System and method for predicting and detecting postoperative complications

Abstract

Disclosed herein are a computer-implemented method and system for assessing a risk value of a target patient. The method includes receiving target patient data at a server via a connection mechanism, and estimating one or more risk values associated with the target patient data at the server using one or more risk assessment models. The target patient data may include historical data, patient population level data, and real-time data. The risk assessment model may include one or more of a historical data risk assessment model, a real-time data risk assessment model, or a combined risk assessment model. A method for training such models is described. The described system and method preferably enable continuous patient monitoring based on a patient's historical data and continuous, real-time physiological data.

Description

Technical Field

[0001] This application claims priority to U.S. Provisional Patent Application No. 63 / 357,370, filed Jun. 30, 2022, which is hereby incorporated by reference in its entirety.

[0002] The present disclosure relates to systems, methods, and devices for monitoring, predicting, and detecting different forms of postoperative complications. The present invention relates to patient monitoring and risk assessment, and more particularly, to computer-implemented systems and methods for monitoring and assessing patient risk, particularly after surgery.

Background Art

[0003] Surgical procedures can use open and minimally invasive techniques on users such as patients to identify and treat medical conditions or improve bodily functions. Surgeries performed for various reasons have an inherent risk of developing postoperative complications such as bleeding, infection, and leakage.

[0004] One of the most dangerous complications of surgery is a complication known as anastomotic leakage. Anastomotic leakage can occur after an anastomosis where two organs are surgically connected and is most commonly seen in gastrointestinal surgery. Anastomotic leakage can lead to leakage of luminal contents into the peritoneal cavity, which can cause a chain of life-threatening complications. This usually presents in the form of severe sepsis, peritonitis, and a pathological condition, which can lead to death.

[0005] Using conventional techniques, it can take an average of 3 to 7 days for a leakage to be diagnosed. This is very dangerous, especially considering that a one-hour delay can cause a significant increase in the patient's pathological condition and risk of death.

[0006] Every year, 70 million major abdominal surgery (MAS) procedures are performed worldwide. MAS includes pancreatic, hepatobiliary, and colorectal surgeries with primary anastomosis. These surgeries have a complication rate of 30 - 60%, 20% of which have significant adverse effects that require invasive treatment and intensive patient monitoring. The occurrence of such complications can have serious and long - term tragic consequences, including a significant degree of morbidity and death in the patients affected by the disease.

[0007] Post - operative complications can further include, but are not limited to, bleeding, post - operative leaks, ischemia, infection, and sepsis.

[0008] Typically, healthcare facilities wait for clinical factors such as abdominal pain, fever, and pulse symptoms to occur before making a diagnosis. Existing techniques for detecting post - operative complications such as anastomotic leaks are non - specific, inefficient, time - consuming, expensive, and / or may lack the ability to provide real - time detection of complications.

[0009] In addition, even if there is an abundance of available patient data, current technologies cannot provide caregivers with an accurate prediction of whether a patient will develop post - operative complications.

[0010] Current patient monitoring systems often rely on physician evaluations that can be biased and may not be provided with information based on historical data. They may not be configured to receive continuous real - time data from patients such as biosignal data or post - operative data (including but not limited to vital measurements).

[0011] Alternatively, patient monitoring systems in the prior art may rely on static models that may not be continuously updated based on newly available patient data.

[0012] Such models are typically trained on data that may have missing values. Common methods for overcoming the missing data problem include imputing the missing values or completely removing inputs with missing values from the dataset.

[0013] Removing patients with missing values limits the number of available training patients and can be problematic to impute when the number of imputed values is large compared to the number of available values and not obvious for a complete collection of sensor measurements.

[0014] U.S. Patent Application Publication No. 20220361823 discloses a wearable blood pressure biosensor, system, and method for short-term blood pressure prediction of mean blood pressure for a future time period. The system can include a model trained to predict future systolic values, diastolic blood pressure values, and / or trends. The model can be trained on data related to the user's personalized body characteristics. However, the model does not receive data from multiple sensors that receive data from multiple patients - rather, they are treated individually for each user, and thus the model is developed from scratch for every new user.

[0015] U.S. Patent No. 10,463,312 discloses embodiments of a method and system for predicting a patient's death. The method includes classifying historical data into a first category and a second category. The method further includes determining a first test parameter and a second test parameter based on at least one of the first patient's sample data and the historical data corresponding to at least one of the first category and the second category. The method further includes determining a probability score based on the cumulative distribution of at least one of the first test parameter and the second test parameter. The method further includes classifying the sample data into one of the first category and the second category based on the probability score. Further, the method includes predicting the death of the first patient based at least on the classification of the first patient's sample data. The death prediction method is limited in that it does not associate biological data with a given time stamp and inherently cannot provide real-time predictions. Since the model is trained under static physiological conditions, it is configured to predict a patient's death based on static, measured biological data and does not associate the time at which the data was measured with the probability of death.

[0016] U.S. Patent Application Publication No. 20220238235 discloses a computer-implemented method for providing patient outcome tracking that may include generating a patient event trigger, wherein during a patient's recovery period, the event trigger may correspond to a value of a patient biomarker that exceeds or is below a threshold while the patient is performing postoperative activities related to the patient's recovery. The computer-implemented method may include receiving actual patient biomarker data from a patient sensor system while the patient is performing postoperative activities. If the actual patient biomarker data includes a value that exceeds or is below a threshold while the patient is performing postoperative activities, the method may include triggering an event trigger. The method may include generating a notification alert corresponding to the event trigger. Outside of a hospital, the patient sensor system comprises medical wearables such as a wristwatch and a headband, and the method is configured to receive postoperative sensor information related to externally measured data such as heart rate, rather than related to a patient's continuously flowing biological fluid. In that sense, it can monitor how a patient is recovering based on these external factors, but may not determine postoperative risks specific to the surgery based on continuous measurements of biomarkers from in-line devices such as leakage or infection.

[0017] New methods and systems have been developed to collect continuous biomarker data from patients in order to detect postoperative complications earlier compared to the standard of care (SOC). This system includes a new method for continuously calculating the probability of risk for a particular patient to develop postoperative complications at that particular point in time. This increases patient outcomes and enables healthcare providers to detect postoperative complications earlier than the SOC.

[0018] Despite significant advancements in postoperative care and the development of numerous new monitoring techniques, to date, there has been no device or technology specifically designed for the early detection or prediction of the occurrence of postoperative complications after surgery. Accordingly, there may be a need for systems and methods that can perform continuous patient monitoring and risk assessment using patient history data, real-time data, or a combination of both.

Prior Art Documents

Patent Documents

[0019]

Patent Document 1

Patent Document 2

Patent Document 3

Patent Document 4

Summary of the Invention

Means for Solving the Problems

[0020] The following presents a simplified summary of the general inventive concept described herein to provide a basic understanding of some aspects of the present disclosure. This summary is not an extensive overview of the present disclosure. It is not intended to limit key or critical elements of the embodiments of the present disclosure or to delineate their scope beyond that explicitly or implicitly described by the following description and the claims.

[0021] The leakage incidence rate from surgical procedures can vary, in some cases, from 1% to 40%. The causes of the occurrence of anastomotic leakage have yet to be studied and no clear cause has been identified. However, there are risk factors associated with a higher incidence rate, such as, but not limited to, age, gender, organ tension, local ischemia, medical history, and surgical errors.

[0022] Physiological changes such as necrosis after tissue ischemia can occur before and during the onset of anastomotic leakage. These changes may appear as slight changes in a patient's vital parameters that may not be detected by current standards of care.

[0023] An object of the present invention is to provide a system and method for predicting and detecting postoperative complications.

[0024] The present disclosure provides systems, methods, and devices for analyzing body fluids and luminal fluids including, but not limited to, ascites, peritoneal drainage fluid, pleural drainage fluid, gastric fluid, fecal material, bile fluid, urine, amniotic fluid, dialysate, sebum, or blood. The fluid can be continuously monitored for changes and trends in specific analytes and biological properties. Examples of these properties and analytes include, but are not limited to, pH, lactate, electrolytes, impedance, conductivity, dissolved oxygen, dissolved CO2, temperature, inflammatory markers, enzymes, bacterial proteins, RNA, or lipids. The systems, methods, and devices disclosed herein can be used for a variety of diagnostic applications including, but not limited to, postoperative leakage, ischemia, infection, and sepsis.

[0025] In some embodiments, sensors such as biosensors can be disposed on a catheter, which can be inserted into the body and can allow fluid to be injected into or withdrawn from the body. The catheter can be placed in proximity to the surgical site to monitor the biological fluid environment in the vicinity of that area. The fluid can be detected locally directly without the need for negative pressure, or it can assist in transporting the fluid through the catheter using negative pressure. Any number of sensors can be disposed on the surface of the catheter such that they are in direct contact with the biological fluid surrounding the area of interest such as a suture line in the case of an anastomosis. The sensors can also be disposed inside the catheter, balloon, pump, or any tubing where fluid can be collected.

[0026] In a further embodiment, the sensor can be housed within a system that can be placed inline with a catheter. The catheter can be placed proximal to the surgical site to monitor the ascitic fluid environment in proximity to the area. The system can be an extension of an existing catheter system. The system can be placed at any time or at a later date when the catheter is in place.

[0027] Advantageously, early management to address complications and leaks using the techniques disclosed herein can significantly reduce the risks associated with such complications.

[0028] Existing techniques for addressing complications can include using interventional radiology techniques to address the patient and existing complications. In the case of a leak, this can include techniques such as placing a drain, placing a stent, performing staple line suturing, etc. These interventions can be performed endoscopically without the need for a second surgery. Monitoring the user situation using systems and techniques as disclosed herein can enable a more effective treatment plan and earlier intervention in the event that complications reappear.

[0029] In addition, the techniques disclosed herein can enable home monitoring of the postoperative course as more patients are transferred to an outpatient monitoring setting. Further, the techniques disclosed herein can enable the user to continuously monitor the status of the patient surgery. It can be an improvement over existing diagnostic tests that take a sample at a particular point in time that may not indicate the patient's status.

[0030] According to one aspect of the present invention, there is provided a method implemented on a computer for monitoring drainage fluid from a patient, the method comprising receiving biosignal data from one or more biosensors coupled to the patient, the sensors being in fluid communication with the drainage fluid and storing pH data in memory; calculating a cumulative average value of the percentage of data points below a predetermined baseline threshold for a current postoperative period; and marking the current postoperative period as "high risk" if the cumulative average value exceeds the predetermined baseline threshold and as "low risk" otherwise.

