Detecting and Differentiating Nociceptive Events from Hemodynamic Drug Administration Events

Abstract

A method for monitoring a patient's arterial pressure and alerting medical personnel of the patient's current nociception includes a hemodynamic monitor receiving sensed hemodynamic data representative of the patient's arterial pressure waveform. The hemodynamic monitor performs waveform analysis of the sensed hemodynamic data to calculate a plurality of signal measurements. A detection input feature, a hemodynamic agent detection input feature, and a stable detection input feature are extracted from the plurality of signal measurements. A first probability is determined based on the detection input features and the hemodynamic agent detection input feature. A second probability is determined based on the detection input features and the stable detection input feature. The hemodynamic monitor compares the second probability to the first probability to determine an output probability of the patient's current nociception. The hemodynamic monitor activates a sensory alarm in response to the output probability meeting a predetermined criterion.

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of U.S. Provisional Application No. 63 / 309,394, filed February 11, 2022, and entitled “DETECTING AND DIFFERENTIATING NOCICEPTION EVENTS FROM HEMODYNAMIC DRUG ADMINISTRATION EVENTS,” the disclosure of which is hereby incorporated by reference in its entirety. [Background technology]

[0002] The present disclosure relates generally to hemodynamic monitoring, and more particularly to using monitored hemodynamic data to detect nociception in a patient.

[0003] Nociception is the process by which nerve endings called nociceptors detect noxious stimuli and send signals to the central nervous system that are interpreted as pain. Harmful injury initiates a "sharp" signal from the source of injury. Locally, nociception initiates an inflammatory response. The signal then travels through neurons to the spinal column, where a muscle reflex is triggered. The signal continues to the brain, where the nociceptive signal triggers a sympathetic nervous system response when it reaches the lower part of the brain. Nociception can trigger a sympathetic nervous system response without or before consciousness is reached, and thus an unconscious patient during surgery or in intensive care may experience pain. To prevent patients from waking up from surgery or intensive care in pain, medical personnel administer painkillers to patients at various times before and / or during surgery and in intensive care. However, pain thresholds and tolerances vary from patient to patient, and because patients cannot communicate verbally or provide feedback while unconscious, it can be difficult to know how much painkiller to administer. Giving a patient too little pain medication during surgery can result in the patient waking up in pain after surgery. Giving a patient too much pain medication during surgery can result in the patient experiencing nausea, drowsiness, impaired thinking ability, and impaired function.

[0004] Considering the negative consequences of administering too little pain medication to a patient, and the negative consequences of administering too much pain medication to a patient, a solution is needed that gives medical personnel the ability to detect or predict nociception in unconscious patients during surgery. Accurately detecting nociception in patients during surgery can help medical personnel know the appropriate amount of pain medication to administer to a patient without giving the patient too much pain medication so that the patient does not wake up from surgery in significant pain. Summary of the Invention

[0005] In one example, a system for monitoring a patient's arterial pressure and providing alerts to a medical professional of a patient's nociception includes a hemodynamic sensor that generates hemodynamic data representative of the patient's arterial pressure waveform. The system also includes a system memory that stores nociception software code, and a user interface with a sensory alarm that provides a sensory signal to alert a medical professional of a patient's nociceptive event. A hardware processor in the system is configured to execute the nociception software code to perform a waveform analysis of the hemodynamic data to determine a plurality of signal measurements. The hardware processor is also configured to extract detection input features from the plurality of signal measurements that are indicative of the patient's nociceptive event. The hardware processor is also configured to extract hemodynamic drug detection input features from the plurality of signal measurements that are indicative of the patient's hemodynamic drug administration event. The hardware processor is also configured to extract hemodynamic drug prediction input features from the plurality of signal measurements that predict effects on the patient from a future hemodynamic drug administration event. The hardware processor is also configured to extract stable detection input features from the plurality of signal measurements that are indicative of a stable episode of the patient. The hardware processor is also configured to determine a first probability based on the hemodynamic drug prediction input features and the stable detection input features. The first probability represents a probability that the patient will experience an effect from a future hemodynamic drug administration event. The hardware processor is also configured to determine a second probability based on the detection input features and the hemodynamic drug detection input features. The second probability represents a probability that the patient will experience a current nociceptive event for a current hemodynamic drug administration event. The hardware processor is also configured to determine a third probability based on the detection input features and the stable detection input features. The third probability represents a probability that the patient will experience a current nociceptive event for a stable episode. The hardware processor is also configured to compare the third probability to the first probability and the second probability to determine an output probability of the patient's current nociceptive event. The hardware processor activates a sensory alarm of the user interface in response to the output probability meeting a predetermined detection criterion.

[0006] In another example, a method for monitoring a patient's arterial pressure and providing a medical professional with an alert of the patient's current nociception includes receiving hemodynamic data sensed by a hemodynamic monitor representing the patient's arterial pressure waveform. The method further includes performing a waveform analysis of the sensed hemodynamic data by the hemodynamic monitor to calculate a plurality of signal measurements of the sensed hemodynamic data. A detection input feature indicative of a patient's current nociceptive event is extracted from the plurality of signal measurements by the hemodynamic monitor. A hemodynamic drug detection input feature indicative of a patient's current hemodynamic drug administration event is extracted from the plurality of signal measurements by the hemodynamic monitor. A current hemodynamic drug administration event is defined as the patient experiencing an increase in heart rate and blood pressure due to administration of a compound that alters cardiovascular hemodynamics. A stable detection input feature indicative of a patient stable episode in which the patient is not experiencing a nociceptive event or a hemodynamic drug administration event is extracted by the hemodynamic monitor from the plurality of signal measurements. A first probability is determined by the hemodynamic monitor based on the detection input feature and the hemodynamic drug detection input feature. The first probability represents a probability that the patient will experience a current nociceptive event for a current hemodynamic drug administration event. The second probability is determined by the hemodynamic monitor based on the detection input features and the stable detection input features. The second probability represents a probability that the patient will experience a current nociceptive event for a stable episode. The hemodynamic monitor compares the second probability to the first probability to determine an output probability of the patient's current nociceptive event. The hemodynamic monitor activates a sensory alarm and generates a sensory signal in response to the output probability satisfying a predetermined detection criterion. [Brief description of the drawings]

[0007] [Figure 1] FIG. 1 is a perspective view of an exemplary hemodynamic monitor that analyzes a patient's arterial pressure and provides medical personnel with a risk score and warnings for the patient's nociceptive events. [Diagram 2] FIG. 2 is a perspective view of an example of a minimally invasive pressure sensor for sensing hemodynamic data representative of a patient's arterial pressure. [Diagram 3]FIG. 3 is a perspective view of an example of a non-invasive sensor for sensing hemodynamic data representative of a patient's arterial pressure. [Figure 4] FIG. 4 is a block diagram illustrating an exemplary hemodynamic monitoring system that determines a risk score representing the probability of a current nociceptive event, a future nociceptive event, a current hemodynamic drug administration event, a future hemodynamic drug administration event, and / or a stable period for a patient based on a set of input features derived from signal measurements of the patient's arterial pressure waveform. [Diagram 5] FIG. 5 is an illustration of a clinical dataset with clinical annotations used for machine training of the hemodynamic monitoring system. [Figure 6] FIG. 6 is a graph showing systolic blood pressure and heart rate over time from the clinical data set of FIG. 5, illustrating nociceptive events and administration of analgesic medication. [Figure 7] FIG. 7 is a graph showing systolic blood pressure and heart rate over time from the clinical data set of FIG. 5, illustrating stable episodes. [Figure 8] FIG. 8 is a graph showing systolic blood pressure and heart rate over time from the clinical data set of FIG. 5, illustrating hemodynamic drug administration events and administration of vasopressors. [Figure 9] FIG. 9 is a flow diagram for extracting a set of input features derived from waveform features of a patient's arterial pressure waveform for training a machine learning model of a hemodynamic monitoring system. [Figure 10] FIG. 10 is a graph showing example arterial pressure waveform traces, including example indicia corresponding to signal measurements used to extract input features that determine a patient's risk score. [Figure 11] FIG. 11 is a diagram of the three machine learning models used to determine the probability of a nociceptive event in a patient. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0008] As described herein, the hemodynamic monitoring system implements a multi-model approach to identify nociceptive events in a patient. A first model calculates the probability of an effect caused by administration of a hemodynamic drug (hereinafter referred to as a "hemodynamic drug administration event"). A second model calculates the probability of a nociceptive event for a hemodynamic drug administration event. A third model calculates the probability of a nociceptive event for a stable episode. The combined output of the three models calculates the probability that the patient is experiencing a nociceptive event rather than a hemodynamic drug administration event.

[0009] The machine learning predictive model of the hemodynamic monitoring system is trained using a clinical dataset containing arterial pressure waveforms that are labeled with clinical annotations of administration of analgesics, vasopressors, inotropes, fluids, and other medications that alter cardiovascular hemodynamics. The hemodynamic monitoring system is described in detail below with reference to Figures 1-11.

[0010] FIG. 1 is a perspective view of a hemodynamic monitor 10 that determines a score representative of the patient's probability of a current nociceptive event and / or a score representative of the patient's probability of a future nociceptive event. As illustrated in FIG. 1, the hemodynamic monitor 10 includes a display device 12 that, in the example of FIG. 1, presents a graphical user interface including control elements (e.g., graphical control elements) that enable user interaction with the hemodynamic monitor 10. The hemodynamic monitor 10 may also include a number of input and / or output (I / O) connectors configured for wired connection (e.g., electrical connection and / or communicative connection) with one or more peripheral components, such as one or more hemodynamic sensors, as described further below. For example, as shown in FIG. 1, the hemodynamic monitor 10 may include an I / O connector 14. Although the example of FIG. 1 shows five separate I / O connectors 14, it should be understood that in other examples, the hemodynamic monitor 10 may include less than five I / O connectors or more than five I / O connectors. In yet other embodiments, hemodynamic monitor 10 may not include I / O connector 14, but rather may communicate wirelessly with various peripheral devices.

[0011] As described further below, hemodynamic monitor 10 includes one or more processors and computer readable memory storing nociceptive detection and prediction software code executable to generate a score representative of the patient's probability of a current (i.e., present) nociceptive event and / or a score representative of the patient's probability of a future nociceptive event. Hemodynamic monitor 10 can receive sensed hemodynamic data representative of the patient's arterial pressure waveform, for example, via one or more hemodynamic sensors connected to hemodynamic monitor 10 via I / O connector 14. As described further below, hemodynamic monitor 10 executes the nociceptive prediction software code to obtain a plurality of nociceptive profiling parameters (e.g., input features) using the received hemodynamic data, which may include one or more vital sign parameters characterizing the patient's vital sign data, as well as differential and combination parameters derived from the one or more vital sign parameters.

