions and / or data.

[0145] In alternative implementations, the secondary memory 2210 may additionally or alternatively include other similar means for allowing computer programs or other instructions to be loaded into the computing device 2200. Such means may include, for example, a removable storage unit 2222 and interface 2220. Examples of removable storage units 2222 and interfaces 2220 include program cartridges and cartridge interfaces (such as those found in video game console devices), removable memory chips (such as EPROMs or PROMs) and associated sockets, removable solid-state storage drives (such as USB flash drives, flash memory devices, solid-state drives or memory cards), and other removable storage units 2222 and interfaces 2220 that allow software and data to be transferred from the removable storage unit 2222 to the computer system 2200.

[0146] The computing device 2200 also includes at least one communication interface 2224. The communication interface 2224 allows software and data to be transferred between the computing device 2200 and external devices via a communication path 2226. In various embodiments of the invention, the communication interface 2224 allows data to be transferred between the computing device 2200 and a data communication network, such as a public or private data communication network. The communication interface 2224 may be used to exchange data between different computing devices, such as computing devices 2200 forming part of an interconnected computer network. Examples of the communication interface 2224 may include a modem, a network interface (such as an Ethernet card), a communication port (such as serial, parallel, printer, GPIB, IEEE 1394, RJ45, etc.), an antenna with associated circuitry, etc. The communication interface 2224 may be wired or wireless. The software and data transferred through the communication interface 2224 are in the form of signals, which may be electronic, electromagnetic, optical, or other signals capable of being received by the communication interface 2224. These signals are provided to the communications interface via communications path 2226.

[0147] As shown in FIG. 22, the computing device 2200 further includes a display interface 2202 that performs operations for rendering images on an associated display 2230, and an audio interface 2232 that performs operations for playing audio content via associated speaker(s) 2234.

[0148] The term "computer program product" as used herein may refer, in part, to removable storage medium 2218, removable storage unit 2222, a hard disk installed in storage drive 2212, or a carrier wave that carries software to communication interface 2224 via communication path 2226 (wireless link or cable). A computer readable storage medium refers to any non-transitory, non-volatile tangible storage medium that provides recorded instructions and / or data to computing device 2200 for execution and / or processing. Examples of such storage media include magnetic, CD-ROM, DVD, Blu-ray disk, hard disk drive, ROM or IC, solid state storage drive (such as USB flash drive, flash memory device, solid state drive or memory card), hybrid drive, magneto-optical disk, or computer readable card such as PC card, etc., whether such device is internal or external to computing device 2200. Examples of transitory or non-tangible computer-readable transmission media that may be involved in providing software, application programs, instructions, and / or data to computing device 2200 include wireless or infrared transmission channels, as well as network connections to other computers or networked devices, and the Internet or intranets, including email transmissions and information stored on websites and the like.

[0149] Computer programs (also referred to as computer program code) are stored in main memory 2208 and / or secondary memory 2210. Computer programs may also be received via communications interface 2224. Such computer programs, when executed, enable computing device 2200 to implement one or more features of the embodiments discussed herein. In various embodiments, the computer programs, when executed, cause a processor 2204Such computer programs thus represent controllers of the computer system 2200.

[0150] The software may be stored in a computer program product and loaded into the computing device 2200 using the removable storage drive 2214, the storage drive 2212, or the interface 2220. The computer program product may be a non-transitory computer readable medium. Alternatively, the computer program product may be downloaded to the computer system 2200 via communications path 2226. The software, when executed by the processor 2204, causes the computing device 2200 to perform the operations necessary to execute the method illustrated in FIG.

[0151] The embodiment of FIG. 202 It should be understood that the features of the computing device 2200 are presented merely as examples to illustrate the operation and structure of the computing device 2200. Thus, in some embodiments, one or more features of the computing device 2200 may be omitted. Also, in some embodiments, one or more features of the computing device 2200 may be combined together. Additionally, in some embodiments, one or more features of the computing device 2200 may be divided into one or more component parts.

[0152] It will be appreciated by those skilled in the art that numerous variations and / or modifications may be made to the invention as shown in the specific embodiments without departing from the spirit or scope of the invention as broadly described, and the present embodiments are therefore to be considered in all respects as illustrative and not restrictive.