[0031] According to another aspect of the present invention, there is provided a system for monitoring a user, the system comprising a processor and a memory in communication with the processor, the memory storing instructions that, when executed by the processor, cause the processor to perform the method as described above.

[0032] According to one aspect, there is provided a monitoring device comprising an input port attachable for fluid communication with a catheter, the catheter being for insertion into a user's body to receive fluid from the user's body; an output port generally parallel to the input port and in fluid communication with a fluid reservoir; a fluid flow path defining fluid communication between the input port and the output port; and a biosensor in communication with a computer device for continuously measuring biosignal data of the fluid in the fluid flow path, the biosensor including an electrode pair.

[0033] In some embodiments, the computer device is configured to determine the user's state based at least in part on the biosignal data.

[0034] In some embodiments, the biosensor includes an impedance sensor for detecting the conductivity of the fluid in the fluid flow path.

[0035] In some embodiments, the biosensor includes a pH sensor for detecting the pH level in the fluid in the fluid flow path.

[0036] In some embodiments, the biosensor includes at least one of a lactate sensor, an amylase sensor, a urea sensor, or a creatinine sensor.

[0037] In some embodiments, the device further comprises a flow rate sensor for continuously determining the flow rate of the fluid within the fluid flow path over time.

[0038] In some embodiments, the device further comprises an optical-based sensor including an optical transmitter and an optical receiver for detecting transmission of light through the fluid within the fluid flow path.

[0039] In some embodiments, the optical-based sensor is configured to detect the color of the fluid based at least in part on the detected wavelength.

[0040] In some embodiments, the device further comprises a temperature sensor for detecting the temperature of the fluid within the fluid flow path.

[0041] In some embodiments, the biosensor is disposed on a substrate in fluid communication with the fluid flow path.

[0042] In some embodiments, electrode pairs are sequentially arranged along the length of the fluid flow path.

[0043] According to another aspect, a method implemented on a computer for monitoring a user is provided, the method comprising continuously receiving biosignal data from a biosensor in fluid communication with a fluid, determining the state of the user based at least in part on the biosignal data, and predicting a future occurrence of a complication based at least in part on the state of the user.

[0044] In some embodiments, the method further includes receiving a user profile that is a user profile including information related to a surgical procedure performed on the user, and a future occurrence of a complication is predicted at least in part based on the user profile.

[0045] In some embodiments, the method further includes updating the user profile at least in part based on the biosignal data.

[0046] In some embodiments, the method further includes continuously receiving flow rate data from a flow rate sensor in fluid communication with a fluid from the user's body, and determining a velocity of the fluid flow at least in part based on the flow rate data, wherein the user's condition is determined at least in part based on the velocity of the flow.

[0047] In some embodiments, the method further includes determining a change in the velocity of the fluid flow over time and a change in the biosignal data over time, and predicting a future occurrence is at least in part based on the change in the velocity of the flow and the change in the biosignal data.

[0048] In some embodiments, the flow rate data is received in near real-time.

[0049] In some embodiments, the biosignal data is received in near real-time.

[0050] In some embodiments, the method further includes receiving optical data related to transmission of light through the fluid from an optical-based sensor in fluid communication with the fluid.

[0051] In some embodiments, the method further includes determining a color of the fluid at least in part based on the optical data.

[0052] In some embodiments, the method further includes receiving temperature data of the fluid from a temperature sensor in fluid communication with the fluid. In some embodiments, the method further includes modulating the biosignal data based at least in part on the temperature data.

[0053] In some embodiments, the method further includes determining a risk factor of the user based on a correlation with a trend of biosignal data of other users.

[0054] In some embodiments, the state of the user is at least partially based on determining whether the biosignal data is within a threshold boundary.

[0055] According to a further aspect, there is provided a system for monitoring a user, comprising a processor and a memory communicating with the processor, the memory storing instructions that, when executed by the processor, cause the processor to perform a method as described herein.

[0056] Current SOCs rely on a combination of standardized management guidelines and the individual experience of healthcare providers. There is active research regarding various risk factors and the use of new biomarkers for early detection of postoperative complications.

[0057] Embodiments of the present invention include methods for calculating a continuous real-time numerical score based on the pH of a drainage fluid related to the probability that a patient will develop a postoperative complication. These methods are new and do not exist in current SOCs. The methods enable healthcare providers (HCPs) to obtain the real-time status of their patients for earlier, closer monitoring, intervention, or drainage compared to SOCs.

[0058] The method can integrate multiple biomarker data or other source data available to the HCP (e.g., EMR / EHR, other sensors, or medical devices) to provide a more accurate probability score.

[0059] The method can be deployed in firmware or in the cloud and can be updated / improved as needed or when newly acquired data becomes available.

[0060] The method can be made to detect different types of postoperative complications.

[0061] The method can be used with pH data from any source or device, including point measurements from continuous monitoring devices or benchtop pH meters.

[0062] The method can further be used with other biomarker or biosignal data from any source or device, including in-line and / or continuous monitoring devices.

[0063] According to one aspect of the present invention, a method implemented on a computer for evaluating a risk value of a target patient is disclosed, the method including receiving target patient data at a server via a connection mechanism and estimating, at the server, one or more risk values associated with the target patient data using one or more risk assessment models.

[0064] In one embodiment, the target patient data includes one or more of historical data, patient population level data, and real-time data.

[0065] In one embodiment, the one or more risk assessment models include one or more of a historical data risk assessment model, a real-time data risk assessment model, and a combined risk assessment model, and the combined risk assessment model includes one or more risk assessment models.

[0066] In one embodiment, a historical data risk assessment model is trained by receiving historical data corresponding to a plurality of patients in a server comprising one or more processors and a memory, wherein the one or more processors include one or more of a mapping engine and a normalization engine, historical data of each patient among the plurality of patients corresponds to an indicator of whether the patient has encountered a complication, mapping, via the mapping engine, the historical data of each patient among the plurality of patients to a numerical value, normalizing, via the normalization engine, the historical data of each patient among the plurality of patients, and performing a regression analysis to determine a relationship between the historical data and a risk value on the preprocessed data to generate preprocessed historical data.

[0067] In one embodiment, a real-time data risk assessment model is trained by receiving time-stamped data corresponding to signal data measured by a plurality of sensors connected to a plurality of patients in a server comprising one or more processors and a memory, wherein the one or more processors include one or more of a filtering and enhancement engine and a normalization engine, time-stamped data for each patient among the plurality of patients corresponds to an indicator of whether the patient has encountered a complication, filtering and enhancing, via the filtering and enhancement engine, the time-stamped data for each patient among the plurality of patients, normalizing, via the normalization engine, the time-stamped data for each patient among the plurality of patients, and performing one or more regression analyses to determine one or more relationships between the real-time data and a risk value on the preprocessed time-stamped data to generate preprocessed time-stamped data.

[0068] According to one embodiment of the present invention, the historical data includes one or more of preoperative risk factors, medical records, surgical history, individual health indicators, and surgical parameters of a target patient or a target patient population.

[0069] According to one embodiment of the present invention, the real-time data includes sensor data from one or more sensors that continuously measure signals related to the physiological conditions of a target patient.

[0070] In one embodiment, the method further includes notifying the user of the risk value.

[0071] In one embodiment, the risk value is continuously estimated.

[0072] According to one embodiment of the present invention, the risk value includes the probability that the target patient will develop postoperative complications.

[0073] According to one aspect of the present invention, a system for evaluating the risk value of a target patient is disclosed. The system includes a server, one or more processors, and a memory. The one or more processors are communicably coupled to a database, which includes historical data and time-stamped data from multiple patients. The one or more processors are configured to receive target patient data via a connection mechanism. The memory includes instructions that, when executed by the one or more processors, cause the server to receive the target patient data at the server, preprocess or process the target patient data via the processor, and estimate one or more risk values related to the target patient data using one or more risk assessment models.

[0074] According to one embodiment of the present invention, the target patient data includes one or more of historical data, patient population level data, and real-time data.

[0075] According to one embodiment of the present invention, the one or more risk assessment models include one or more of a historical data risk assessment model, a real-time data risk assessment model, and a combined risk assessment model, and the combined risk assessment model includes one or more risk assessment models.

[0076] According to one embodiment of the present invention, a historical data risk assessment model receives historical data corresponding to a plurality of patients in a server including one or more processors and a memory, wherein the one or more processors include one or more of a mapping engine and a normalization engine, and the historical data of each patient among the plurality of patients corresponds to an indicator of whether the patient has encountered a complication; maps the historical data of each patient among the plurality of patients to a numerical value via the mapping engine; normalizes the historical data of each patient among the plurality of patients via the normalization engine; and performs a regression analysis for determining a relationship between the historical data and a risk value on the preprocessed data, thereby generating preprocessed historical data, and is trained thereby.

[0077] According to one embodiment of the present invention, a real-time data risk assessment model receives time-stamped data corresponding to signal data measured by a plurality of sensors connected to a plurality of patients in a server including one or more processors and a memory, wherein the one or more processors include one or more of a filtering and enhancement engine and a normalization engine, and the time-stamped data for each patient among the plurality of patients corresponds to an indicator of whether the patient has encountered a complication; filters and enhances the time-stamped data for each patient among the plurality of patients via the filtering and enhancement engine; normalizes the time-stamped data for each patient among the plurality of patients via the normalization engine; and performs one or more regression analyses for determining one or more relationships between the real-time data and a risk value on the preprocessed time-stamped data, thereby generating preprocessed time-stamped data, and is trained thereby. According to one embodiment of the present invention, the risk value includes the probability that a target patient will develop a postoperative complication.

[0078] According to one embodiment of the present invention, the system further includes a display system for displaying risk values.

[0079] According to one aspect of the present invention, a non-transitory computer-readable storage medium is disclosed, the computer-readable storage medium including instructions that, when executed by a computer, cause the computer to perform steps of receiving target patient data via a connection mechanism, preprocessing or processing the target patient data via a processor, and estimating one or more risk values associated with the target patient data using one or more risk assessment models.

[0080] According to one embodiment of the present invention, the target patient data includes one or more of historical data, patient population level data, and real-time data.

[0081] Other features will become apparent from the drawings in conjunction with the following description.

[0082] The advantages and features of the present invention include convenient and early management for dealing with complications such as anastomotic leakage using the techniques disclosed herein that can significantly reduce the risks associated with such complications. The advantages and features will be better understood by reference to the following more detailed description and the claims in conjunction with the accompanying drawings in which like elements are identified by like reference numerals.

[0083] Some embodiments of the present disclosure will be provided by way of example only with reference to the accompanying drawings.