[0012] As shown in Figure 1, hemodynamic monitor 10 can present a graphical user interface on display device 12. Display device 12 can be a liquid crystal display (LCD), a light emitting diode (LED) display, an organic light emitting diode (OLED) display, or other display device suitable for providing information to a user in a graphical format. In some implementations, such as the implementation of Figure 1, display device 12 can be a touch-sensitive and / or presence-sensitive display device configured to receive user input in the form of gestures, such as touch gestures, scroll gestures, zoom gestures, swipe gestures, or other gesture input.

[0013] Figure 2 is a perspective view of a hemodynamic sensor 16 that may be attached to a patient to sense hemodynamic data representative of the patient's arterial pressure. The hemodynamic sensor 16 illustrated in Figure 2 is an example of a minimally invasive hemodynamic sensor that may be attached to the patient via a fluid-filled tube connected to a radial artery catheter inserted in the patient's arm, for example. In another example, the hemodynamic sensor 16 may be attached to the patient via a femoral artery catheter inserted in the patient's leg.

[0014] As shown in FIG. 2, the hemodynamic sensor 16 includes a housing 18, a fluid input port 20, a catheter-side fluid port 22, and an I / O cable 24. The fluid input port 20 is configured to be connected via tubing or other hydraulic connection to a fluid source, such as a saline bag or other fluid input source. The catheter-side fluid port 22 is configured to be connected via tubing or other hydraulic connection to a catheter (e.g., a radial artery catheter or a femoral artery catheter) inserted into a patient's arm (i.e., a radial artery catheter) or a patient's leg (i.e., a femoral artery catheter). The I / O cable 24 is configured to be connected to the hemodynamic monitor 10 (FIG. 1), for example, via one or more of the I / O connectors 14. The housing 18 of the hemodynamic sensor 16 encloses one or more pressure transducers, communication circuitry, processing circuitry, and corresponding electronic components for sensing fluid pressure corresponding to the patient's arterial pressure, which is transmitted to the hemodynamic monitor 10 (FIG. 1) via the I / O cable 24.

[0015] In operation, a column of fluid (e.g., saline) is introduced from a fluid source (e.g., a saline bag) through hemodynamic sensor 16 via fluid input port 20 to catheter side fluid port 22 toward a catheter inserted into a patient. Arterial pressure is transmitted through the fluid column to a pressure sensor located within housing 16 that senses the pressure of the fluid column. Hemodynamic sensor 16 converts the sensed pressure of the fluid column into an electrical signal via a pressure transducer and outputs a corresponding electrical signal to hemodynamic monitor 10 (FIG. 1) via I / O cable 24. Hemodynamic sensor 16 thus transmits analog sensor data (or a digital representation of the analog sensor data) to hemodynamic monitor 10 (FIG. 1) that represents a substantially continuous, beat-to-beat monitoring of the patient's arterial pressure.

[0016] FIG. 3 is a perspective view of a hemodynamic sensor 26 for sensing hemodynamic data representative of a patient's arterial pressure. The hemodynamic sensor 26 illustrated in FIG. 3 is an example of a non-invasive hemodynamic sensor that may be attached to a patient via one or more finger cuffs to sense data representative of the patient's arterial pressure. As shown in FIG. 3, the hemodynamic sensor 26 includes an inflatable finger cuff 28 and a cardiac reference sensor 30. The inflatable finger cuff 28 includes an inflatable blood pressure bladder configured to inflate and deflate as controlled by a pressure controller (not shown) that is pneumatically connected to the inflatable finger cuff 28. The inflatable finger cuff 28 also includes an optical (e.g., infrared) transmitter and an optical receiver that are electrically connected to the pressure controller (not shown) to measure the changing volume of the artery below the finger cuff.

[0017] In operation, the pressure controller continuously adjusts the pressure in the finger cuff to maintain a constant volume in the finger's artery (i.e., the arterial unloaded volume) as measured via the optical transmitter and receiver in the inflatable finger cuff 28. The pressure applied by the pressure controller to continuously maintain the unloaded volume represents the blood pressure in the finger and is communicated by the pressure controller to the hemodynamic monitor 10 shown in FIG. 1. The cardiac reference sensor 30 measures the hydrostatic height difference between the level at which the finger is held and the reference level for pressure measurement, which is typically the cardiac level. Thus, the hemodynamic sensor 26 transmits sensor data representing a substantially continuous beat-to-beat monitoring of the patient's arterial pressure waveform.

[0018] 4 is a block diagram of a hemodynamic monitoring system 32 that determines a nociceptive score representing the probability of a current or future nociceptive event for a patient 36 based on a set of nociceptive profiling parameters (also referred to as input features) derived from the patient's arterial blood pressure. The hemodynamic monitoring system 32 monitors the arterial pressure of the patient 36 and provides an alert to a medical professional 38 when the nociceptive score of the patient 36 exceeds a predetermined threshold. The medical professional 38 can respond to the alert by administering an appropriate analgesic to the patient 36 to mitigate the current or future nociceptive event.

[0019] As shown in Figure 4, the hemodynamic monitoring system 32 includes the hemodynamic monitor 10 and a hemodynamic sensor 34. The hemodynamic monitoring system 32 can be implemented in a patient care environment, such as an ICU, an OR, or other patient care environment. As shown in Figure 4, the patient care environment can include a patient 36 and a healthcare worker 38 trained to utilize the hemodynamic monitoring system 32.

[0020] As discussed above with respect to FIG. 1, hemodynamic monitor 10 may be an integrated hardware unit including, for example, system processor 40, system memory 42, display device 12, analog-to-digital (ADC) converter 44, and digital-to-analog (DAC) converter 46. In other embodiments, any one or more of the components and / or described functionality of hemodynamic monitor 10 may be distributed among multiple hardware units. For example, in some embodiments, display device 12 may be a separate display device that is remote from and operably coupled to hemodynamic monitor 10. Although generally illustrated and described in the embodiment of FIG. 4 as an integrated hardware unit, it should be understood that hemodynamic monitor 10 may include any combination of devices and components that are electrically, communicatively, or otherwise operably connected to perform the functions attributed to hemodynamic monitor 10 herein.

[0021] As shown in FIG. 4, the system memory 42 stores nociceptive software code 48 that forms a predictive model for the hemodynamic monitor 10. The nociceptive software code 48 includes a first module 50 for extracting and calculating waveform features from the arterial pressure of the patient 36, a second module 51 for extracting input features from the waveform features, and a third module 52 for calculating a probability of nociception for the patient 36 based on the input features. The display device 12 provides a user interface 54 that includes control elements 56 that enable user interaction with the hemodynamic monitor 10 and / or other components of the hemodynamic monitoring system 32. As illustrated in FIG. 4, the user interface 54 also provides a sensory alarm 58 for providing alerts to medical personnel of current or predicted future nociceptive events for the patient 36, as will be further described below. The sensory alarm 58 may be implemented as one or more of a visual, audible, tactile, or other type of sensory alarm. For example, the sensory alarm 58 may be triggered as any combination of a flashing and / or colored graphic presented by the user interface 54 on the display device 12, a display of a nociceptive score via the user interface 54 on the display device 12, an audible warning sound such as a siren or repetitive tone, and a tactile alarm configured to cause the hemodynamic monitor 10 to vibrate or otherwise deliver a physical impulse that is perceptible to the medical professional 38 or other user.

[0022] The hemodynamic sensor 34 can be attached to the patient 36 to sense hemodynamic data representative of the arterial pressure waveform of the patient 36. The hemodynamic sensor 34 is operatively connected (e.g., electrically and / or communicatively connected via a wired or wireless connection or both) to the hemodynamic monitor 10 to provide the sensed hemodynamic data to the hemodynamic monitor 10. In some examples, the hemodynamic sensor 34 provides the hemodynamic data representative of the arterial pressure waveform of the patient 36 to the hemodynamic monitor 10 as an analog signal, which is converted by the ADC 44 into digital hemodynamic data representative of the arterial pressure waveform. In other examples, the hemodynamic sensor 34 can provide the sensed hemodynamic data to the hemodynamic monitor 10 in a digital format, in which case the hemodynamic monitor 10 may not include or utilize the ADC 44. In yet another embodiment, hemodynamic sensor 34 can provide hemodynamic data representative of the arterial pressure waveform of patient 36 to hemodynamic monitor 10 as an analog signal that is analyzed by hemodynamic monitor 10 in its analog form.

[0023] Hemodynamic sensor 34 may be a non-invasive or minimally invasive sensor attached to patient 36. For example, hemodynamic sensor 34 may take the form of minimally invasive hemodynamic sensor 16 (FIG. 2), non-invasive hemodynamic sensor 26 (FIG. 3), or other minimally invasive or non-invasive hemodynamic sensor. In some examples, hemodynamic sensor 34 may be non-invasively attached to an extremity of patient 36, such as the wrist, arm, finger, ankle, toe, or other extremity of patient 36. In this manner, hemodynamic sensor 34 may take the form of a small, lightweight, comfortable hemodynamic sensor suitable for extended wear by patient 36 to provide substantially continuous beat-to-beat monitoring of arterial pressure of patient 36 over extended periods of time, such as minutes or hours.

[0024] In certain embodiments, the hemodynamic sensor 34 may be configured to sense the arterial pressure of the patient 36 in a minimally invasive manner. For example, the hemodynamic sensor 34 may be attached to the patient 36 via a radial artery catheter inserted in the arm of the patient 36. In other embodiments, the hemodynamic sensor 34 may be attached to the patient 36 via a femoral artery catheter inserted in the leg of the patient 36. Such minimally invasive techniques may similarly enable the hemodynamic sensor 34 to provide substantially continuous beat-to-beat monitoring of the arterial pressure of the patient 36 over extended periods of time, such as minutes or hours.

[0025] The system processor 40 is a hardware processor configured to execute nociceptive software code 48, which implements a first module 50, a second module 51, and a third module 52 to generate a nociceptive score representing the probability of a current nociceptive event or the probability of a future nociceptive event for the patient 36. Examples of the system processor 40 may include any one or more of a microprocessor, a controller, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other equivalent discrete or integrated logic circuitry.

[0026] The system memory 42 may be configured to store information within the hemodynamic monitor 10 during operation. In some examples, the system memory 42 is described as a computer-readable storage medium. In some examples, the computer-readable storage medium may include a non-transitory medium. The term "non-transitory" may indicate that the storage medium is not embodied in a carrier wave or propagating signal. In certain examples, the non-transitory storage medium may store data that may change over time (e.g., in RAM or cache). The system memory 42 may include volatile and non-volatile computer-readable memory. Examples of volatile memory include random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), and other forms of volatile memory. Examples of non-volatile memory include, for example, magnetic hard disks, optical disks, flash memory, or forms of electrically programmable memory (EPROM) or electrically erasable programmable (EEPROM) memory.