[0153] In addition, some or all of the above embodiments can be described as, but are not limited to, the following supplementary notes. (Appendix 1) 1. A method for adaptively detecting an acute hypotensive episode (AHE) event in a patient, comprising: processing, by a processor, signals relating to the patient to determine at least one mean arterial pressure (MAP) threshold for the patient; detecting, by the processor, an AHE event in the patient based on the at least one MAP threshold; A method comprising: (Appendix 2) The step of processing the signal by the processor to determine the at least one mean arterial pressure (MAP) threshold for the patient comprises: determining a variation of at least one of said signals from a baseline of said signals; The method of claim 1, comprising: (Appendix 3) The step of detecting, by the processor, the AHE event in the patient includes: Comparing one or more MAP values ​​of said signal to said at least one MAP threshold. The method of claim 1, comprising: (Appendix 4) The step of detecting, by the processor, the AHE event in the patient includes: determining whether a duration associated with the one or more MAP values ​​is greater than a preconfigured threshold duration; The method of claim 3, comprising: (Appendix 5) The step of detecting, by the processor, the AHE event in the patient includes: Determining whether the at least one MAP threshold is lower than a first preconfigured MAP threshold. The method of claim 1, comprising: (Appendix 6) The step of detecting, by the processor, the AHE event in the patient includes: determining whether a baseline average MAP value of the signal is lower than a second preconfigured MAP threshold; The method of claim 1, comprising: (Appendix 7) The step of detecting, by the processor, the one or more AHE events in the patient includes: detecting an onset and duration of the AHE event in the patient based on the at least one MAP threshold and the signal. 2. The method of any one of the preceding clauses, comprising: (Appendix 8) storing in a database the onset and duration of the AHE event in the patient, and one or more MAP values ​​associated with the AHE event; 8. The method of claim 7, further comprising: (Appendix 9) and receiving a MAP waveform, the signal including the MAP waveform. 13. The method of any one of the preceding clauses. (Appendix 10) receiving at least one clinical data, the at least one clinical data including at least one of blood constituents, urine constituents, hemodynamic waveform data, demographic data, comorbidity data, blood pressure, central venous pressure, frailty score, and perioperative management data, the signal including the at least one clinical data; 13. The method of any one of the preceding clauses. (Appendix 11) determining, by the processor, a prediction of a subsequent AHE event for the patient based on the at least one MAP threshold and the signal. The method of any one of the preceding clauses, further comprising: (Appendix 12) receiving an input regarding a time frame, the prediction including a prediction of an occurrence of the subsequent AHE event for the patient within the time frame; 12. The method according to claim 11. (Appendix 13) The step of determining, by the processor, the prediction of the subsequent AHE event in the patient comprises: determining a MAP value of the subsequent AHE event, the MAP value being related to the severity level of the subsequent AHE event; 12. The method according to claim 11. (Appendix 14) extracting at least one feature in the signal, and the determination of the prediction of the subsequent AHE event is based on the at least one feature of the signal. 14. The method according to any one of claims 11 to 13. (Appendix 15) The step of determining, by the processor, the prediction of the subsequent AHE for the patient includes: determining two or more predictions based on the at least one feature extracted from a plurality of signals for a plurality of patients; comparing the two or more predictions to determine an optimal prediction of a subsequent AHE event for the plurality of patients based on the at least one feature extracted from the plurality of signals, the optimal prediction being one or a combination of the two or more predictions, and the prediction of the subsequent AHE for the patient is based on the optimal prediction of the subsequent AHE event for the plurality of patients; 15. The method of claim 14, comprising: (Appendix 16) 1. An apparatus for adaptively detecting an acute hypotensive episode (AHE) event in a patient, comprising: At least one processor; at least one memory containing computer program code; The at least one memory and the computer program code are configured, together with at least one processor, to cause the device to at least perform the method according to any one of claims 1 to 15. Device.