Brief Description of the Drawings

[0084]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

MODE FOR CARRYING OUT THE INVENTION

[0085] The various implementation forms and aspects of this specification will be described with reference to the details discussed below. The following description and drawings are examples of the specification and should not be construed as limiting the specification. A number of specific details are described to provide a complete understanding of the various implementation forms of this specification. However, in some cases, well-known or conventional details are not described to provide a concise discussion of the implementation forms of this specification.

[0086] Various devices and processes are described below to provide examples of implementation forms of the system disclosed in this specification. The implementation forms described below do not limit any implementation form of the claims, and the implementation forms of the claims may include processes or devices different from those described below. The implementation forms of the claims are not limited to devices or processes having all the features of any one device or process described below or to features common to a plurality or all of the devices or processes described below. The devices or processes described below may not be an implementation form of the subject matter of any claim.

[0087] Furthermore, numerous specific details are set forth in order to provide a thorough understanding of the implementations described herein. It will be understood by those of ordinary skill in the art, however, that the implementations described herein may be practiced without these specific details. In other instances, well-known methods, procedures, and components have not been described in detail so as not to obscure the implementations described herein.

[0088] As used herein, an element may be described as "configured to" perform one or more functions or "configured for" such functions. In general, an element configured to or configured for a function is enabled to perform that function, suitable for performing that function, made to perform that function, operable to perform that function, or otherwise has the ability to perform that function.

[0089] For the purposes of this specification, it is understood that the language "at least one of X, Y, and Z" and "one or more of X, Y, and Z" can be construed as only X, only Y, only Z, or any combination of multiple items X, Y, and Z (e.g., XYZ, XY, YZ, ZZ, etc.). Similar logic applies to any occurrence of the language "at least one of..." and "one or more of.." with respect to multiple items.

[0090] In some embodiments, the sensor may include an electrochemical or solid state sensor having different forms including, but not limited to, potentiometric measurements, voltammetry, conductivity measurements, capacitance, amperometric measurements or ion-sensitive field effect transistors (ISFETs). In some embodiments, the sensor may be piezoelectric or a micro-electro-mechanical system (MEMS). The sensor may include terminals connected to an active, counter, reference or pseudo-reference electrode, depending on the type of sensor being used. The sensor may be of different types including, but not limited to, pH sensors, ion-sensitive sensors, temperature sensors, lactate sensors, electrolyte sensors, impedance sensors, light-based sensors, microbial sensors, protein sensors, DNA-based sensors, carbohydrate sensors, enzyme sensors, oxygen sensors such as P02 (partial pressure of oxygen) sensors, amylase sensors, urea sensors, creatinine sensors, motion sensors (e.g., accelerometers), pressure sensors and flow sensors.

[0091] The sensors may be connected in series or in parallel and may be arranged sequentially, for example, along the length of a fluid flow path.

[0092] In some embodiments, the sensor may include one or more temperature sensors such as a thermistor or a resistance temperature detector.

[0093] In use, the temperature sensor may be subject to a change in resistance that correlates with a change in temperature. Thus, the temperature may be determined by exciting with a current and determining the resistance of the thermistor by measuring the voltage (or vice versa).

[0094] In some embodiments, the sensor may include an essentially electrochemical pH sensor that enables a biological sample to be converted into an electrical signal that can later be measured, monitored, and analyzed to determine whether a postoperative complication has occurred. A system of electrodes (active, counter, and reference) combined with each other may be manufactured on a biocompatible substrate. The electrodes may be manufactured from biocompatible materials such as gold, platinum, titanium, and silver, and may then be functionalized with an active polyaniline (PANI), polyaniline / polyurethane (PAIN / PU), polyurethane, polymer, or other suitable layer. In one example, the m-biosensors are sized 500 pm × 500 pm and are capable of being placed on or incorporated into a catheter to monitor changes in pH over time.

[0095] In some embodiments, the sensor may include an optically based sensor such as an optoelectronic sensor that uses a combination of an optical transmitter or source and a detector in the ultraviolet to infrared spectrum to measure the light absorption or transmission characteristics of a fluid. Single-wavelength or multi-wavelength light rays may be used.

[0096] In some embodiments, the sensor includes an impedance sensor that is typically operated with alternating current (AC) excitation and can be used to evaluate the user's situation. The impedance sensor may include an electrode pair and may include excitation and readout circuitry.

[0097] Evaluating postoperative risk The systems, methods, and devices disclosed herein can be used to monitor, detect, and predict different forms of postoperative complications such as leaks that can occur after surgery. Embodiments may include sensing and diagnostic devices that use sensors, such as sensors on a catheter or in-line device, to detect or predict the presence of fluid in a lumen when a leak occurs, for example.

[0098] In some embodiments, the systems, methods, and devices disclosed herein enable monitoring of biological fluids of a sample that may indicate a surgical leak, and include sensors such as biosensors that can be used to detect biosignal data and are disposed in a location proximate to a surgical site.

[0099] The biosignal data can be received from prior art detection systems, such as those described in U.S. Patent Application Publication No. 2022 / 0265175, which is owned by the assignee of the present application and is hereby incorporated by reference in its entirety.

[0100] In some embodiments, the sensor can include electrochemical or solid-state sensors having different forms, including but not limited to potentiometry, voltammetry, conductivity measurement, capacitance, amperometry, or ion-sensitive field effect transistors (ISFETs). In some embodiments, the sensor can be piezoelectric or a microelectromechanical system (MEMS). The sensor can include terminals that connect to an active, counter, reference, or pseudo-reference electrode depending on the type of sensor being used. The sensor can be of different types including, but not limited to, pH sensors, ion-sensitive sensors, temperature sensors, lactate sensors, electrolyte sensors, impedance sensors, fluid sensors, light-based sensors, microbial sensors, protein sensors, inflammation sensors, carbohydrate sensors, enzyme sensors, oxygen sensors such as P02 (partial pressure of oxygen) sensors, amylase sensors, urea sensors, creatinine sensors, pressure sensors, and flow sensors.

[0101] The sensors can be connected in series or in parallel and can be arranged sequentially, for example, along the length of a fluid flow path.

[0102] In some embodiments, the sensor can include a temperature sensor such as a thermistor.

[0103] In use, a thermistor may experience a change in resistance correlated with a change in temperature. Thus, temperature can be determined by determining the resistance of the thermistor by exciting it with a current and measuring the voltage (or vice versa).

[0104] A temperature sensor can be used to account for some artifacts and error sources in biosignal measurement results. The temperature sensor can be used to compensate or modulate signals from other temperature-dependent sensors such as impedance and pH. Since biological fluids tend to be at a higher temperature than the ambient temperature, an increase in the fluid temperature detected by the temperature sensor can indicate the influx of new fluid.

[0105] An array of a temperature sensor and a heating element can be used to measure fluid flow using the principle of thermal mass fluid transport.

[0106] In some embodiments, the sensor can include a flow sensor, such as a flow meter, for measuring the volume or mass flow rate of a fluid, such as a liquid or gas within a user's body.

[0107] In some embodiments, the sensor can include an essentially electrochemical pH sensor that enables a biological sample to be converted into an electrical signal that can later be measured, monitored, and analyzed to determine whether postoperative complications are developing. A system of electrodes (active, counter, and reference) combined with each other can be fabricated on a biocompatible substrate. The electrodes can be fabricated from biocompatible materials such as gold, platinum, titanium, and silver and then functionalized with an active polyaniline (PANI), polyaniline / polyurethane (PAIN / PU), polyurethane, polymer, or other suitable layer. In one example, m-biosensors are 500 pm × 500 pm in size and are capable of being placed on a catheter to monitor changes in pH over time.

[0108] The pH sensor can be formed from a conductive polymer made from aniline monomers. The sensitivity of a suitable conductive polymer to pH levels can enable its use as a pH-sensitive component within the pH sensor.

[0109] The pH sensor can be adjusted and / or controlled by an electronic device that controls the potential and current differences of a three-electrode system comprising a potentiostat, specifically a working electrode (WE), a reference electrode (RE), and a counter electrode (CE). This electrical device has a number of applications that can be used to manufacture pH sensors such as cyclic voltammetry (CV), chronoamperometry, and chronopotentiometry.

[0110] The pH sensor can be configured to detect pH values within a threshold or boundary or deviations from such a boundary.

[0111] In some embodiments, the sensor can include an optoelectronic sensor such as a light-based sensor that uses a combination of a light transmitter or source and a detector in the ultraviolet to infrared spectrum to measure the light absorption or transmission characteristics of a fluid. Single-wavelength or multi-wavelength light rays can be used.

[0112] The light-based sensor can include a combination of a light transmitter and a detector within the ultraviolet to infrared spectrum and can be used to measure the light absorption or transmission characteristics of a fluid.

[0113] The light absorption or transmission characteristics can indicate changes in body fluids and luminal fluids, including but not limited to protein composition and concentration, pH, conductivity, inflammatory markers, and cellular activity due to the manifestation of complications or diseases. This also enables the measurement of the color of fluids that can indicate bleeding (red), bile leakage (yellow-green), fecal leakage (brown), gastric leakage (green), urine leakage (yellow), and other fluids of specific colors.

[0114] In some embodiments, a single wavelength or multi - wavelength light beam may be used. Changes detected in the absorption or transmission characteristics of a fluid within a particular spectral band or wavelength may enable measurement of the color of the fluid. Serous fluids (e.g., peritoneal and pleural fluids) are normally pale yellow, so a change in color may indicate bleeding (red), bile leak (yellow - green), fecal leak (brown), gastric leak (green), urine leak (yellow), or other fluids of a particular color.

[0115] In some embodiments, an optical - based sensor may include a combination of optical transmitters and detectors in the ultraviolet to infrared spectrum to measure the scattering of light by the fluid in order to measure its turbidity. Serous fluids are normally clear in appearance and have low turbidity. An increase in turbidity, measured as an increase in light measured at right angles by a photodetector for example, may indicate white blood cells and microorganisms within the fluid due to infection.

[0116] The optical - based sensor may include multiple light sources and receivers. For example, a single broadband light source may be used in combination with multiple band - specific photodiodes (e.g., red, green, and blue). In this way, the absorption / transmission characteristics of the fluid may be measured across the same number of bands as there are photodetectors present. Similarly, multiple light sources may be used in combination with a single broadband photodetector, whereby each light source is subsequently turned on and the transmitted light is appropriately measured by the photodetector. Finally, the light sources and photodetectors may also use dynamic filters to enable the emission or detection of specific bands of light instead of multiple sources or photodetectors.