[0027] Display device 12 may be a liquid crystal display (LCD), a light emitting diode (LED) display, an organic light emitting diode (OLED) display, or other display suitable for providing information to a user in a graphical format. User interface 54 may include graphical and / or physical control elements that allow user input to interact with hemodynamic monitor 10 and / or other components of hemodynamic monitoring system 32. In some examples, user interface 54 may take the form of a graphical user interface (GUI) that presents graphical control elements presented, for example, on a touch-sensitive and / or presence-sensitive display screen of display device 12. In such examples, user input may be received in the form of gesture input, such as touch gestures, scroll gestures, zoom gestures, or other gesture input. In certain examples, user interface 54 may take the form of and / or include physical control elements, such as physical buttons, keys, knobs, or other physical control elements configured to receive user input that interacts with components of hemodynamic monitoring system 32.

[0028] During operation, the hemodynamic sensor 34 senses hemodynamic data representative of the arterial pressure waveform of the patient 36. The hemodynamic sensor 34 provides the hemodynamic data (e.g., as analog sensor data) to the hemodynamic monitor 10. The ADC 44 converts the analog hemodynamic data to digital hemodynamic data representative of the patient's arterial pressure waveform.

[0029] The nociceptive software code 48 may include nociceptive detection software code. The system processor 40 executes the nociceptive detection software code of the nociceptive software code 48 to determine a nociceptive detection score representative of the probability of a current nociceptive event for the patient 36 using the received hemodynamic data. For example, the system processor 40 may execute a first module 50 to perform a waveform analysis of the hemodynamic data to determine a plurality of signal measurements. The system processor 40 executes a second module 51 to extract nociceptive detection input features from the plurality of signal measurements that detect a nociceptive event for the patient 36. The system processor 40 executes a third module 52 to determine a nociceptive detection score representative of the probability of a nociceptive event for the patient 36 based on the nociceptive detection input features. If the nociceptive detection score meets a predetermined detection criterion, the system processor 40 activates a sensory alarm 58 of the user interface 54 to transmit a first sensory signal to alert the medical personnel 38 that the patient 36 is currently experiencing a current nociceptive event. A medical professional 38 may respond to the alert by administering a painkiller to the patient 36, or by administering any other form of treatment to the patient 36, to alleviate the current nociceptive event.

[0030] The nociceptive software code 48 may also include nociceptive prediction software code. The system processor 40 executes the nociceptive prediction software code of the nociceptive software code 48 to determine a nociceptive prediction score representative of the probability of a future nociceptive event for the patient 36 using the received hemodynamic data. For example, the system processor 40 may execute a first module 50 to perform a waveform analysis of the hemodynamic data to determine a plurality of signal measurements. The system processor 40 executes a second module 51 to extract nociceptive prediction input features from the plurality of signal measurements that predict a future nociceptive event for the patient 36. The system processor 40 executes a third module 52 to determine a nociceptive prediction score representative of the probability of a future nociceptive event for the patient 36 based on the nociceptive prediction input features. If the nociceptive prediction score meets a predetermined prediction criterion, the system processor 40 activates a sensory alarm 58 on the user interface 54 to transmit a second sensory signal to alert the medical personnel 38 that the patient 36 will soon experience a future nociceptive event. A medical professional 38 can respond to this alert by administering a painkiller to the patient 36, or by administering any other form of treatment to the patient 36, to mitigate or prevent the onset of a predicted future nociceptive event.

[0031] In addition to detecting current nociceptive events and predicting future nociceptive events in the patient 36, the hemodynamic monitoring system 32 can identify when the patient 36 is experiencing a current nociceptive event and when the patient 36 is only responding to a hemodynamic drug previously administered to the patient 36 by the medical professional 38 (i.e., a hemodynamic drug administration event). A hemodynamic drug administration event is defined as an event where the patient 36 experiences an increase in heart rate and blood pressure due to administration of a compound (e.g., a vasopressor, inotrope, fluid, and / or other agent) that alters cardiovascular hemodynamics and induces a sympathetic response that mimics the sympathetic response of nociception, but is not a nociceptive event in the patient 36. The nociceptive software code 48 includes hemodynamic drug detection software code for detecting the presence of a hemodynamic drug administration event in the patient 36. The system processor 40 executes the hemodynamic drug detection software code of the nociceptive software code 48 to determine a hemodynamic drug detection score using the received hemodynamic data, the hemodynamic drug detection score representing the probability that a hemodynamic drug administration event is responsible for an increase in the heart rate and blood pressure of the patient 36. For example, the system processor 40 may execute a first module 50 to perform a waveform analysis of the hemodynamic data to determine a plurality of signal measurements. The system processor 40 executes a second module 51 to extract hemodynamic drug detection input features from the plurality of signal measurements that detect a current effect of a hemodynamic drug administration event in the patient 36. The system processor 40 executes a third module 52 to determine a hemodynamic drug detection score for the patient 36 based on the hemodynamic drug detection input features. If the hemodynamic drug detection score meets a predetermined hemodynamic detection criterion, the system processor 40 activates a sensory alarm 58 in the user interface 54 to transmit a third sensory signal to alert the medical personnel 38 that the patient 36 is experiencing a hemodynamic drug administration event rather than a current nociceptive event. The hemodynamic drug detection score and the third sensory signal help prevent the medical professional 38 from confusing hemodynamic drug administration events with nociceptive events and prevent the medical professional 38 from administering pain medication to the patient 36 unnecessarily.

[0032] The nociception software code 48 also includes hemodynamic drug prediction software code for detecting the onset of an effect from a future hemodynamic drug administration event for the patient 36 (i.e., a sympathetic response that affects a hemodynamic parameter). The system processor 40 executes the hemodynamic drug prediction software code of the nociception software code 48 to determine a hemodynamic drug prediction score using the received hemodynamic data, the hemodynamic drug prediction score representing a probability that a hemodynamic drug administration event is responsible for an increase in the heart rate and an increase in blood pressure for the patient 36. For example, the system processor 40 may execute a first module 50 to perform a waveform analysis of the hemodynamic data to determine a plurality of signal measurements. The system processor 40 executes a second module 51 to extract predictive input features (hereinafter referred to as "hemodynamic drug prediction input features") associated with the administration of a hemodynamic drug that detects the onset of an effect from a hemodynamic drug administration event for the patient 36 from the plurality of signal measurements. The system processor 40 executes a third module 52 to determine a hemodynamic drug prediction score for the patient 36 based on the hemodynamic drug prediction input features. If the hemodynamic drug prediction score meets predetermined hemodynamic prediction criteria, the system processor 40 activates a sensory alarm 58 on the user interface 54 and transmits a fourth sensory signal to alert the medical practitioner 38 that the patient 36 will experience a future effect from the hemodynamic drug administration event. The hemodynamic drug prediction score and the fourth sensory signal help prevent the medical practitioner 38 from confusing a future hemodynamic drug administration event with a future nociceptive event and from administering unnecessary analgesic medication to the patient 36.

[0033] The system memory 42 of the hemodynamic monitor 10 may also include stable detection software code for detecting stable episodes of the patient 36. A stable episode is defined as a period during which the patient 36 does not experience a nociceptive event or a hemodynamic drug administration event. The stable detection software code may be a subpart of the nociceptive software code 48. The system processor 40 executes the stable detection software code to extract stable detection input features from the plurality of signal measurements. The stable detection software code may use a second module 51 to extract stable detection input features from the plurality of signal measurements. The stable detection input features detect stable episodes of the patient 36. The system processor 40 executes a third module 52 to determine a stable score for the patient 36 based on the stable detection input features. The system processor 40 outputs the stable score for the patient 36 to a user interface 54 of the display device 12.

[0034] The system processor 40 may execute the first module 50 to extract a single batch of signal measurements for a given unit of time, which may be used by the second module 51 to extract all of the nociception detection input features, nociception prediction input features, hemodynamic drug detection input features, hemodynamic drug prediction input features, and stability detection input features for that unit of time. The second module 51 may simultaneously extract all of the nociception detection input features, nociception prediction input features, hemodynamic drug detection input features, hemodynamic drug prediction input features, and stability detection input features from the multiple signal measurements. The system processor 40 may execute the third module 52 to simultaneously determine a nociception detection score, a nociception prediction score, a hemodynamic drug detection score, a hemodynamic drug prediction score, and a stability score. The nociception software code 48 of the hemodynamic monitor 10 may utilize a classification-based machine learning model with binary positive vs. negative labels in some implementations. The processor 40, in certain examples, can output the nociceptive detection score and the hemodynamic drug detection score together on the display device 12 to compare and contrast the two probabilities and help the medical professional 38 better understand whether a nociceptive event or a hemodynamic drug administration event is causing an increase in blood pressure and heart rate in the patient 36.

[0035] Alternatively, the nociceptive software code 48 of the hemodynamic monitor 10 may utilize a multi-class machine learning model having three labels: nociceptive events, hemodynamic drug events, and stable episodes. For example, the processor 40 may output the nociceptive detection score along with both the stable score and the hemodynamic drug detection score to the display device 12 such that all three probabilities are compared to one another: the probability that the patient is undergoing a current nociceptive event, the probability that the patient is experiencing a current hemodynamic drug administration event, and the probability that the patient is stable. As discussed below with reference to Figures 5-8, the machine learning model of the hemodynamic monitor 10 may be trained to recognize and / or predict these types of events and episodes in the arterial pressure waveform of the patient 36 using a clinical dataset that includes arterial pressure waveforms and clinical annotations of administration of compounds (e.g., vasopressors, inotropes, fluids, and / or other agents) that alter cardiovascular hemodynamics.

[0036] FIG. 5 is an illustration of a clinical dataset 60 used for data mining and machine training of the hemodynamic monitor 10. The clinical dataset 60 includes a first dataset 61 that includes a collection of arterial pressure waveforms recorded from previous patients. The first dataset 61 may be collected by an invasive hemodynamic sensor, such as the hemodynamic sensor 16 shown in FIG. 2, or may be collected by a non-invasive hemodynamic sensor, such as the hemodynamic sensor 26 shown in FIG. 3. The clinical dataset 60 also includes a second dataset 62 that includes a log of instances in which compounds that alter cardiovascular hemodynamics (e.g., vasopressors, inotropes, fluids, and / or other medications) were administered to the patients of the first dataset 61 while their arterial pressure waveforms were being recorded. A medical professional can enter the administration information directly into the same hemodynamic monitor that is collecting the first dataset 61, such that the first dataset 61 and the second dataset 62 are collected together at the same time. As shown in FIGS. 6-8, information from a second dataset 62 is annotated and labeled to the collection of arterial pressure waveforms from a first dataset 61.