[0154] This application claims priority to Singapore Patent Application No. 10202108729V, filed on August 10, 2021, the disclosure of which is incorporated herein by reference in its entirety. [Explanation of symbols]

[0155] 100 Flowchart 200 Block Diagram 201 Source 202 Equipment 204 Personalized AHE Definition Module 206 Hypotension Episode Prediction Module 208 Data Storage 210 Monitoring Module 211 User Interface 300 Block Diagram 301 Signal Source 304 Personalized AHE Definition Module 311 Data Storage 313 Patient Monitoring Unit 314 Invasive Blood Pressure Monitoring Devices 315 Non-invasive or minimally invasive blood pressure monitoring devices 601 MAP waveform 602 Smoothed MAP waveform 700 Processed MAP Waveforms 800 Flowchart 901 Processed MAP waveform 1001 Processed MAP waveform 1002 A subsample of the processed MAP waveform 1001 1003 Processed MAP waveform 1101 Processed MAP waveform 1102 A subsample of the processed MAP waveform 1101 1103 Processed MAP waveform 1201 Processed MAP waveform 1202 A subsample of the processed MAP waveform 1201 1203 Processed MAP waveform 1401 Data Storage 1402 Personalized AHE Definition Module 1403 Other clinical data 1404 Hypotension Prediction Module 1411 User Interface 1600 Block Diagram 1601 Predictive Modeling Engine 1602,1603 Input 1802,1804 signal 1803,1805 Table 1908 Table 2000 equipment 2001 Input data for predictive modeling module 2002 Predictive Modeling Module 2003 Predictive model output 2101 training data set 2102 Stacking Ensemble Model 2103 Final forecast result 2111 Stage 1 Model 2112 Stage 2 Model 2200 Computing Devices 2202 Display Interface 2204 Processor 2206 Communication Infrastructure 2208 Main Memory 2210 Secondary Memory 2212 Storage Drive 2214 Removable Storage Drive 2218 Removable storage media 2220 Interface 2222 Removable Storage Unit 2224 Communication Interface 2226 Communication Path 2230 Display 2232 Audio Interface 2234 Speaker

Claims

1. 1. A method for adaptively detecting an acute hypotensive episode (AHE) event in a patient, comprising: processing, by a processor, signals relating to the patient to determine at least one mean arterial pressure (MAP) threshold for the patient; detecting, by the processor, an AHE event in the patient based on the at least one MAP threshold; determining, by the processor, a prediction of a subsequent AHE event for the patient based on the at least one MAP threshold and the signal; Including, Determining, by the processor, the prediction of the subsequent AHE event in the patient includes: determining a MAP value for the subsequent AHE event, the MAP value being related to a severity level of the subsequent AHE event; method.

2. Processing the signals by the processor to determine the at least one MAP threshold for the patient includes: determining a variation of at least one of said signals from a baseline of said signals; The method of claim 1 , comprising:

3. Detecting, by the processor, the AHE event in the patient, comprises: Comparing one or more MAP values ​​of said signal to said at least one MAP threshold. The method of claim 1 , comprising:

4. Detecting, by the processor, the AHE event in the patient, comprises: Determining whether a duration associated with the one or more MAP values ​​is greater than a preconfigured threshold duration. The method of claim 3 , comprising:

5. Detecting, by the processor, the AHE event in the patient, comprises: Determining whether the at least one MAP threshold is lower than a first preconfigured MAP threshold. The method of claim 1 , comprising:

6. Detecting, by the processor, the AHE event in the patient, comprises: determining whether a baseline average MAP value of the signal is lower than a second preconfigured MAP threshold; The method of claim 1 , comprising:

7. Detecting, by the processor, the one or more AHE events in the patient includes: detecting an onset and duration of the AHE event in the patient based on the at least one MAP threshold and the signal. The method of claim 1 , comprising:

8. storing in a database the onset and duration of the AHE event of the patient, and one or more MAP values ​​associated with the AHE event; The method of claim 7 further comprising:

9. and receiving a MAP waveform, the signal including the MAP waveform. The method of claim 1.

10. 1. An apparatus for adaptively detecting an Acute Hypotensive Episode (AHE) event in a patient, comprising: At least one processor; at least one memory containing computer program code; The at least one memory and the computer program code are configured, together with at least one processor, to cause the device to at least perform the method according to any one of claims 1 to 9. Device.

Images