[0117] In some embodiments, the sensor may include an impedance sensor that is normally operated with alternating - current (AC) excitation to evaluate the user's condition. The impedance sensor may include an electrode pair and may include excitation and read - out circuitry.

[0118] In some embodiments, the impedance sensor can be configured to perform AC excitation within a well-defined and constant fluid geometry (constrained by a flow channel or housing) such that a normalized impedance (or a specific impedance) and admittance are determined.

[0119] In some embodiments, the impedance of the fluid can be measured across a range of frequencies (from Hz to MHz) to separate the contributions of individual electrolytes and infer the ionic composition of the fluid. The user state can be based at least in part on the ionic composition of the fluid.

[0120] The measured impedance values can be transformed to determine the conductivity of the fluid (e.g., the real component of the impedance). Conductivity can, in itself, reveal properties of the fluid and thus can directly provide clinical values.

[0121] For example, conductivity can indicate the unique properties and composition of the specimen.

[0122] Impedance can be affected by fluid volume and geometry, and thus the measured impedance can be used to locate and track particles and bubbles within the fluid flow path.

[0123] In some embodiments, the impedance sensor can be used to account for some artifacts and error sources in the biological signal measurements.

[0124] In some embodiments, the impedance sensor can be used to detect a rapid and dramatic increase in impedance beyond the range of body fluids that indicates the presence of bubbles in the flow path. Bubbles are a problem for catheter-based measurements because they cause artifacts in the readings.

[0125] In some embodiments, the impedance sensor can be used to detect a sharp increase in impedance that can indicate the presence and quantity of inhomogeneous materials and particles (e.g., blood clots, fibrin).

[0126] In some embodiments, the impedance sensor can be used to detect blood clotting (characterized typically by a sharp increase in impedance followed by a slow but sustained increase in impedance) and thus the presence of blood and the risk of flow path obstructions.

[0127] In some embodiments, an array of impedance sensors disposed along the flow path can be used to detect and track bubbles, inhomogeneous materials, and / or particles as they travel through the flow path using the techniques described herein.

[0128] In some embodiments, the sensor can include an amylase sensor.

[0129] In use, the systems, methods, and devices can monitor trends and changes in physical and chemical biomarkers, including but not limited to pH, temperature, flow rate, pressure, lactate, lactic acid, nitrate, glucose, alkali ions, oxygen, bicarbonate, inflammatory proteins, bacterial proteins, and other biomarkers that may be associated with or correlated to leakage.

[0130] A single sensor or sensor array can be disposed along the wall of a catheter, inside a dedicated lumen, or within an in-line device, enabling the device to detect and monitor whether leakage is occurring.

[0131] In some embodiments, the catheter can be used as a carrier device for sensors for monitoring internal compartments of the body, such as the peritoneal or pleural cavity, without applying negative pressure. The catheter can be connected to a balloon or a mechanical pump to apply negative pressure to facilitate fluid drainage. The catheter can also be connected to a fluid supply, such as saline, to perform treatment and diagnostic functions, such as dialysis or irrigation methods.

[0132] In some embodiments, multiple sensors can be spaced along the length of the catheter. The multiple sensors arranged along the catheter can enable multiple regions to be sensed and the spatial progression of leakage to be tracked.

[0133] The catheter can be formed of a tube having a hollow or solid body and made of a medical material such as a suitable polymer. In some embodiments, the catheter can be a flexible substrate.

[0134] In some embodiments, the catheter can be formed of a low-friction material.

[0135] The catheter can have different designs. The catheter can be cylindrical, rectangular, flat, or T-shaped in cross-section and can have a single lumen or multiple lumens.

[0136] In some embodiments, the sensor can be disposed inside a reservoir from which fluid can be collected from the user's body. The reservoir can include elements such as a balloon, pump, or other container that can hold biological fluid. The sensor disposed within the reservoir can be used simultaneously with the sensor disposed within the catheter. This can enable a greater number of sensors to be used to determine postoperative complications such as various different conditions or fluid leakage, infection, inflammation, or other dangerous complications.

[0137] Sensors such as biosensors can be connected to a monitor such as an electronic data acquisition system (DAQ) that can continuously process data obtained from the sensors and can be placed inside or outside the user's body. The connection can be established through different methods including, but not limited to, wires and connectors embedded in at least one lumen designed to allow the wires and connectors to run through the interior of the catheter. The connection can also be established wirelessly by transmitting data acquired in vivo from the biosensor to a receiver placed outside the body via a transmission system.

[0138] In some embodiments, each of the plurality of sensors communicates independently with the monitor.

[0139] The monitor can have a screen that allows the readings to be directly observed on the device. The monitor can also use various visual or audio cues such as small LEDs or alarm sounds to notify various events.

[0140] The data acquired by the monitor can also be communicated to a computer system via a wired or wireless medium for further analysis and visualization. The communicated data can be processed, left as is, or summarized.

[0141] In some embodiments, the data collected by the monitor can be analyzed to identify trends related to the onset of different complications. This can be done by evaluating single or multiple data sets obtained from one or more sensors over time to diagnose and determine the stage of onset of the complication.

[0142] If one or more of the sensors demonstrate a biological trend related to surgical leakage, an alarm signal can be sent from the monitor to a computer-based system to enable the user to determine appropriate medical action.

[0143] In one example, a slow decrease in local pH may indicate a small leakage into the wound or poor blood supply. If a simultaneous slow increase in lactate concentration is observed, it may indicate a lack of blood supply (i.e., ischemia). If the lactate concentration is constant, it may indicate a slow leakage.

[0144] In another example, a sharp decrease in pH may indicate a large leakage. If the pH returns to its baseline, it may suggest that the wound has healed despite the leakage. If the pH continues to drop or remains low, it may indicate a significant leakage that the body has difficulty recovering from.

[0145] The systems and methods disclosed herein may perform monitoring, detection and diagnosis, and prediction. For example, monitoring may indicate data sensed by sensors such as biosensors. Detection and diagnosis can, via algorithms, detect the user's condition and / or make diagnostic determinations such as, for example, at a relevant confidence level, leakage, what type of leakage, and where the leakage is coming from. Prediction can examine different trends and process signals using the sensed data to predict possible future leakage, for example, at a relevant confidence level. As such, embodiments of the systems and methods disclosed herein can identify the physiological differences between the occurrence of leakage and the precursors of leakage.

[0146] The systems and methods disclosed herein can be used to perform clinical functions. In one example, a catheter system can be connected to a mechanical element that can apply a negative pressure that allows fluid to be discharged from the user's body, in addition to its diagnostic function. Such clinical functions can both be performed at locations within the user's body, such as inside the gastrointestinal tract or within the peritoneal cavity.

[0147] Techniques for applying negative pressure can include, but are not limited to, balloons, mechanical pumps, vacuum systems, or other devices that can draw fluid from the body to the outside. In some embodiments, the fluid being discharged can assist in diagnostic applications by causing a constant flow rate across a sensor. In some embodiments, clinical functions can be performed by pumping fluid into the user's body.

[0148] As used herein, the term "body fluid" can refer to fluids derived from within the human body, fluids excreted or secreted by the body (e.g., blood, gastric juice, and ascites), and similar fluids. By extension, the term "luminal fluid" refers to a subset of body fluids present within internal cavities, intestines, blood vessels, tubular organs, and many other thin-membrane-bound organs, such as gastric juice, intestinal fluid, fecal material, urine, bile fluid, and other similar fluids.

[0149] As used herein, the terms "biomarker" and "aptamer" can refer to molecules, substances, and chemical or physical properties that can be measured or detected as biological signals within body fluids. They include, but are not limited to, pH, temperature, electrolyte concentration, fluid flow rate, pressure, lactate, lactic acid, nitrate, alkali ions, inflammatory proteins, bacterial proteins, specific cells, molecules, genes, gene products, enzymes, hormones, inflammatory proteins, and glucose.

[0150] As used herein, the terms "biosensor" and "sensor" can refer to devices or systems that convert these signals into measurable electrical signals to detect or respond to biomarkers or biological signals. Biosensors and sensors used herein can include, but are not limited to, pH sensors, lactate sensors, amylase sensors, lactate sensors, glucose sensors, temperature sensors, pressure sensors, enzyme sensors, protein sensors, biological sensors, ion sensors, electrolyte sensors, impedance sensors, conductivity sensors, flow rate sensors, and other forms of electrochemical and solid-state sensors.

[0151] In some embodiments, signal data received from a sensor device can be collected over time and associated with a user in order to develop a user profile. In some embodiments, user profile information, such as signal data associated with one or more users, can be applied to machine learning techniques in order to develop a model of such signal data.

[0152] The user profile can include information regarding a surgical procedure performed on the user and the date and time of the surgical procedure, the location of the surgical procedure, the date and time of insertion of the sensor device, the location of insertion of the sensor device, the user's age, height, weight, medical history, condition or illness (e.g., diabetes), current or past medications being used by the user, or other current or historical factors related to the user, the surgery, or device details.

[0153] In some embodiments, the user profile includes a list of medications used by the user. This can be used to identify potential sources of error caused by medications that change one or more thresholds of the biosignals being measured using the sensors described herein.

[0154] In some embodiments, the user profile includes procedures performed on the patient. Such a list of procedures can be used to further analyze the potential list of complications that the user may be subject to from the given risks of each procedure. Additionally, such a list of procedures can be used to identify the anatomical structures of biological fluids in proximity to the procedure location.

[0155] In some embodiments, information related to a surgical procedure performed on the user includes the date and time of the surgical procedure. Additionally, the date and time of the surgery can be used to analyze the user's condition assuming a complete timeline of the user's recovery.

[0156] In some embodiments, user profile data can be input by the user, a health management facility, or a health management expert. For example, a surgeon can input information related to an operation performed and details regarding sensor devices used after the operation (e.g., operating parameters, number and type of sensors, etc.).

[0157] In some embodiments, a user profile can be automatically generated from, for example, a health record indicating surgical details or the user's electronic health record. These can be received from a computer device communicating with system 100.

[0158] In some embodiments, a user profile can include information identifying factors related to a user state determined from collected signal data.

[0159] In one example, a user state indicated by sensor data occurring during a temporary period suggesting, for example, a temporary spike can be discarded as not indicative of that particular state and more as an anomaly.

[0160] In some embodiments, the future occurrence of complications such as anastomotic leakage can be predicted at least in part based on the time at which the user state occurs and the length of time the user state occurs.

[0161] Machine learning algorithms can be applied to previously acquired signal data related to a user state. For example, pattern recognition can be performed on previously acquired signal data related to a particular user state. Machine learning can generate a user state classification model trained by previously acquired signal data.