[0037] FIG. 6 is a graph showing a plot of systolic blood pressure over time (hereinafter referred to as an "SBP plot") and a plot of heart rate over time (hereinafter referred to as an "HR plot"). Before the clinical data set 60 can be used to machine train the hemodynamic monitor 10, an SBP plot and an HR plot are determined for each of the arterial pressure waveforms collected in the clinical data set 60. The SBP plot and the HR plot shown in FIG. 6 are examples from one of the arterial pressure waveforms (not shown) of the clinical data set 60. After the SBP plot and the HR plot are determined for each of the arterial pressure waveforms collected in the clinical data set 60, both the SBP plot and the HR plot are annotated to indicate when a compound that alters cardiovascular hemodynamics was administered to the clinical patient. For example, the SBP plot and the HR plot shown in FIG. 6 include an analgesic label 64, which is a vertical bar that extends across the SBP plot and the HR plot at the same position in time. The painkiller label 64 in FIG. 6 indicates that the clinical patient was administered a painkiller during the time period represented by the SBP and HR plots in FIG. 6. After the SBP and HR plots have been annotated and labeled to indicate administration of the drug to the clinical patient, a nociceptive data segment 66 is identified and labeled on the SBP and HR plots. In this particular case, a drug that induces a sympathetic response was not administered to the patient prior to the increase in systolic blood pressure and heart rate, causing a nociceptive event (labeled as data segment 66). A painkiller was administered after the increase in systolic blood pressure and heart rate to reduce the effects of the nociceptive event, after which both parameters returned to their respective baseline values.

[0038] As shown in FIG. 6, the nociceptive data segment 66 is identified on the SBP plot and the HR plot by finding a time segment in both the SBP plot and the HR plot during which the systolic blood pressure of the clinical patient has increased by at least a threshold amount (e.g., 20% or other threshold amount) compared to a previous time period, the heart rate of the clinical patient has also increased by at least a threshold amount (e.g., 20% or other threshold amount) compared to a previous time period, and an infusion of a compound that alters cardiovascular hemodynamics by inducing a sympathetic response (e.g., a vasopressor, an inotrope, a fluid, and / or other agent) has not been initiated prior to the increase in blood pressure and heart rate. In FIG. 6, both the SBP plot and the HR plot increase by more than 20% at a start point 68, which occurs prior to the analgesic label 64, thus indicating the start of the nociceptive data segment 66 of FIG. 6. The nociceptive data segment 66 of FIG. 6 continues until both the SBP plot and the HR plot begin to decline as a result of the analgesic administered in good time to the clinical patient at the analgesic label 64. The decline in the SBP plot and the HR plot is indicated by an end point 70. Both the start point 68 and the end point 70 are labeled on the HR plot and the SBP plot, and the time segment between the start point 68 and the end point 70 is designated as one nociceptive data segment 66. Once the analgesic label 64 and the nociceptive data segment 66 have been identified on the SBP plot and the HR plot, the arterial pressure waveform used to generate the SBP plot and the HR plot of FIG. 6 may also be annotated and labeled to indicate when the analgesic label 64 and the nociceptive data segment 66 occurred on the arterial pressure waveform. Once labeled with the analgesic label 64 and the nociceptive data segment 66, the arterial pressure waveform is ready to be used for data mining and machine training of the hemodynamic monitor 10 to detect nociceptive events. As discussed further below with reference to FIGS. 9-10 , waveform analysis is performed on the clinical dataset 60 including the nociceptive data segment 66 to calculate a number of signal measurements that are then used to calculate a nociceptive detection input feature that best detects the probability of a current nociceptive event.

[0039] The predictive data segment 71 of FIG. 6 may also be used in machine training of the hemodynamic monitor 10 to predict future nociceptive events. The predictive data segment 71 may be identified within the clinical dataset 60 by identifying a prior period prior to the onset of an increase in the SBP and HR plots. In the example of FIG. 6, the prior period occurs prior to the start point 68. The prior period prior to the start point 68 is labeled as the predictive data segment 71. In some examples, the predictive data segment 71 includes a period beginning 15 minutes prior to the nociceptive data segment 66 and ending immediately prior to the nociceptive data segment 66. In other examples, the predictive data segment 71 may include a greater or lesser period prior to the nociceptive data segment 66. Once the predictive data segment 71 has been identified and labeled on the SBP and HR plots, the arterial pressure waveform used to generate the SBP and HR plots of FIG. 6 may also be annotated and labeled to indicate when the predictive data segment 71 occurred on the arterial pressure waveform. Once labeled with the predictive data segments 71, the arterial pressure waveform is ready to be used for data mining and machine training of the hemodynamic monitor 10 to predict nociceptive events. As discussed further below with reference to Figures 9-10, waveform analysis is performed on the clinical dataset 60, including the predictive data segments 71, to calculate a number of signal measurements that are then used to calculate nociceptive predictive input features that best detect the probability of the onset of a future nociceptive event.

[0040] FIG. 7 is a graph showing another SBP plot and HR plot derived from an arterial pressure waveform segment (not shown) from clinical data set 60. The arterial pressure waveform segment that generated the SBP plot and HR plot of FIG. 7 can be identified as a stable data segment 72 and used for data mining and machine training of hemodynamic monitor 10 to detect when a patient is experiencing a stable episode without nociception. If the SBP plot does not have an increase greater than a threshold amount (e.g., 20% or other threshold amount), the HR plot does not have an increase greater than a threshold amount (e.g., 20% or other threshold amount), and an infusion of a compound that alters cardiovascular hemodynamics is not performed, then the arterial pressure waveform segment of clinical data set 60 is identified as a stable data segment 72. As shown in the example of FIG. 7, the SBP plot does not include an increase greater than 20% between start point 74 and end point 76. The HR plot of the example of FIG. 7 also does not include an increase greater than 20% between start point 74 and end point 76. The HR and SBP plots of FIG. 7 also do not include any annotations or labels indicating the infusion of a compound that alters cardiovascular hemodynamics in the clinical patient between the start point 74 and the end point 76. Given the above-mentioned characteristics of the HR and SBP plots in the example of FIG. 7, the HR and SBP plots of FIG. 7 are labeled as a stable data segment 72 between the start point 74 and the end point 76. The arterial pressure waveform segment (not shown) that generated the HR and SBP plots of FIG. 7 is also labeled as a stable data segment 72 between the start point 74 and the end point 76. Once labeled with a stable data segment 72, the arterial pressure waveform segment is ready to be used for stable data mining and stable machine training of the hemodynamic monitor 10. As will be further discussed below with reference to FIGS. 9-10, waveform analysis is performed on the clinical dataset 60 including the stable data segment 72 to calculate a number of signal measurements that are then used to calculate a stable detection input feature that best detects the probability of a stable episode.

[0041] FIG. 8 is a graph showing another SBP plot and HR plot derived from an arterial pressure waveform segment (not shown) from the clinical data set 60. The arterial pressure waveform segment that generated the SBP plot and HR plot of FIG. 8 can be identified as a hemodynamic drug administration data segment 78 (hereinafter referred to as "HDA data segment 78") and used in data mining and machine training of the hemodynamic monitor 10 to detect when the patient is experiencing a current hemodynamic drug administration event. An arterial pressure waveform segment of the clinical data set 60 is identified as an HDA data segment 78 if the arterial pressure waveform segment includes an infusion of a compound that alters cardiovascular hemodynamics into the clinical patient and there is an increase in both the SBP plot and the HR plot after the infusion by at least a threshold amount (e.g., 20% or other threshold amount). In the example of FIG. 8, a vasopressor infusion label 80 on the SBP plot and the HR plot indicates that the clinical patient was administered a vasopressor. Immediately after the vasopressor infusion label 80, the SBP plot increased by at least 20% between the start point 82 and the end point 84. Between the start point 82 and the end point 84, the HR plot also increased by at least 20%, thus indicating that an HDA data segment 78 occurred between the start point 82 and the end point 84. The HR and SBP plots of FIG. 8 are labeled as the HDA data segment 78 between the start point 82 and the end point 84. The arterial pressure waveform segment (not shown) that generated the HR and SBP plots of FIG. 8 is also labeled as the HDA data segment 78 between the start point 82 and the end point 84. Once labeled with the HDA data segment 78, the arterial pressure waveform segment is ready to be used for hemodynamic drug detection data mining and machine training of the hemodynamic monitor 10. As discussed further below with reference to FIGS. 9-10, waveform analysis is performed on the clinical dataset 60 including the HDA data segment 78 to calculate a number of signal measurements that are then used to calculate a hemodynamic drug detection input feature that best detects the probability of a current hemodynamic drug administration event.

[0042] The hemodynamic drug prediction data segment 85 (hereinafter "HDP data segment 85") of FIG. 8 may also be used for machine training of the hemodynamic monitor 10 to predict the hemodynamic effect on the patient 36 of a future drug administration event. The HDP data segment 85 may be identified in FIG. 8 within the clinical dataset 60 by identifying a previous period of time prior to both the infusion of a vasopressor (as indicated by label 80) and the onset of both the increase in the SBP and HR plots. In the example of FIG. 8, the previous period occurs prior to start point 82. The previous period prior to start point 82 is labeled as the HDP data segment 85. In some examples, the HDP data segment 85 includes a period of time beginning 15 minutes prior to the HDA data segment 78 and ending immediately prior to the HDA data segment 78. In other examples, the HDP data segment 85 may include a greater or lesser period of time prior to the HDA data segment 78. Once the HDP data segments 85 have been identified and labeled on the SBP and HR plots, the arterial pressure waveform used to generate the SBP and HR plots of FIG. 8 may also be annotated and labeled to indicate when the HDP data segments 85 occurred on the arterial pressure waveform. Once labeled with the HDP data segments 85, the arterial pressure waveform is ready to be used for data mining and machine training of the hemodynamic monitor 10 to predict the onset of a future hemodynamic drug delivery event. As discussed further below with reference to FIGS. 9-10 , waveform analysis is performed on the clinical dataset 60, including the HDP data segments 85, to calculate a number of signal measurements that are then used to calculate hemodynamic drug prediction input features that best detect the probability of onset of effect from a future hemodynamic drug delivery event.

[0043] FIG. 9 is a flow diagram of a method 86 for data mining of the clinical dataset 60 from FIGS. 5-8 for machine training of a machine learning model of the hemodynamic monitor 10. The method 86 of FIG. 9 is considered with reference to FIG. 10 as well. The method 86 is applied to each of the nociceptive data segment 66 (shown in FIG. 6), the predictive data segment 71 (shown in FIG. 6), the stable data segment 72 (shown in FIG. 7), the HDA data segment 78 (shown in FIG. 8), and the HDP data segment 85 (shown in FIG. 8) of the clinical dataset 60 to train the hemodynamic monitor to find input features as previously described with reference to FIG. 4. The method 86 is described as being applied to the nociceptive data segment 66 (shown in FIG. 6).

[0044] To machine train the hemodynamic monitor 10 to identify the nociceptive detection input features described in FIG. 4, the nociceptive detection input features are first determined by applying a method 86 to the nociceptive data segments 66 of the clinical dataset 60. A first step 88 of the method 86 is to perform a waveform analysis of the nociceptive data segments 66 of the arterial waveforms collected in the dataset 60 to calculate a plurality of signal measurements of the nociceptive data segments. Performing a waveform analysis of the nociceptive data segments 66 may include identifying individual cardiac cycles in each of the arterial pressure waveforms of the nociceptive data segments 66. FIG. 10 provides an exemplary graph showing an example trace of an arterial pressure waveform with individual cardiac cycles identified and expanded. Performing a waveform analysis of the nociceptive data segments 66 may then include identifying dicrotic notches in each of the individual cardiac cycles of each of the arterial pressure waveforms of the nociceptive data segments 66, similar to the example shown in FIG. 10. Next, waveform analysis of the nociceptive data segment 66 involves identifying the systolic upstroke phase, the systolic downstroke phase, and the diastole phase in each individual cardiac cycle of each arterial pressure waveform of the nociceptive data segment 66, similar to the example shown in FIG. 10.

[0045] Signal measurements are extracted from each of the systolic up phase, systolic decay phase, and diastole from each of the individual cardiac cycles of each of the arterial pressure waveforms of the nociceptive data segment 66. The signal measurements may correspond to hemodynamic effects from each of the systolic up phase, systolic decay phase, and diastole from each of the individual cardiac cycles. These hemodynamic effects may include contractility, aortic compliance, stroke volume, vascular tone, afterload, and the entire cardiac cycle. The signal measurements calculated or extracted by the waveform analysis of the first step 88 of method 86 include mean, maximum, minimum, duration, area, standard deviation, derivative, and / or morphological measurements from each of the systolic up phase, systolic decay phase, and diastole from each of the individual cardiac cycles. The signal measurements may also include heart rate, respiratory rate, stroke volume, pulse pressure, pulse pressure variability, stroke volume variability, mean arterial pressure (MAP), systolic pressure (SYS), diastolic pressure (DIA), heart rate variability, cardiac output, peripheral resistance, vascular compliance, and / or left ventricular contractility extracted from each individual cardiac cycle of each arterial pressure waveform of the nociceptive data segment 66.

[0046] After the signal measurements for the nociceptive data segment 66 have been determined, step 90 of method 86 is performed on the signal measurements of the nociceptive data segment 66. Step 90 of method 86 is to calculate combination measurements between each of the signal measurements of the nociceptive data segment 66. Calculating combination measurements between the signal measurements of the nociceptive data segment 66 may include performing steps 92, 94, 96, and 98 shown in FIG. 9 for all of the signal measurements of the nociceptive data segment 66.

[0047] Step 92 of method 86 is to arbitrarily select three signal measurements from the signal measurements of the nociceptive data segment. Next, in step 94 of method 86 shown in FIG. 9, powers of different orders are calculated for each of the three signal measurements to generate a power of the three signal measurements. In step 96 of method 86, the powers of the three signal measurements are then multiplied together to generate a product of the powers of the three signal measurements. Step 98 of method 86 is to perform a receiver operating characteristic (ROC) analysis of the product of the powers to arrive at a combined measurement for the three signal measurements. Steps 92, 94, 96, and 98 are repeated until all of the combined measurements have been calculated between all of the signal measurements of the nociceptive data segment 66.

[0048] The signal measurements having the most predictive top combined measurements (i.e., combined measurements that meet the threshold prediction criteria) are selected in step 100 of method 86 to perform machine learning. The signal measurements having the most predictive top combined measurements are the top signal measurements for the nociceptive data segment 66 and are labeled as nociceptive detection input features. Once the nociceptive detection input features have been determined, the hemodynamic monitor 10 is trained or programmed to perform waveform analysis on the arterial pressure waveform of the patient 36 (shown in FIG. 4 ), extract nociceptive detection input features from the arterial pressure waveform of the patient 36, and use these nociceptive detection input features to determine the probability that the patient 36 is currently experiencing a current nociceptive event.

[0049] Similar to how method 86 was applied to nociceptive data segment 66, method 86 is applied to predicted data segment 71, stable data segment 72, HDA data segment 78, and HDP data segment 85 of clinical dataset 60, respectively, to determine nociceptive predictive input features, stable detection input features, hemodynamic drug detection input features, and hemodynamic drug predictive input features.

[0050] FIG. 11 is an illustration of an output function 108 having three machine models (first model 102, second model 104, and third model 106) that can be trained with various input features identified using method 86 (FIG. 9) to determine the probability of a nociceptive event, and more specifically, to distinguish nociceptive events from hemodynamic drug events. Model training involves binary classification labeling of data segments obtained from clinical dataset 60. Output function 108, including first model 102, second model 104, and third model 106, is included in nociceptive software code 48 and stored in system memory 42 (both shown in FIG. 4). First module 50, second module 51, and third module 52 of nociceptive software code 48 can be used by system processor 40 of hemodynamic monitor 10 to execute first model 102, second model 104, and third model 106 to calculate output function 108.

[0051] The first model 102 can be trained to utilize the HDP data segments 85 and the stable data segments 72 to predict effects from future administration of hemodynamic drugs (e.g., vasopressors and inotropes). As already discussed above with reference to FIG. 8, each HDP data segment 85 is identified within the clinical dataset 60 by identifying a previous period prior to the onset of the increase in the SBP and HR plots of FIG. 8. In the example of FIG. 8, each HDP data segment 85 includes a period beginning 15 minutes prior to the HDA data segment 78 and ending immediately prior to the HDA data segment 78. In other examples, the HDP data segment 85 may include a greater or lesser period prior to the HDA data segment 78. After the HDP data segments 85 are identified within the clinical dataset 60, the first model 102 is machine trained to identify hemodynamic drug prediction input features using steps 88, 90, 92, 94, 96, 98, and 100 of the method 86 described above with reference to FIGS. 9 and 10. Once the first model 102 has been machine trained to identify hemodynamic drug predictive input features from the HDP data segments 85, the first model 102 labels the hemodynamic drug predictive input features as positive.

[0052] As already discussed above with reference to Figure 7, each of the stable data segments 72 is identified within the clinical dataset 60 by identifying an arterial pressure waveform segment where there is no increase in the SBP plot greater than a threshold amount (e.g., 20% or other threshold amount), no increase in the HR plot greater than a threshold amount (e.g., 20% or other threshold amount), and no infusion of a compound that alters cardiovascular hemodynamics is performed. After the stable data segments 72 are identified within the clinical dataset 60, the first model 102 is machine trained to identify stable detection input features using steps 88, 90, 92, 94, 96, 98, and 100 of the method 86 described above with reference to Figures 9 and 10. Once the first model 102 is machine trained to identify stable detection input features from the stable data segments 72, the first model 102 labels the stable detection input features as negative.

[0053] With the first model 102 trained and the hemodynamic drug prediction input features labeled as positive and the stable detection input features labeled as negative, the arterial pressure waveform of the patient 36 can be fed into the first model 102, which extracts the positive values ​​of the hemodynamic drug prediction input features and the negative values ​​of the stable detection input features of the patient 36. The first model 102 calculates a probability P1, which is the probability of a hemodynamic drug administration event of the patient 36, by taking the sum between the positive values ​​of the hemodynamic drug prediction input features and the negative values ​​of the stable detection input features. If the value of the probability P1 is positive, the first model 102 indicates that the patient 36 will experience a hemodynamic drug administration event if such a drug is administered. If the value of the probability P1 is negative, the first model 102 indicates that the patient 36 is experiencing a stable event. Both the positive values ​​of the hemodynamic drug prediction input features and the negative values ​​of the stable detection input features of the patient 36 may be normalized by the first model 102 before the first model 102 calculates the sum between the positive values ​​of the hemodynamic drug prediction input features and the negative values ​​of the stable detection input features. Normalizing both the positive values ​​of the hemodynamic drug detection (or prediction) input features and the negative values ​​of the stable detection input features of the patient 36 before calculating the probability P1 ensures that both the hemodynamic drug prediction input features and the stable detection input features are appropriately emphasized or weighted within the first model 102.

[0054] The second model 104 may be trained to distinguish whether the patient 36 is experiencing a current nociceptive event or a current hemodynamic drug event. The nociceptive data segments 66 and the HDA data segments 78 from the clinical dataset 60 are used to machine train the second model 104. As already discussed above with reference to FIG. 6, each nociceptive data segment 66 is identified within the clinical dataset 60 by finding time segments in both the SBP and HR plots during which the clinical patient's systolic blood pressure increased by at least a threshold amount (e.g., 20% or other threshold amount) compared to the previous period, the clinical patient's heart rate also increased by at least a threshold amount (e.g., 20% or other threshold amount) compared to the previous period, and an infusion of a compound that alters cardiovascular hemodynamics (e.g., a vasopressor, inotrope, fluid, and / or other agent) was not initiated prior to the increase in blood pressure and heart rate. After the nociceptive data segments 66 have been identified within the clinical dataset 60, the second model 104 is machine trained to identify nociceptive detection input features using steps 88, 90, 92, 94, 96, 98, and 100 of the method 86 described above with reference to Figures 9 and 10. Once the second model 104 has been machine trained to identify nociceptive detection input features from the nociceptive data segments 66, the second model 104 labels the nociceptive detection input features as positive.

[0055] As already discussed above with reference to Figure 8, each HDA data segment 78 is identified in the clinical dataset 60 when the arterial pressure waveform data segment includes an infusion of a compound that alters cardiovascular hemodynamics into a clinical patient with a subsequent increase in both the SBP plot and the HR plot of at least a threshold amount (e.g., 20% or other threshold amount). After the HDA data segment 78 is identified in the clinical dataset 60, the second model 104 is machine trained to identify hemodynamic drug detection input features using steps 88, 90, 92, 94, 96, 98, and 100 of the method 86 described above with reference to Figures 9 and 10. Once the second model 104 is machine trained to identify hemodynamic drug detection input features from the HDA data segment 78, the second model 104 labels the hemodynamic drug detection input features as negative.

[0056] With the second model 104 trained and the nociceptive detection input features labeled as positive and the hemodynamic drug detection input features labeled as negative, the arterial pressure waveform of the patient 36 may be fed into the second model 104, which extracts the positive values ​​of the nociceptive detection input features and the negative values ​​of the hemodynamic drug detection input features of the patient 36. The second model 104 calculates a probability P2, which is the probability of the patient 36's current nociceptive event relative to the patient's current hemodynamic drug administration event, by taking the sum between the positive values ​​of the nociceptive detection input features and the negative values ​​of the hemodynamic drug detection input features. If the value of the probability P2 is positive, the second model 104 indicates that the patient 36 is experiencing a current nociceptive event. If the value of the probability P2 is negative, the second model 104 indicates that the patient 36 is experiencing a current hemodynamic drug administration event. Both the positive values ​​of the nociceptive detection input features and the negative values ​​of the hemodynamic drug detection input features of the patient 36 may be normalized by the second model 104 before the second model 104 calculates the sum between the positive values ​​of the nociceptive detection input features and the negative values ​​of the hemodynamic drug detection input features. Normalizing both the positive values ​​of the nociceptive detection input features and the negative values ​​of the hemodynamic drug detection input features of the patient 36 before calculating the probability P2 ensures that both the nociceptive detection input features and the hemodynamic drug detection input features are appropriately emphasized or weighted within the second model 104.