[0162] A further description of such a model will be given in more detail later.

[0163] Leakage can be predicted by analyzing changes in the flow rate of fluid around the sensor of a sensor device. For example, how fast the change in flow rate occurs can indicate how fast the leakage is flowing.

[0164] In another example, leakage can be predicted based on the accumulation of lactate detected by a sensor.

[0165] In another example, leakage can be predicted based on the consumption of oxygen detected by a sensor.

[0166] In another example, leakage can be predicted based on the detected pH change, and may include an analysis of why the pH has changed to distinguish different causes or conditions of such pH changes.

[0167] In some embodiments, the future occurrence of anastomotic leakage can be predicted based on whether the user's condition exceeds or is below a predetermined threshold. Such a threshold may be, for example, a pH value.

[0168] As described above, a secondary state or a second user state can be determined from the sensor data. The future occurrence of leakage can be predicted at least partially based on the second user state. The second user state can also indicate a risk factor or risk level of the leakage state.

[0169] In some embodiments, a risk model can determine a confidence level as to whether leakage has occurred based on a combined analysis of the user profile and the processed biosignals. Weighted coefficients based on the user's profile and current state can be used to describe the context of the algorithm input / output according to the likelihood of leakage onset. These weights may be dynamic over time and can be updated when the user's state is updated. In one example, if a user has undergone bariatric surgery, a higher weight can be applied to the gastric leakage detection algorithm compared to the colorectal leakage detection algorithm.

[0170] In some embodiments, predictive analysis of signals such as biosignals from a biosignal sensor, such as a sensor device or an in-line monitoring device, may include diagnosis, e.g., detection of leaks or identification of disease characteristics by examination of symptoms monitored by the sensor.

[0171] In some embodiments, triage conditions or risk levels for predicting future occurrences of leaks may be based on signal data, user state, and user profile. The generated data may include triage conditions or risk levels.

[0172] Machine learning algorithms may be applied to previously acquired signal data, user profile data, and user state data. For example, pattern recognition may be performed on previously acquired signal data related to a particular leak prediction.

[0173] Data related to predicting future occurrences of leaks may include a notification of the prediction.

[0174] In some embodiments, signal data may be collected to build trends across several patients, and cross-correlation techniques may be used to identify similarities between a patient's data and previous patients. And a match or correlation may indicate a risk factor.

[0175] Advantageously, better predictions may be possible because there is better fidelity of data acquired over time, for example continuously as a flow rate via a sensor device, and in one example in real-time or near real-time.

[0176] An example application of the systems and methods disclosed herein can be demonstrated by looking at a patient suffering from colorectal cancer, where the tumor needs to be removed. The surgeon may decide to perform an anastomosis after removing the tumor from the body. The surgeon may then place a catheter with a biosensor in one of the paracolic gutters if they think it is the most likely area to collect fluid from a leak. The catheter system used by the surgeon may be equipped with fluid, pH, and lactate sensors. After the catheter is placed, the surgeon may also choose to use absorbent sutures to keep the catheter in place. The catheter can then be connected to a monitor placed outside the body. The catheter can also be connected to a balloon that will apply negative pressure to drain fluid from the peritoneum.

[0177] The monitor will then inform the user and the surgeon that the patient's condition appears normal and confirm that the connection has been established with the biosensor. The patient can then be kept in the hospital overnight and then discharged on the second day. The patient can be discharged with the monitor and catheter. For three days after the surgery and after the patient has been discharged, the biosensor can detect clinically relevant simultaneous increases in pH changes and lactate concentrations. The monitor can then send a signal to the patient to seek treatment, or the system can delay the signal to wait for more significant changes. The data can be relayed wirelessly to a medical facility, at which point an expert can also view the data and decide whether to have the patient come to the medical facility. The system can then detect significant pH changes and a relatively high flow rate of fluid in the abdominal cavity. The monitor can then warn the patient by informing the patient that a leak has been detected and that the patient needs immediate medical intervention. The medical facility can also receive the notification. If the patient is in the hospital, the medical facility can view the data obtained from the biosensor and make clinical decisions to support the patient before complications increase. The surgeon may also decide to take corrective medical actions, including but not limited to reoperation of the patient. The system can be reused after the corrective action has been taken.

[0178] Conveniently, the sensor device described herein can be removable from the user or patient in an outpatient setting, for example, by a nurse in the user's home, and without the user having to undergo additional surgical procedures. In one example, this can occur 10 to 20 days after surgery.

[0179] In some embodiments, integrating the sensor with a catheter (and specifically a drain) can enable the sensor device to be removed in an outpatient setting.

[0180] The method calculates a continuous real-time numerical score based on the pH of the drainage fluid related to the probability that a patient will develop postoperative complications such as leakage.

[0181] A set of pH measurement results is obtained from the patient (i.e., from the patient's biological fluid) in a clinical trial either continuously or at individual time intervals. The data for each patient will be labeled as "Leak: AL" or "non Leak: nAL". The continuous data can be further divided into time intervals of any length. For this example, the interval was 1-hour intervals. The determination of the baseline threshold is performed using Receiver Operator Characteristic (ROC) analysis based on the pH every hour after surgery from the clinical data.

[0182] Figure 1 shows the ROC analysis of the data. Essentially, the data from each hour is examined, and the baseline pH value for that 1-hour interval is determined via ROC analysis, thereby enabling the discrimination between AL patient data and nAL patient data. The baseline is based on the greatest area under the curve (AUC) of the ROC curve or on the highest possible sum of sensitivity and specificity. The representative ROC curve with the highest Youden index threshold is highlighted with a black dot.

[0183] Figure 2 shows the baseline cut-off points for each interval. This step is repeated until all available data has been processed. The result can be a list of baseline cut-off points for each interval (shown graphically in Figure 2, indicating the baseline pH threshold per hour). Each point represents a pH value threshold that will distinguish patients with nAL and AL by other methods that can, for example, maximize the sum of the sensitivity and specificity of the data or estimate the highest true positive rate with the lowest false positive rate when applied to the data in that postoperative time interval.

[0184] In Figure 3, the average pH is shown as a function of postoperative time.

[0185] Figure 4 shows one embodiment of a method for calculating a continuous real-time numerical score based on the pH of the drainage fluid related to the probability that a patient will develop postoperative complications.

[0186] The method enables healthcare providers (HCPs) to obtain the real-time status of their patients for earlier, closer monitoring, intervention, or drainage compared to SOC.

[0187] The method includes continuous learning depicted in the flowchart of Figure 4. The method learns from datasets of clinical trials and ongoing use as in Figure 2. Baseline thresholds (OT:threshold) and baseline threshold cumulative averages (OTCA:threshold cumulative average) are first set based on the biomarker statistics of patients with complications (e.g., leaks) versus patients without complications from dataset 414 from the clinical trial.

[0188] The method continuously learns based on data collected in memory 412 from the use of the embodiment by patients (see Figure 2) and whether they encountered complications.

[0189] The method measures one or more biomarker data points (pH is used as an example to illustrate the embodiments, but the method is not limited to pH) at regular intervals (for example, every 5 minutes). The data points are stored in the memory 412.

[0190] The method uses receiver operating characteristic (ROC) analysis to calculate a baseline threshold (OT) value for the pH data points 404.

[0191] The method determines, for each interval, the percentage of data points 406 that are below the calculated baseline threshold. The cumulative average (CA) of the percentage below the calculated baseline threshold is calculated for each interval 408.

[0192] A baseline threshold for the calculated cumulative average (OTCA) is calculated using receiver operating characteristic (ROC) analysis 410.

[0193] Figure 5 depicts a method 502 for determining the probability of postoperative complications after the method has been initialized.

[0194] The method measures one or more biomarkers at regular intervals (for example, every 5 minutes) during the current postoperative period.

[0195] The method calculates the percentage of data points within the current postoperative period interval that are below the OT and calculates the CA of the percentage of data points below the OT 506.

[0196] The current postoperative time interval is identified as "high risk" 512 if the CA exceeds the OTCA 508. Otherwise, it is marked as "low risk" 510.

[0197] If the number of consecutive postoperative hours marked as "high risk" exceeds a predetermined limit (e.g., 10), the method notifies the HCP that the patient has or is at high risk of developing postoperative complications by sending a message, resulting in noise in the device, etc.

[0198] If the patient develops a complication, the accumulated dataset of this patent is marked as such to improve the accuracy of the CA and OTCA thresholds.

[0199] The postoperative period can be 30 minutes, 1 hour, half a day, etc.

[0200] An IIR (infinite impulse response) or FIR (finite impulse response) time filter can be used instead of the CA to calculate the OTCA. For example, a moving average, Gaussian window, or exponential filter, etc.

[0201] Other non - linear statistical methods such as the median or mode can be used to determine the OT.

[0202] Other pre - operative, intra - operative, and post - operative data (e.g., American Society of Anesthesiologists (ASA) score, demographics, surgical details, etc.) can also be integrated in the method.

[0203] The method can combine multiple biomarker data or other source data available to the HCP (e.g., EMR / EHR, heart rate, other sensors or medical devices) to give a more accurate probability score. Different weights can be applied to different biomarkers to adjust their impact on the output risk / prediction. Other metadata about the patient can be integrated into the method. Statistical reliability can be added to the measured probability.

[0204] The method can be deployed within the firmware, enabled for updates and improvements, and executed on a processor or on a cloud server. The method can be adapted to detect different types of postoperative complications. The method can use pH data from any source.

[0205] Developing a model for assessing postoperative risk The disclosed embodiments can include a computer-implemented method for predicting a patient's postoperative risk value based on analyzing continuous patient data and patient history data against a separate risk model that includes real-time data and risk values of a plurality of patients and historical data and risk values of a plurality of patients.

[0206] In some embodiments, the computer-implemented method can be used in conjunction with a target patient's existing medical records, electronic medical records (EMRs), preoperative patient data, and the like.

[0207] As in the previous section, the biosignal data can be received from prior art detection systems, such as those described in U.S. Patent Application Publication No. 2022 / 0265175, which is owned by the assignee of the present application and is hereby incorporated by reference in its entirety.

[0208] Generally, for the purposes of the present disclosure, a target patient is a patient being monitored for complications and / or risks associated with a pre-operative, post-operative, or intra-operative state.

[0209] In addition, a plurality of patients can provide time-stamped data and historical data for training the various risk models disclosed herein. This data can be obtained from clinical trials, medical records, and the like.