[0057] The third model 106 may be trained to detect whether the patient 36 is experiencing a current nociceptive event or a stable event. The nociceptive data segments 66 and the stable data segments 72 from the clinical dataset 60 are used to machine train the third model 106. The third model 106 is machine trained to apply steps 88, 90, 92, 94, 96, 98, and 100 of the method 86 described above with reference to Figures 9 and 10 to the nociceptive data segments 66 to identify nociceptive detection input features. Once the third model 106 is machine trained to identify nociceptive detection input features from the nociceptive data segments 66, the third model 106 labels the nociceptive detection input features as positive. The third model 106 is also machine trained to apply steps 88, 90, 92, 94, 96, 98, and 100 of the method 86 described above with reference to Figures 9 and 10 to the stable data segments 72 to identify stable detection input features. Once the third model 106 has been machine trained to identify stable-detection input features from the stable data segments 72, the third model 106 labels the stable-detection input features as negative.

[0058] With the third model 106 trained and the nociceptive detection input features labeled as positive and the stable detection input features labeled as negative, the arterial pressure waveform of the patient 36 may be fed into the third model 106, which extracts the positive values ​​of the nociceptive detection input features and the negative values ​​of the stable detection input features of the patient 36. The third model 106 calculates a probability P3, which is the probability of the patient 36's current nociceptive event relative to a stable event, by taking the sum between the positive values ​​of the nociceptive detection input features and the negative values ​​of the stable detection input features. If the value of the probability P3 is positive, the third model 106 indicates that the patient 36 is experiencing a current nociceptive event. If the value of the probability P3 is negative, the third model 106 indicates that the patient 36 is experiencing a stable event. Both the positive values ​​of the nociceptive detection input features and the negative values ​​of the stable detection input features of the patient 36 may be normalized by the third model 106 before the third model 106 calculates the sum between the positive values ​​of the nociceptive detection input features and the negative values ​​of the stable detection input features. Normalizing both the positive values ​​of the nociceptive detection input features and the negative values ​​of the stable detection input features of the patient 36 before calculating the probability P3 ensures that both the nociceptive detection input features and the stable detection input features are appropriately emphasized or weighted within the third model 106.

[0059] The probabilities P1, P2, and P3 are combined to generate an output function 108. The output function 108 represents the probability that the patient 36 is experiencing a nociceptive event rather than a hemodynamic drug event. The predicted probability of nociception in the patient 36 generated by the output function 108 may be displayed by the hemodynamic monitor 10. The probability P1 from the first model 102 and the probability P2 from the second model 104 act as a false positive check for the probability P3 to verify whether the current nociceptive event detected by the probability P3 is in fact a current nociceptive event in the patient 36 and is not a current hemodynamic drug event that has been falsely identified as a nociceptive event. For example, if probability P3 indicates that the patient 36 is experiencing a nociceptive event, but probability P1 indicates that the patient 36 will soon experience a future hemodynamic drug delivery event, and probability P2 indicates that the patient 36 is experiencing a current hemodynamic drug delivery event, then output function 108 indicates on display 12 of hemodynamic monitor 10 that the patient 36 is experiencing a current hemodynamic drug delivery event and not a nociceptive event. In another example, if probability P3 indicates that the patient 36 is experiencing a nociceptive event, probability P1 indicates that the patient 36 will not soon experience a future hemodynamic drug delivery event, and probability P2 indicates that the patient 36 is experiencing a current nociceptive event, output function 108 and system processor 40 activate sensory alarm 58 of user interface 54 to send a first sensory signal alerting medical personnel 38 that the patient 36 is currently experiencing a current nociceptive event. A medical professional 38 may respond to the alert by administering a painkiller to the patient 36, or by administering any other form of treatment to the patient 36, to alleviate the current nociceptive event.

[0060] Discussion of Possible Embodiments

[0061] The following is a non-exclusive description of possible embodiments of the present invention.

[0062] 1. A system for monitoring a patient's arterial pressure and providing alerts to a healthcare professional of nociception in the patient, the system comprising: a hemodynamic sensor that generates hemodynamic data representative of the patient's arterial pressure waveform; a system memory that stores nociception software code; a user interface including a sensory alarm that provides a sensory signal to alert the healthcare professional of a nociceptive event in the patient; and a hardware processor that executes the nociception software code to perform waveform analysis of the hemodynamic data to determine a plurality of signal measurements, extracting from the plurality of signal measurements hemodynamic drug detection input features indicative of the nociceptive event in the patient, extracting from the plurality of signal measurements hemodynamic drug detection input features indicative of a hemodynamic drug administration event in the patient, extracting from the plurality of signal measurements hemodynamic drug prediction input features indicative of a future hemodynamic drug administration event for the patient, and extracting from the plurality of signal measurements stable detection input features indicative of a stable episode for the patient. a hardware processor configured to: extract from a fixed value a probability representing a probability that the patient will experience an effect from the future hemodynamic drug administration event; determine a first probability based on the hemodynamic drug prediction input features and the stable detection input features, the first probability representing a probability that the patient will experience an effect from the future hemodynamic drug administration event; determine a second probability based on the detection input features and the hemodynamic drug detection input features, the second probability representing a probability that the patient will experience the current nociceptive event for the current hemodynamic drug administration event; determine a third probability based on the detection input features and the stable detection input features, the third probability representing a probability that the patient will experience the current nociceptive event for the stable episode; compare the third probability to the first probability and the second probability to determine an output probability of the current nociceptive event for the patient; and activate the sensory alarm of the user interface in response to the output probability meeting a predetermined detection criterion.

[0063] The system of the preceding paragraph may optionally additionally and / or alternatively include any one or more of the following features, configurations, and / or additional components.

[0064] The detection input features of the nociceptive software code are determined by detection machine training, the detection machine training comprising: collecting a clinical dataset comprising clinical annotations of an arterial pressure waveform and administration of a cardiovascular hemodynamic altering compound; identifying nociceptive data segments within the clinical dataset, each of the nociceptive data segments comprising an increase in blood pressure of at least a first threshold amount compared to a previous period of time and an increase in heart rate of at least a second threshold amount compared to the previous period of time, without an infusion of the cardiovascular hemodynamic altering compound having commenced prior to the increase in blood pressure and the increase in heart rate; the method comprising: identifying a start point and an end point of the increase in the heart rate; labeling the nociceptive data segment after the start point and during the increase in the blood pressure and the increase in the heart rate; performing a waveform analysis of the labeled nociceptive data segment to calculate a plurality of signal measures of the nociceptive data segment; determining the detected input feature by calculating a combination measure between the plurality of signal measures of the nociceptive data segment, selecting a top signal measure having a most predictive combination measure from the plurality of signal measures of the nociceptive data segment and labeling the top signal measure as the detected input feature.

[0065] The hemodynamic drug detection input features of the nociceptive software code are determined by hemodynamic drug detection machine training, the hemodynamic drug detection machine training comprising: identifying hemodynamic drug administration data segments within the clinical dataset, each of the hemodynamic drug administration data segments including an infusion of a compound that alters cardiovascular hemodynamics, an increase in blood pressure after the infusion of at least a third threshold amount, and an increase in heart rate after the infusion of at least a fourth threshold amount; identifying a start point and an end point of the increase in blood pressure and the increase in heart rate in each of the hemodynamic drug administration data segments; The method includes labeling the hemodynamic drug administration data segment after a starting point and during the increase in the blood pressure and the increase in the heart rate, performing a waveform analysis of the labeled hemodynamic drug administration data segment to calculate a plurality of signal measurements of the hemodynamic drug administration data segment, and determining the hemodynamic drug detection input feature by calculating a combination measurement between the plurality of signal measurements of the hemodynamic drug administration data segment and selecting a signal measurement having a most predictive combination measurement from the plurality of signal measurements of the hemodynamic drug administration data segment as the hemodynamic drug detection input feature.

[0066] The hemodynamic drug prediction input features of the nociceptive software code are determined by hemodynamic drug prediction machine training, the hemodynamic drug prediction machine training including identifying a previous period prior to the onset of the increase in the blood pressure and the increase in the heart rate in each of the hemodynamic drug administration data segments, labeling the previous period of each of the hemodynamic drug administration data segments as a hemodynamic drug prediction data segment, performing a waveform analysis of the hemodynamic drug prediction data segments to calculate a plurality of signal measurements for the hemodynamic drug prediction data segments, and determining the hemodynamic drug prediction input features by calculating combination measurements between the plurality of signal measurements for the hemodynamic drug prediction data segments and selecting a signal measurement having a most predictive combination measurement from the plurality of signal measurements for the hemodynamic drug prediction data segment as the hemodynamic drug prediction input feature.

[0067] The stable detection input features of the nociceptive software code are determined by stable detection machine training, the stable machine training comprising: identifying stable data segments in the clinical dataset, each of the stable data segments comprising a stable blood pressure without an increase greater than the first threshold amount over a set period of time, a stable heart rate without an increase greater than the second threshold amount over the set period of time, and no infusion of a compound that alters cardiovascular hemodynamics; identifying a start point and an end point of the stable blood pressure and the stable heart rate; labeling the stable data segments from the start point to the end point of the stable blood pressure and the stable heart rate; performing a waveform analysis of the labeled stable data segments to calculate a plurality of stable signal measurements for the stable data segments; and determining the stable detection input features by calculating combination measurements between the plurality of stable signal measurements and selecting the stable signal measurement having the most predictive combination measurement from the plurality of stable signal measurements as the stable detection input feature.

[0068] Performing waveform analysis of the labeled nociceptive data segments to calculate the plurality of signal measurements of the nociceptive data segments includes identifying individual cardiac cycles in the arterial pressure waveform of the clinical dataset, identifying a dicrotic notch in each of the individual cardiac cycles, identifying a systolic up phase, a systolic decay phase, and a diastole in each of the individual cardiac cycles, and extracting signal measurements from each of the systolic up phase, the systolic decay phase, and the diastole from each of the individual cardiac cycles.

[0069] The signal measurements correspond to hemodynamic effects from each of the systolic upstroke phase, the systolic downstroke phase, and the diastole phase from each of the individual cardiac cycles, the hemodynamic effects including contractility, aortic compliance, stroke volume, vascular tone, afterload, and the total cardiac cycle.

[0070] The signal measurements include mean, maximum, minimum, duration, area, standard deviation, derivative, and / or morphological measurements from each of the systolic upstroke phase, the systolic downstroke phase, and the diastole from each of the individual cardiac cycles.

[0071] The signal measurements include heart rate, respiratory rate, stroke volume, pulse pressure, pulse pressure variation, stroke volume variation, mean arterial pressure (MAP), systolic pressure (SYS), diastolic pressure (DIA), heart rate variability, cardiac output, peripheral resistance, vascular compliance, and / or left ventricular contractility extracted from each of the individual cardiac cycles. Calculating combination measures between the plurality of signal measurements of the nociceptive data segment includes performing step 1 by arbitrarily selecting three signal measurements from the plurality of signal measurements of the nociceptive data segment; performing step 2 by calculating different order powers for each of the three signal measurements to generate a power of the three signal measurements; performing step 3 by multiplying the powers of the three signal measurements together to generate a product of powers of the three signal measurements; performing step 4 by performing a Receiver Operating Characteristic (ROC) analysis of the product of powers to arrive at combination measures for the three signal measurements; and repeating steps 1, 2, 3, and 4 until all combination measures have been calculated between all of the plurality of signal measurements of the nociceptive data segment.

[0072] The hemodynamic sensor is a non-invasive hemodynamic sensor attachable to an extremity of the patient.

[0073] The hemodynamic sensor is a minimally invasive arterial catheter-based hemodynamic sensor.

[0074] The hemodynamic sensor generates the hemodynamic data as an analog hemodynamic sensor signal representative of the patient's arterial pressure waveform.

[0075] An analog-to-digital converter converts the analog hemodynamic sensor signal into digital hemodynamic data representative of the arterial pressure waveform of the patient.

[0076] 1. A method for monitoring a patient's arterial pressure and providing an alert to a medical professional of the patient's current nociception, comprising: receiving hemodynamic data sensed by a hemodynamic monitor representing the patient's arterial pressure waveform; performing, by the hemodynamic monitor, a waveform analysis of the sensed hemodynamic data to calculate a plurality of signal measurements of the sensed hemodynamic data; extracting, by the hemodynamic monitor, detection input features from the plurality of signal measurements indicative of a current nociceptive event of the patient; extracting, by the hemodynamic monitor, hemodynamic drug detection input features from the plurality of signal measurements indicative of a current hemodynamic drug administration event of the patient, the current hemodynamic drug administration event being defined as the patient experiencing an increase in heart rate and blood pressure due to administration of a compound that alters cardiovascular hemodynamics; and determining, by the hemodynamic monitor, that the patient is not experiencing a nociceptive event or a hemodynamic drug administration event. extracting stable detection input features from the plurality of signal measurements indicative of a stable episode of the patient; determining, by the hemodynamic monitor, a first probability based on the detection input features and the hemodynamic drug detection input features, the first probability representing a probability that the patient will experience the current nociceptive event for the current hemodynamic drug administration event; determining, by the hemodynamic monitor, a second probability based on the detection input features and the stable detection input features, the second probability representing a probability that the patient will experience the current nociceptive event for the stable episode; comparing, by the hemodynamic monitor, the second probability to the first probability to determine an output probability of the current nociceptive event for the patient; and activating a sensory alarm and generating a sensory signal by the hemodynamic monitor in response to the output probability satisfying a predetermined detection criterion.

[0077] The method of the preceding paragraphs may optionally, additionally and / or alternatively include any one or more of the following features, configurations, and / or additional components.

[0078] extracting, by the hemodynamic monitor, hemodynamic drug prediction input features from the plurality of signal measurements that predict an effect on the patient from a future hemodynamic drug administration event; determining, by the hemodynamic monitor, a third probability based on the hemodynamic drug prediction input features and the stable detection input features, the third probability representing a probability that the patient will experience an effect from the future hemodynamic drug administration event for the stable episode; comparing, by the hemodynamic monitor, the second probability to the first probability and the third probability to determine the output probability of the current nociceptive event for the patient; and activating a sensory alarm and generating a sensory signal, by the hemodynamic monitor, in response to the output probability satisfying the predetermined detection criterion.

[0079] Nociceptive detection training of the hemodynamic monitor to determine the detection input features, the nociceptive detection machine training comprising: collecting a clinical dataset comprising clinical annotations of an arterial pressure waveform and administration of a compound that alters cardiovascular hemodynamics; identifying nociceptive data segments within the clinical dataset, each of the nociceptive data segments comprising an increase in blood pressure of at least a first threshold amount compared to a previous period of time and an increase in heart rate of at least a second threshold amount compared to the previous period of time, and no infusion of the compound that alters cardiovascular hemodynamics was initiated prior to the increase in blood pressure and the increase in heart rate; 20. The method of claim 19, further comprising: identifying a start point and an end point of the increase; labeling the nociceptive data segment after the start point and during the increase in blood pressure and the increase in heart rate; performing a waveform analysis of the labeled nociceptive data segment to calculate a plurality of signal measures for the nociceptive data segment; and determining the detection input features by calculating combination measures between the plurality of signal measures for the nociceptive data segment, selecting a top signal measure having a most predictive combination measure from the plurality of signal measures for the nociceptive data segment, and labeling the top signal measure as the detection input feature.

[0080] training the hemodynamic monitor for determining the hemodynamic drug detection input features, comprising: identifying hemodynamic drug administration data segments within the clinical dataset, each of the hemodynamic drug administration data segments including an infusion of a compound that alters cardiovascular hemodynamics, an increase in blood pressure after the infusion of at least a third threshold amount, and an increase in heart rate after the infusion of at least a fourth threshold amount; identifying a start point and an end point of the increase in blood pressure and the increase in heart rate in each of the hemodynamic drug administration data segments; and detecting an end point of the increase in blood pressure and the increase in heart rate after the start point and before the increase in blood pressure and the end point of the increase in heart rate. and determining the hemodynamic drug detection input feature by performing waveform analysis of the labeled hemodynamic drug administration data segment to calculate a plurality of signal measurements of the hemodynamic drug administration data segment during the increase in heart rate, and calculating combination measurements between the plurality of signal measurements of the hemodynamic drug administration data segment and selecting a signal measurement having a most predictive combination measurement from the plurality of signal measurements of the hemodynamic drug administration data segment as the hemodynamic drug detection input feature.

[0081] 4. Hemodynamic drug prediction machine training of the hemodynamic monitor to determine the hemodynamic drug prediction input features, the hemodynamic drug prediction machine training comprising: identifying a previous period prior to the onset of the increase in blood pressure and the increase in heart rate for each of the hemodynamic drug administration data segments; labeling the previous period of each of the hemodynamic drug administration data segments as a hemodynamic drug prediction data segment; performing waveform analysis of the hemodynamic drug prediction data segments to calculate a plurality of signal measurements for the hemodynamic drug prediction data segments; and determining the hemodynamic drug prediction input features by calculating combination measurements between the plurality of signal measurements for the hemodynamic drug prediction data segments and selecting a signal measurement having a most predictive combination measurement from the plurality of signal measurements for the hemodynamic drug prediction data segment as the hemodynamic drug prediction input feature.

[0082] 23. The hemodynamic monitor of claim 22, wherein the stable detection input feature is determined by: identifying stable data segments in the clinical dataset, each of the stable data segments comprising a stable blood pressure that has not increased by more than a first threshold amount over a set period of time, a stable heart rate that has not increased by more than a second threshold amount over the set period of time, and no infusion of a compound that alters cardiovascular hemodynamics; identifying a start point and an end point of the stable blood pressure and the stable heart rate; labeling the stable data segments from the start point to the end point of the stable blood pressure and the stable heart rate; performing a waveform analysis of the labeled stable data segments to calculate a plurality of stable signal measurements for the stable data segments; and determining the stable detection input feature by calculating a combination measurement between the plurality of stable signal measurements and selecting a stable signal measurement having a most predictive combination measurement from the plurality of stable signal measurements as the stable detection input feature.

[0083] Although the invention has been described with reference to exemplary embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from its essential scope. Therefore, it is not intended that the invention be limited to the particular embodiments disclosed, but the invention is intended to include all embodiments falling within the scope of the appended claims.

Claims

1. 1. A system for monitoring arterial blood pressure in a patient and providing alerts to a healthcare professional regarding nociception in said patient, comprising: a hemodynamic sensor for generating hemodynamic data representative of an arterial pressure waveform of the patient; a system memory storing nociceptive software code; a user interface including a sensory alarm that provides a sensory signal to alert the healthcare professional of a nociceptive event in the patient; a hardware processor executing said nociceptive software code, performing a waveform analysis of the hemodynamic data to determine a plurality of signal measurements; extracting detected input features from the plurality of signal measurements indicative of the nociceptive event in the patient; extracting hemodynamic drug detection input features from the plurality of signal measurements indicative of a hemodynamic drug administration event for the patient; extracting hemodynamic drug predictive input features from the plurality of signal measurements that predict a benefit to the patient from a future hemodynamic drug administration event; extracting stability detection input features from the plurality of signal measurements indicative of stable episodes of the patient; determining a first probability based on the hemodynamic drug prediction input features and the stability detection input features, the first probability representing a probability that the patient will experience a benefit from the future hemodynamic drug administration event; determining a second probability based on the detected input features and the hemodynamic agent detected input features, the second probability representing a probability that the patient will experience the current nociceptive event for the current hemodynamic agent administration event; determining a third probability based on the detected input features and the stable detected input features, the third probability representing a probability that the patient will experience the current nociceptive event relative to the stable episode; comparing said third probability to said first probability and said second probability to determine an output probability of said current nociceptive event for said patient; and a hardware processor configured to activate the sensory alarm of the user interface in response to the output probability meeting a predetermined detection criterion.

2. The detection input features of the nociceptive software code are determined by detection machine training, the detection machine training comprising: collecting a clinical dataset including arterial pressure waveforms and clinical annotations of administration of a compound that alters cardiovascular hemodynamics; identifying nociceptive data segments within said clinical dataset, each nociceptive data segment comprising: an increase in blood pressure of at least a first threshold amount compared to a previous period; an increase in heart rate of at least a second threshold amount compared to the previous period; and wherein an infusion of said compound that alters cardiovascular hemodynamics is not initiated prior to said increase in blood pressure and heart rate. identifying a beginning and an end point of the rise in blood pressure and the increase in heart rate; labeling the nociceptive data segments after the onset point and during the increase in blood pressure and the increase in heart rate; performing a waveform analysis of the labeled nociceptive data segments to calculate a plurality of signal measures of the nociceptive data segments; determining the detected input features by calculating combination measures between the plurality of signal measures for the nociceptive data segment, selecting top signal measures from the plurality of signal measures for the nociceptive data segment having the most predictive combination measures, and labeling the top signal measures as the detected input features.

3. The hemodynamic agent detection input features of the nociceptive software code are determined by hemodynamic agent detection machine training, the hemodynamic agent detection machine training comprising: identifying hemodynamic drug administration data segments within the clinical dataset, each of the hemodynamic drug administration data segments comprising: injecting a compound that alters cardiovascular hemodynamics; an increase in blood pressure after said infusion of at least a third threshold amount; an increase in heart rate after said infusion of at least a fourth threshold amount; identifying a beginning and an end point of the rise in blood pressure and the increase in heart rate in each of the hemodynamic drug delivery data segments; labeling the hemodynamic drug administration data segments after the starting point and during the increase in the blood pressure and the increase in the heart rate; performing a waveform analysis of the labeled hemodynamic drug delivery data segments to calculate a plurality of signal measurements of the hemodynamic drug delivery data segments; The system of claim 1 or 2, further comprising: determining the hemodynamic drug detection input feature by calculating a combination measurement between the plurality of signal measurements of the hemodynamic drug administration data segment and selecting a signal measurement having a most predictive combination measurement from the plurality of signal measurements of the hemodynamic drug administration data segment as the hemodynamic drug detection input feature.