[0210] In some embodiments, the computer-implemented method may be used in conjunction with an in-line device that can continuously measure real-time data of a target patient or multiple patients using sensors, and the real-time data is sent via a network to one or more risk assessment models included in a server to provide a risk value for the target patient or to train one or more risk assessment models to evaluate risk based on the real-time data.

[0211] The in-line device may include an in-line device having a catheter, or a device incorporated within the catheter, or a device disposed along the catheter.

[0212] In some embodiments, the computer-implemented method may be used in conjunction with an embedded medical device, a benchtop device, a point of care (POC), and a handheld device.

[0213] In some embodiments, the sensor may include different forms of electrochemical or solid-state sensors including, but not limited to, potentiometric measurements, voltammetry, conductivity measurements, capacitance, amperometric measurements, or ion-sensitive field effect transistors (ISFETs). In some embodiments, the sensor may be piezoelectric or a microelectromechanical system (MEMS). The sensor may include terminals connected to an active, counter, reference, or pseudo-reference electrode depending on the type of sensor being used. The sensor may be of different types including, but not limited to, a pH sensor, an ion-sensitive sensor, a temperature sensor, a lactate sensor, an electrolyte sensor, an impedance sensor, a light-based sensor, a microbial sensor, a protein sensor, a DNA-based sensor, a carbohydrate sensor, an enzyme sensor, an oxygen sensor such as a P02 (partial pressure of oxygen) sensor, an amylase sensor, a urea sensor, a creatinine sensor, a motion sensor (e.g., an accelerometer), a pressure sensor, and a flow sensor.

[0214] The sensors can be connected in series or in parallel and can be arranged sequentially, for example, along the length of a fluid flow path.

[0215] In some embodiments, the sensor can include one or more temperature sensors, such as a thermistor or a resistance temperature detector.

[0216] In use, the temperature sensor can be subject to a change in resistance that correlates with a change in temperature. Thus, the temperature can be determined by exciting with a current and measuring the voltage to determine the resistance of the thermistor (or vice versa).

[0217] In some embodiments, the sensor can include an essentially electrochemical pH sensor that enables a biological sample to be converted into an electrical signal that can be subsequently measured, monitored, and analyzed to determine whether postoperative complications are developing. A system of electrodes (active, counter, and reference) combined with each other can be fabricated on a biocompatible substrate. The electrodes can be fabricated from biocompatible materials such as gold, platinum, titanium, and silver and can then be functionalized with an active polyaniline (PANI), polyaniline / polyurethane (PAIN / PU), polyurethane, polymer, or other suitable layer. In one example, the m-biosensors are 500 pm × 500 pm in size and can be placed on or incorporated into a catheter to monitor changes in pH over time.

[0218] In some embodiments, the sensor can include a light-based sensor such as an optoelectronic sensor that uses a combination of a light transmitter or source and a detector in the ultraviolet to infrared spectrum to measure the light absorption or transmission characteristics of a fluid. A single wavelength or multiple wavelength light beams can be used.

[0219] In some embodiments, the sensor can include an impedance sensor that is typically operated with alternating current (AC) excitation and can be used to evaluate the user's situation. The impedance sensor can include an electrode pair and can include excitation and readout circuitry.

[0220] Advantageously, the systems and methods described below according to different embodiments function as an agnostic interface that allows data generated by different types of sources to be used. Thus, the system and method can be configured to interface with existing systems that measure patient biosignals.

[0221] FIG. 6 shows a method 600 implemented on a computer for evaluating a patient's risk value.

[0222] Generally, a method 600 implemented on a computer for evaluating the risk value of a target patient 620 first includes receiving, at a server 612, one or more of patient real-time data 602 and patient history data 606 from the target patient 620 via a connection mechanism, such as a network 618. The method further includes estimating, at the server 612, a risk value 616 of the measured signal of the target patient via a combined risk assessment model 610 that includes one or more real-time data risk assessment models 604 and one or more history data risk assessment models 608. The method further includes displaying or notifying a user of the risk value 616 via a display system 614.

[0223] The display system can include a computer, a smartphone, a physical printout, etc., as well as any other means for displaying data known in the art. The display system can be connected to the server via a network, a cable, or the like.

[0224] Notification to the user can include a notification sound, a visual notification, or the like, such as via a tactile, visual, auditory, or digital notification system.

[0225] Alternatively, the risk value can be stored without displaying or notifying the user of the risk value 616.

[0226] The risk assessment models 604 and 608 can preferably be trained separately.

[0227] Training the risk assessment models 604, 608 generally includes sending data to a server, associating known or measured data (e.g., by calibration, mapping, etc.) with a numerical value or indication of whether a complication has occurred, normalizing the data (e.g., by normalizing or subtracting the mean and dividing by the standard deviation), and performing one or more regression analyses on the normalized data to provide a model that associates data points with risk factors.

[0228] For example, the real-time data risk assessment model 604 can be trained using continuous sensor data from a plurality of sensors that measure biosignals from a first plurality of patients. The continuous sensor data can be sent, for example, from an in-line device 622 that can continuously monitor patient 620, and the patient real-time data 602 is sent to the server 612 via a connection mechanism such as the network 618, cable, USB device, or the like.

[0229] Alternatively, the continuous sensor data can be sent from an embedded device, bench-top device, POC device, handheld device, and / or medical wearable (i.e., smart watch).

[0230] The training data can include associating the time after surgery with the sensor data, whereby the time-stamped data can be associated with an indication of whether a patient among a plurality of patients has suffered a complication, and as a result, a regression curve can be developed associating the probability of suffering a complication at a given time after surgery with the sensor data.

[0231] For example, the time-stamped data used to train the real-time risk assessment model 604 may include pH versus postoperative time for each of a plurality of patients. The time-stamped data may be classified, for example, based on whether it corresponds to a patient who has encountered a complication. The time-stamped data may be classified, for example, based on when a given patient encountered a complication. The time-stamped data may be classified by a combination of the foregoing methods and any other methods known in the art.

[0232] The historical data risk assessment model 608 may be trained using historical data from a second plurality of patients including preoperative risk factors. The second plurality of patients may overlap with the patients from the first plurality of patients.

[0233] The historical data risk assessment model may be trained, additionally or alternatively, with patient or population level historical data.

[0234] The historical data may include, but is not limited to, one or more of: preoperative risk factors (e.g., demographics, weight, gender, age, etc.), medical records, surgical history, etc. of the subject patient or subject patient population.

[0235] For example, the historical data risk assessment model 608 may be trained with preoperative data of a plurality of patients, such as their ASA scores.

[0236] Training the combined risk assessment model 610 may include mapping the decision rules associated with the two individual models 604, 608 to the risk value 616. It may include rules for making decisions based on the availability of data within the two individual models 604, 608.

[0237] For example, the combined risk assessment model 610 can include an engine that makes a determination to select whether to use one or both of the risk assessment models 604, 608 before calculating the risk value 616. Some patients may not have continuous patient real-time data 602, in which case only the patient history data 606 including preoperative data or medical records 624 can be input into the method 600. In this case, the decision-making engine can choose to evaluate the risk based only on the historical data risk assessment model 608.

[0238] This enables the use of partially complete patient data in both risk assessment model training and validation (for example, when information about a patient's preoperative risk factors is available but their sensor data is not, and vice versa).

[0239] This is an alternative to more common methods such as imputing missing values or dropping patients with missing values.

[0240] Dropping patients with missing values limits the number of available training patients, and imputation can be problematic when the number of imputed values is large compared to the number of available values and not obvious for a complete collection of sensor measurements.

[0241] Furthermore, the models 604, 608, and 610 can be continuously updated by training them with the target patient data.

[0242] In general, but not necessarily, the risk assessment models 604, 608, and 610 are trained using machine learning algorithms or techniques.

[0243] These may include, for example, deep learning architectures such as deep belief networks (DBNs), stacked auto encoders (SAEs), convolutional neural networks (CNNs), or recurrent neural networks (RNNs) being used. Other examples may include restricted Boltzmann machines (RBMs), social restricted Boltzmann machines (SRBMs), fuzzy restricted Boltzmann machines (FRBMs), the TTRBM model of deep belief networks (DBNs), or similar techniques being used, the AE, FAE, GAE, DAE, BAE models of statistically adjusted end use (SAE) models being used, models such as the AlexNet, ResNet, Inception, VGG16, ECNN models of CNNs being used, the bidirectional recurrent neural network (BiRNN), long short-term memory (LSTM) network, gated recurrent unit (GRU) of RNNs also being used, including but not limited to these. Additional techniques specific to time series modeling may be used, including but not limited to dynamic time warping, change point detection, autoregressive integrated moving average (ARIMA).

[0244] In some embodiments, other types of algorithms such as physics-based mathematical calculations and basic multiple linear regression models may also be relied upon in conjunction with or complementary to their architectures and learning algorithms. This may further include the cumulative average (CA) method as described above and illustrated in FIGS. 4 and 5.

[0245] In one non-limiting example, method 600 may estimate the risk value 616 of anastomotic leakage after surgery.

[0246] The combined risk assessment model 610 generally uses the real-time data risk assessment model 604 and the historical data risk assessment model 608 to estimate the continuous real-time risk value 616 of patients who develop anastomotic leakage after general surgical anastomosis. In this embodiment, the historical data risk assessment model 608 includes a "preoperative risk factor model", while the real-time data risk assessment model 604 includes a pH model. This may preferably enable users, including healthcare providers, to obtain the real-time situation of their patients for earlier and closer monitoring, intervention, or discharge compared to the current standard of treatment.

[0247] For the target patient 620, their historical data 702, including preoperative risk factors such as an ASA score of 1, 2, 3, 4, or 5 and surgical procedures, such as laparoscopy and robotic laparotomy in this case, is transmitted to the server 612 via the network 618 and preprocessed. The preprocessing may include determining preoperative risk factors (e.g., ASA score) based on the patient's medical record. The preprocessing may include mapping or assigning numerical values to the preoperative risk factors. The historical data may also include the patient's individual health indicators, such as gender, height, weight, or past surgeries, past surgical complications, surgical parameters (including but not limited to duration, blood loss, etc.) and vital data (heart rate, respiratory rate, oxygen saturation, etc.).

[0248] The pre-processed historical data can then be input into the historical data - risk assessment model 608 using the standardized values calculated for the data used to train the model 608. The historical data - risk assessment model 608 then preferably uses a regression model that associates the probability of leakage with a given preoperative risk factor to estimate a risk value 616 based on the preoperative risk factor, resulting in an estimated probability of anastomotic leakage risk.

[0249] For the same patient, their patient real - time data 602 can be pre - processed for a first point in time or a first time period, in which case the real - time data is pH sensor measurement results obtained from a pH sensor device that can be in fluid communication with the patient 620. For example, the pH sensor device can continuously measure the pH of the patient's body fluid. Pre - processing of the pH sensor measurement results can include obtaining the average of the pH measurement results over a time period. This can be achieved by filtering and enhancing the raw data using a filtering and enhancement engine.

[0250] The pre - processed patient real - time data 602 can then be input into the real - time data - risk assessment model 604, which is trained using the pH sensor data in this case.

[0251] The real - time data - risk assessment model 604 can estimate a risk value based on the pH value at the first point in time or the first time period to produce an estimated probability of anastomotic leakage risk by using one or more regression models that associate the probability of leakage with a given pH value at a given postoperative time / time period and using the calculated average pH through one or more trained pH models.

[0252] The risk value can alternatively or additionally be estimated from the CA method of FIGS. 4 - 5.

[0253] For the same patient, at a desired time interval (e.g., at each given postoperative time), the combined risk assessment model 610 may preferably use both model risk values to classify the patient's anastomotic leakage risk based on a set of rules that preferably coincide with two risk models.

[0254] In one non-limiting example, the patient's anastomotic leakage risk is "very high" if the probability estimate of the preoperative risk factor model exceeds 0.5 and the probability estimate of the pH model exceeds 0.75, "high" if the probability estimate of the preoperative risk factor model exceeds 0.5 and the probability estimate of the pH model is between 0.5 and 0.75, 1) "low" if the probability estimate of the preoperative risk factor model exceeds 0.5 and the probability estimate of the pH model is less than 0.5 or 2) "low" if the probability estimate of the preoperative risk factor model is less than 0.5 and the probability estimate of the pH model exceeds 0.5, and "very low" if the probability estimate of the preoperative risk factor model is less than 0.5 and the probability estimate of the pH model is less than 0.5, and may be based on the following decision rules.

[0255] Figure 7 shows a system 700 for evaluating a patient's risk value.

[0256] Generally, the system comprises a server having one or more processors and a memory, the one or more processors being communicably coupled to a database 706, either via a network, cable, or the like or without, the database including historical data 702 from a plurality of patients and time-stamped data 716 from a plurality of sensors that measure biosignals from a plurality of patients. The one or more processors are further communicably coupled to subject patient real-time data 712 and subject patient historical data 714 via one or more networks, cables, or the like.

[0257] The memory includes instructions that configure the server 704 to perform the following when executed by one or more processors: 1) At server 704, receive one or more of the target patient real-time data 712 and the target patient history data 714. With respect to the target patient real-time data 712, the server may be configured to continuously receive the data 712 in real time. Alternatively, the server may be configured to continuously receive 712 at specific time intervals before, after, or during the surgery. 2) Preprocess or process one or more of the data 712 and 714 via a processor. 3) Use a first model to estimate a first risk value 616 associated with the target patient 710. For example, the first model may include a history data risk assessment model 608 that may be trained using the history data 702 from the database 706. The history data risk assessment model 608 may provide the risk value 616 based on preoperative risk factors. 4) Use a second model to estimate a second risk value associated with the target patient 710. For example, the second model may include a real-time data risk assessment model 604 that may be trained using the time-stamped data 716 from the database 706. The real-time data risk assessment model 604 may provide the risk value 616 based on real-time sensor measurement results that match the physiological conditions of the target patient 710 at a given time. The second model may continuously estimate the risk value 616, for example, at specific time intervals after the surgery. 5) Use the combined risk assessment model 610 to estimate a final risk value 616 associated with the target patient 710, and thereby determine a risk assessment 708 based on the final risk value 616. The risk assessment 708 may be determined based on a set of rules that match the values 616 estimated in steps 3 and 4. The risk assessment 708 is determined once, continuously in real time, or continuously at specific time intervals.

[0258] The system may further include a display system 614 communicatively coupled to a server such as a smart phone or a computer. One or more processors may be configured to send one or more of the first, second, or third risk values 616 and / or the risk assessment 708 to the display system to notify the user of the risk value. As an alternative or in addition, the system 700 may include an auditory notification system so that noise associated with the risk value 616 notifies the user of the risk value 616.

[0259] The risk assessment 708 may be in the form of, for example, "high", "low", "medium" to notify the risk of postoperative leakage. It may be in the form of a percentage. It may be sent to the patient, their healthcare provider, or other designated user.

[0260] The risk assessment 708 may include errors and / or confidence intervals generated by the quantity and quality of statistical values and data related to various models in the database 706.

[0261] The risk assessment 708 may be used to inform and / or automatically update the postoperative treatment plan for the target patient 710.

[0262] The combined risk assessment model 610 may be trained with one or more risk assessment models. The historical data - risk assessment model 608 may be trained based on various types of historical data including, but not limited to, preoperative risk factors such as ASA score, surgical procedure, patient medical records (i.e., EMR), and preoperative physical condition. The real - time data - risk assessment model 604 may be trained based on various types of time - stamped data 716 including biosensor data and the like.

[0263] FIG. 8 shows a schematic diagram illustrating a method for training a historical data - risk assessment model 800 that uses historical data 702.

[0264] The method includes sending historical data and an indicator of whether a patient associated with the historical data has a complication for a plurality of patients to the server 612, preprocessing the historical data 702 by mapping the historical data to a numerical value with a mapping engine 802, normalizing the data with a normalization engine 804, and inputting the mapped and normalized data into a historical data risk assessment model 608. The model 608 includes one or more regression models trained by associating an indicator of whether a given patient has a complication with the preprocessed historical data of the given patient for a plurality of patients.

[0265] In an example of anastomotic leakage risk assessment, training the historical data risk assessment model 608 may include the following: Preprocessing the historical data via a mapping engine 802 and a normalization engine 804: 1) Collecting historical data 702 from a plurality of patients - in this case, the historical data 702 includes preoperative risk factors such as the ASA score and surgical procedure (laparotomy, laparoscopy, robot), and an indicator of whether a patient associated with the historical data for each patient in the available training data has a complication. 2) Mapping the ASA score to a numerical value (e.g., ASA I: 1, ASA II: 2, ASA III: 3, ASA IV: 4, ASA V: 5) and the surgical procedure to a numerical value (e.g., laparoscopy and robot: 0, laparotomy: 1) via a mapping engine 802, and normalizing a set of training values via a normalization engine 804 to have a mean of 0 and a standard deviation of 1. Training of the historical data risk assessment model 608: 3) Training one or more regression models using the numerical values of the complications and related indicators. For example, a logistic regression model (using L2 penalty and balanced class weights) can be trained to predict a risk value 616 in the form of the probability that a patient has an anastomotic leakage given the numerical surgical procedure and ASA score.

[0266] The above is an example of a historical data risk assessment model 608 trained using preoperative data so as to give a preoperative risk factor history data risk value 806.

[0267] The historical data risk assessment model 608 can also be trained using individual health indicators of a patient such as gender, height, weight, or past surgeries, past surgical complications, surgical parameters (including but not limited to duration, blood loss, etc.), and vital data (heart rate, respiratory rate, oxygen saturation, etc.).

[0268] FIG. 9 shows a schematic diagram illustrating a method for training a real-time data risk assessment model 604 that uses time-stamped data 716.

[0269] The method generally includes sending time-stamped data 902 for a plurality of patients to a server by receiving signals from sensors coupled to the patients and an indicator of whether the patients associated with the time-stamped data 716 have suffered from complications, preprocessing the time-stamped data 716, which may include different techniques for filtering and enhancing the signals received from the sensors according to the characteristics of the time-stamped data 716 via a filtering and enhancement engine 904, normalizing the data with a normalization engine 906, and inputting the filtered, enhanced, and normalized data into the real-time data risk assessment model 604. Filtering the data based on a time-based average may include subsampling sensor measurements at a larger / smaller average (e.g., 30 minutes) or aggregating information using statistical values such as median / average.

[0270] Model 604 includes one or more regression models that are trained for a plurality of patients by associating time-stamped data for a given patient with an indicator of whether the given patient has developed a complication. When a target patient is evaluated in real time, the post-operative time along with the real-time sensor measurement results can be input into the real-time data risk assessment model 604 to estimate a real-time data risk value 908 of whether the target patient may develop a complication for a given post-operative time that can be deployed on the display system 614.

[0271] In an example of anastomotic leakage risk assessment, training the real-time data risk assessment model 604 can include the following: Preprocessing of pH data via the filtering and enhancement engine 904: 1) Removing incorrect sensor measurement results having pH and EC measurement results (from continuous measurement results from the in-line device 622 and any measurement results from the calibration fluid that are in fluid communication with the expected out-of-bounds patient). 2) Removing all pH measurement results for patients who have experienced anastomotic leakage after their diagnosis (this is only performed for patients used to train the model). 3) Thinning the pH signal filtered every 15 minutes on average. This time frame may be smaller or larger depending on the desired resolution of the data. Normalizing the pH data via the normalization engine 804: 4) For a given post-operative time, calculating the average pH sensor readings for the 24 hours prior to the post-operative time, and normalizing the calculated average from the training set to have a mean of 0 and a standard deviation of 1 (when making predictions for new patients, the normalized values calculated for the training data are used). Training the real-time data risk assessment model 604: 5) Using the standardized mean pH sensor readings of each patient within the available training data, train one or more regression models, such as a logistic regression model (using L2 penalty and balanced class weights), to predict a risk value 616 in the form of the probability that a patient has an anastomotic leak given their mean pH at that postoperative time. In some embodiments, multiple models may be trained at specific time intervals, such as hourly, after the surgery at which the risk value prediction is made.

[0272] The foregoing is an example of a real-time data risk assessment model 604 that is trained using continuous sensor measurements related to a patient's real-time state to continuously estimate a postoperative risk factor risk value 616.

[0273] The real-time data risk assessment model 604 can also be trained using data from different sensors measured in real-time or at point measurements (e.g., a blood gas analyzer), quantitative and qualitative data obtained by a healthcare professional (e.g., heart rate, heart rate variability, blood pressure, SpO2, temperature, ECG, etc.), color monitoring of patient fluids (blood, urine, bile, etc.) or any other relevant physiological data related to the patient's condition.

[0274] FIG. 10 is an exemplary process flow diagram of a method 1000 for estimating a patient's risk value.

[0275] The method 1000 starts at step 1002.

[0276] In step 1004, the system acquires historical data from the target patient. In step 1012, the system begins to continuously acquire real-time data from the patient. Generally, steps 1004 and 1012 are executed in parallel, but they can be executed sequentially, or in the case of a patient where one form of data may not be available, only one of these steps may be executed instead.