4. The hemodynamic drug prediction input features of the nociceptive software code are determined by hemodynamic drug prediction machine training, the hemodynamic drug prediction machine training comprising: identifying a previous period prior to the onset of the rise in blood pressure and the increase in heart rate for each of the hemodynamic drug administration data segments; labeling the prior period of each of the hemodynamic drug administration data segments as a hemodynamic drug prediction data segment; performing a waveform analysis of the hemodynamic drug prediction data segment to calculate a plurality of signal measurements of the hemodynamic drug prediction data segment; determining the hemodynamic drug prediction input feature by calculating combination measurements between the plurality of signal measurements of the hemodynamic drug prediction data segment and selecting a signal measurement having a most predictive combination measurement from the plurality of signal measurements of the hemodynamic drug prediction data segment as the hemodynamic drug prediction input feature.

5. The stability detection input features of the nociceptive software code are determined by stability detection machine training, the stability machine training comprising: identifying stable data segments within the clinical dataset, each stable data segment comprising: a stable blood pressure without an increase greater than the first threshold amount over a set period of time; a stable heart rate without an increase of more than the second threshold amount over the set period of time; and identifying that no infusion of a compound that alters cardiovascular hemodynamics is performed. identifying a beginning and an end point of said stable blood pressure and said stable heart rate; labeling the stable data segments from the beginning to the end of the stable blood pressure and the stable heart rate; performing a waveform analysis of the labeled stable data segments to calculate a plurality of stable signal measurements for the stable data segments; determining the stable detection input feature by calculating combination measurements between the plurality of stable signal measurements and selecting a stable signal measurement from the plurality of stable signal measurements having a most predictive combination measurement as being the stable detection input feature.

6. performing a waveform analysis of the labeled nociceptive data segments to calculate the plurality of signal measures of the nociceptive data segments; identifying individual cardiac cycles in the arterial pressure waveform of the clinical data set; identifying a dicrotic notch in each of said individual cardiac cycles; identifying a systolic up phase, a systolic down phase, and a diastole in each of said individual cardiac cycles; and extracting signal measurements from each of the systolic up phase, the systolic down phase, and the diastole from each of the individual cardiac cycles.

7. 7. The system of claim 6, wherein the signal measurements correspond to hemodynamic effects from each of the systolic up phase, the systolic down phase, and the diastole from each of the individual cardiac cycles, the hemodynamic effects including contractility, aortic compliance, stroke volume, vascular tone, afterload, and total cardiac cycle.

8. 8. The system of claim 7, wherein the signal measurements include mean, maximum, minimum, duration, area, standard deviation, derivative, and / or morphological measurements from each of the systolic up phase, the systolic down phase, and the diastole from each of the individual cardiac cycles.

9. 9. The system of claim 8, wherein the signal measurements include heart rate, respiratory rate, stroke volume, pulse pressure, pulse pressure variation, stroke volume variation, mean arterial pressure (MAP), systolic pressure (SYS), diastolic pressure (DIA), heart rate variability, cardiac output, peripheral resistance, vascular compliance, and / or left ventricular contractility extracted from each of the individual cardiac cycles.

10. Calculating the combined measure between the plurality of signal measures of the nociceptive data segment comprises: performing step 1 by arbitrarily selecting three signal measurements from said plurality of signal measurements of said nociceptive data segment; performing step 2 by computing powers of different orders for each of the three signal measurements to generate powers of the three signal measurements; performing step 3 by multiplying the powers of the three signal measurements together to generate a product of the powers of the three signal measurements; performing step 4 by performing a Receiver Operating Characteristic (ROC) analysis of said product of powers to arrive at a combined measurement for said three signal measurements; and repeating steps 1, 2, 3, and 4 until all combination measures have been calculated between all of said plurality of signal measures of said nociceptive data segments.

11. The system of any of claims 1 to 10, wherein the hemodynamic sensor is a non-invasive hemodynamic sensor attachable to an extremity of the patient.

12. The system of any of claims 1 to 10, wherein the hemodynamic sensor is a minimally invasive arterial catheter-based hemodynamic sensor.

13. The system of any preceding claim, wherein the hemodynamic sensor generates the hemodynamic data as an analog hemodynamic sensor signal representative of the arterial pressure waveform of the patient.

14. 14. The system of claim 1, further comprising an analog-to-digital converter for converting an analog hemodynamic sensor signal into digital hemodynamic data representative of the arterial pressure waveform of the patient.

15. 1. A method for monitoring arterial blood pressure in a patient and providing an alert to a healthcare professional about current nociception in said patient, comprising: receiving, by a hemodynamic monitor, sensed hemodynamic data representative of an arterial pressure waveform of the patient; performing, by the hemodynamic monitor, a waveform analysis of the sensed hemodynamic data to calculate a plurality of signal measurements of the sensed hemodynamic data; extracting, by the hemodynamic monitor, detected input features from the plurality of signal measurements indicative of a current nociceptive event of the patient; extracting, by the hemodynamic monitor, from the plurality of signal measurements, a hemodynamic drug detection input feature indicative of a current hemodynamic drug administration event of the patient, the current hemodynamic drug administration event being defined as the patient experiencing an increase in heart rate and blood pressure due to administration of a compound that alters cardiovascular hemodynamics; extracting, by the hemodynamic monitor, stability detection input features from the plurality of signal measurements indicative of stable episodes of the patient in which the patient is not experiencing a nociceptive event or a hemodynamic drug administration event; determining, by the hemodynamic monitor, a first probability based on the sensed input features and the hemodynamic drug sensed input features, the first probability representing a probability that the patient will experience the current nociceptive event for the current hemodynamic drug administration event; determining, by the hemodynamic monitor, a second probability based on the detected input characteristics and the stable detected input characteristics, the second probability representing a probability that the patient will experience the current nociceptive event relative to the stable episode; comparing, by the hemodynamic monitor, the second probability to the first probability to determine a probability of output of the current nociceptive event for the patient; and actuating a sensory alarm and generating a sensory signal by the hemodynamic monitor in response to the output probability meeting a predetermined detection criterion.

16. extracting, by the hemodynamic monitor, hemodynamic drug predictive input features from the plurality of signal measurements that predict an effect on the patient from a future hemodynamic drug administration event; determining, by the hemodynamic monitor, a third probability based on the hemodynamic drug prediction input features and the stable detection input features, the third probability representing a probability that the patient will experience a benefit from the future hemodynamic drug administration event for the stable episode; comparing, by the hemodynamic monitor, the second probability to the first probability and the third probability to determine the output probability of the current nociceptive event for the patient; activating, by the hemodynamic monitor, the sensory alarm and generating the sensory signal in response to the output probability satisfying the predetermined detection criterion; The method of claim 15 further comprising:

17. and further comprising: training said hemodynamic monitor to determine said detection input characteristics, said nociceptive detection machine training comprising: collecting a clinical dataset including arterial pressure waveforms and clinical annotations of administration of a compound that alters cardiovascular hemodynamics; identifying nociceptive data segments within said clinical dataset, each nociceptive data segment comprising: an increase in blood pressure of at least a first threshold amount compared to a previous period; an increase in heart rate of at least a second threshold amount compared to the previous period; and wherein an infusion of said compound that alters cardiovascular hemodynamics is not initiated prior to said increase in blood pressure and heart rate. identifying a beginning and an end point of the rise in blood pressure and the increase in heart rate; labeling the nociceptive data segments after the onset point and during the increase in blood pressure and the increase in heart rate; performing a waveform analysis of the labeled nociceptive data segments to calculate a plurality of signal measures of the nociceptive data segments; and determining the detected input features by calculating combination measures between the plurality of signal measures for the nociceptive data segment, selecting top signal measures from the plurality of signal measures for the nociceptive data segment having the most predictive combination measures, and labeling the top signal measures as the detected input features.

18. and hemodynamic drug detection machine training of the hemodynamic monitor to determine the hemodynamic drug detection input characteristics, the hemodynamic drug detection machine training comprising: identifying hemodynamic drug administration data segments within the clinical dataset, each of the hemodynamic drug administration data segments comprising: injecting a compound that alters cardiovascular hemodynamics; an increase in blood pressure after said infusion of at least a third threshold amount; an increase in heart rate after said infusion of at least a fourth threshold amount; identifying a beginning and an end point of the rise in blood pressure and the increase in heart rate in each of the hemodynamic drug delivery data segments; labeling the hemodynamic drug administration data segments after the starting point and during the increase in the blood pressure and the increase in the heart rate; performing a waveform analysis of the labeled hemodynamic drug delivery data segments to calculate a plurality of signal measurements of the hemodynamic drug delivery data segments; 18. The method of claim 17, comprising determining the hemodynamic drug detection input feature by calculating a combination measurement between the plurality of signal measurements of the hemodynamic drug administration data segment and selecting a signal measurement from the plurality of signal measurements of the hemodynamic drug administration data segment having a most predictive combination measurement as being the hemodynamic drug detection input feature.

19. and hemodynamic drug prediction machine training of the hemodynamic monitor to determine the hemodynamic drug prediction input features, the hemodynamic drug prediction machine training comprising: identifying a previous period of time prior to the onset of the increase in blood pressure and the increase in heart rate for each of the hemodynamic drug administration data segments; labeling the prior period of each of the hemodynamic drug administration data segments as a hemodynamic drug prediction data segment; performing a waveform analysis of the hemodynamic drug prediction data segment to calculate a plurality of signal measurements of the hemodynamic drug prediction data segment; and determining the hemodynamic drug prediction input feature by calculating combination measurements among the plurality of signal measurements of the hemodynamic drug prediction data segment and selecting a signal measurement having a most predictive combination measurement from the plurality of signal measurements of the hemodynamic drug prediction data segment as the hemodynamic drug prediction input feature.

20. and further comprising: training the hemodynamic monitor to determine the stability detection input characteristics, the stability detection machine training comprising: identifying stable data segments within the clinical dataset, each stable data segment comprising: a stable blood pressure without an increase greater than the first threshold amount over a set period of time; a stable heart rate without an increase of more than the second threshold amount over the set period of time; and identifying that no infusion of a compound that alters cardiovascular hemodynamics is performed. identifying a beginning and an end point of said stable blood pressure and said stable heart rate; labeling the stable data segments from the beginning to the end of the stable blood pressure and the stable heart rate; performing a waveform analysis of the labeled stable data segments to calculate a plurality of stable signal measurements for the stable data segments; determining the stable detection input feature by calculating combination measurements between the plurality of stable signal measurements and selecting a stable signal measurement from the plurality of stable signal measurements having a most predictive combination measurement as being the stable detection input feature.

Images