[0277] In step 1006, the historical information is pre - processed and processed using the historical data - risk assessment model 608. This step generally includes estimating the risk value from the processed historical data using the risk model 608. In step 1014, the real - time data is continuously pre - processed and processed using the real - time data - risk assessment model 604. This step generally includes estimating the risk value from the real - time data using the risk model 604. Generally, steps 1006 and 1014 are executed in parallel, but they can be executed sequentially, or in the case of a patient where one form of data is not available, only one of these steps can be executed instead.

[0278] In step 1008, the processed historical data or the estimated risk value from step 1006 is sent to the combined risk assessment model. In parallel or separately, in step 1016, the processed real - time data or the estimated risk value from step 1014 is sent to the combined risk assessment model.

[0279] In step 1010, the combined risk assessment model uses the risk values sent from the separate models in steps 1008 and 1016 to estimate the risk assessment for the patient.

[0280] In step 1018, the method ends.

[0281] Method 1000 may further include continuously evaluating the risk, that is, continuously repeating steps 512 - 1010 using the real - time data generated continuously.

[0282] It may further include notifying or displaying the risk assessment in the form of visual, auditory, or tactile notifications.

[0283] In some embodiments, different machine learning algorithms or techniques may be used alone or in combination in the aforementioned different engines.

[0284] These may include, for example, deep learning architectures such as deep belief networks (DBNs), stacked autoencoders (SAEs), convolutional neural networks (CNNs) or recurrent neural networks (RNNs) being used. Other examples may include restricted Boltzmann machines (RBMs), social restricted Boltzmann machines (SRBMs), fuzzy restricted Boltzmann machines (FRBMs), the TTRBM model of deep belief networks (DBNs) or similar techniques being used, the AE, FAE, GAE, DAE, BAE models of the statistically adjusted end use (SAE) model being used, models such as AlexNet, ResNet, Inception, VGG16, the ECNN model of CNNs being used, bidirectional recurrent neural networks (BiRNNs), long short-term memory (LSTM) networks, gated recurrent units (GRUs) of RNNs also being used, but not limited to these. Additional techniques specific to time series modeling, including but not limited to dynamic time warping, change point detection, and ARIMA, may be used.

[0285] In some embodiments, other types of algorithms, such as physics-based mathematical calculations and basic multiple linear regression models, may also be relied upon in conjunction with or complementary to their architectures and learning algorithms.

[0286] Many of the functional units described herein are labeled "engines" to more specifically emphasize their implementation independence. For example, an engine can be implemented as a hardware circuit including a custom VLSI circuit or gate array, a commercially available semiconductor such as a logic chip, a transistor, or other discrete component. An engine can also be implemented in a programmable hardware device such as a field programmable gate array, programmable array logic, programmable logic device or the like.

[0287] An engine can also be implemented in software for execution by various types of processors. For example, an identified engine of executable code can include one or more physical or logical blocks of computer instructions that can be organized, for example, as objects, procedures, or functions. However, the executable files of an identified engine need not be physically located together, but can include completely different instructions stored in different locations that, when logically combined, include the engine and achieve the defined purpose of the module.

[0288] In fact, an engine of executable code can be a single instruction or multiple instructions and can be distributed across different code segments, between different programs, and across several memory devices. Similarly, operational data may be identified and illustrated herein within an engine, can be implemented in any suitable form, and can be organized within any suitable type of data structure. Operational data can be collected as a single data set or can be distributed across different locations including across different storage devices and can exist at least partially as electronic signals on a simple system or network. When an engine or portion of an engine is implemented in software, the software portion is stored on one or more computer-readable storage media.

[0289] It should be understood that the models described herein can be trained using data from various sources.

[0290] Additional models can be coordinated with preoperative risk factors and pH models (e.g., Vital models) for use in conjunction with or as an alternative to the models described above.

[0291] Higher-performance individual models can be developed for use in conjunction with or as an alternative to the models described above. For example, those that predict probabilities using a preoperative risk factor model or a random forest model that also obtains the amount of blood loss during surgery.

[0292] Filtering data to a time-based average can include subsampling sensor measurements to a larger / smaller average (e.g., 30 minutes) or aggregating using statistical values such as the median / average.

[0293] Alternative methods for reconciling predictions from individual models can be used (e.g., moving to a logistic regression model instead of using a threshold for classification).

[0294] The systems and methods described herein can further include features such as prompts for requesting the input of additional data at critical times to update the model. For example, a prompt for HR at postoperative time 6 that can be used to update the risk assessment.

[0295] The combined risk assessment model or algorithm can be deployed in firmware or in the cloud and can be updated / improved as needed. It can also be made to detect different types of postoperative complications.

[0296] The combined risk assessment model can be used with an in-line device that continuously receives data in real time from a biosensor or sensor in fluid communication with the patient, or that interfaces with other sensor systems, or that receives manual input of data.

[0297] The present disclosure includes a system having a processor for providing various functionalities for processing information and for determining a result based on an input. Generally, the processing can be achieved using a combination of hardware and software elements. The hardware aspects can include a microprocessor, logic electrical circuit configurations, communication / networking ports, digital filters, memory, or a combination of operably connected hardware components including a logic electrical circuit configuration. The processor can be made to perform operations specified by computer-executable code that can be stored on a computer-readable medium.

[0298] The steps of the methods described herein can be achieved via software or a suitable programmable processing device that executes stored instructions or an on-board field programmable gate array (FPGA) or digital signal processor (DSP). Generally, the physical processors and / or machines used by embodiments of the present disclosure for any processing or evaluation can include one or more networked or non-networked general-purpose computer systems, microprocessors, field programmable gate arrays (FPGAs), digital signal processors (DSPs), microcontrollers, etc., programmed according to the teachings of exemplary embodiments understood by those of ordinary skill in the foregoing computer and software arts. As will be understood by those of ordinary skill in the software arts, suitable software can be readily prepared by an ordinary programmer based on the teachings of the exemplary embodiments. Additionally, as will be understood by those of ordinary skill in the electrical arts, the devices and subsystems of the exemplary embodiments can be implemented by the preparation of application-specific integrated circuits. Accordingly, the exemplary embodiments are not limited to any particular combination of hardware electrical circuit configurations and / or software.

[0299] Stored on any one or combination of computer-readable media, exemplary embodiments of the present invention may include software for controlling devices and subsystems of the exemplary embodiments and for processing data and signals, or for enabling devices and subsystems of the exemplary embodiments to interact with a human user or for similar purposes. Such software may include, but is not limited to, device drivers, firmware, operating systems, development tools, application software, and the like. Such computer-readable media may further include a computer program product of an embodiment of the present invention for performing all or part of the processing implemented in an implementation form (when the processing is distributed). The computer code device of the exemplary embodiment of the present invention may include any suitable interpretable or executable code mechanism, including but not limited to scripts, interpretable programs, dynamic link libraries (DLLs), complete executable programs, and the like.

[0300] Common forms of computer-readable media may include, for example, magnetic disks, flash memories, RAMs, PROMs, EPROMs, FLASH-EPROMs, or any other suitable memory chip or medium from which a computer or processor can read.

[0301] Although specific implementations and application examples of the present disclosure have been illustrated and described, it should be understood that the present disclosure is not limited to the exact structures and compositions disclosed herein, and that various modifications, changes, and variations may become apparent from the foregoing without departing from the spirit and scope of the present disclosure.

Claims

1. A method implemented on a computer for evaluating a risk value of a target patient, comprising: receiving target patient data at a server via a connection mechanism; estimating, at the server, one or more risk values associated with the target patient data using one or more risk assessment models; The method includes.

2. The method implemented on a computer according to claim 1, wherein the target patient data includes one or more of historical data, patient population level data, and real-time data.

3. The method implemented on a computer according to claim 2, wherein the one or more risk assessment models include one or more of a historical data risk assessment model, a real-time data risk assessment model, and a combined risk assessment model, and the combined risk assessment model includes one or more risk assessment models.

4. The historical data risk assessment model is receiving historical data corresponding to a plurality of patients at a server including one or more processors and a memory, wherein the one or more processors include one or more of a mapping engine and a normalization engine, and the historical data of each patient among the plurality of patients corresponds to an indicator of whether the patient has encountered a complication; mapping, via the mapping engine, the historical data of each patient among the plurality of patients to a numerical value; normalizing, via the normalization engine, the historical data of each patient among the plurality of patients; performing a regression analysis to determine a relationship between the historical data and the risk value on the preprocessed data; generating the preprocessed historical data thereby; The method implemented on a computer according to claim 3, which is trained thereby.

5. The real-time data risk assessment model is In a server comprising one or more processors and a memory, receiving time-stamped data corresponding to signal data measured by a plurality of sensors connected to a plurality of patients, wherein the one or more processors comprise one or more of a filtering and enhancement engine and a normalization engine, and the time-stamped data for each of the plurality of patients corresponds to an indicator of whether the patient has encountered a complication, filtering and enhancing the time-stamped data for each of the plurality of patients via the filtering and enhancement engine, normalizing the time-stamped data for each of the plurality of patients via the normalization engine, performing one or more regression analyses on the preprocessed time-stamped data to determine one or more relationships between real-time data and risk values, generating the preprocessed time-stamped data thereby, A computer-implemented method according to claim 3, which is trained thereby.

6. The computer-implemented method according to claim 2, wherein the historical data includes one or more of preoperative risk factors, medical records, surgical history, individual health indicators, and surgical parameters of the target patient or target patient population.

7. The computer-implemented method according to claim 2, wherein the real-time data includes sensor data from one or more sensors that continuously measure signals related to the physiological condition of the target patient.

8. Further comprising notifying the user of the risk value. The computer-implemented method according to claim 1.

9. The computer-implemented method according to claim 7, wherein the risk value is continuously estimated.

10. The computer-implemented method according to claim 1, wherein the risk value includes the probability that the target patient will develop a postoperative complication.

11. A system for evaluating the risk value of a target patient, comprising a server, one or more processors, and a memory, wherein the one or more processors are communicably connected to a database, and the database includes historical data and time-stamped data from a plurality of patients, The one or more processors are configured to receive target patient data via a connection mechanism, when the memory is executed by the one or more processors, in the server, receiving the target patient data; preprocessing or processing the target patient data via a processor; estimating one or more risk values related to the target patient data using one or more risk assessment models; A system comprising instructions for configuring the server to perform the above. **Claim 12** The system according to claim 11, wherein the target patient data includes one or more of historical data, patient population level data, and real-time data. **Claim 13** The system according to claim 11, wherein the one or more risk assessment models include one or more of a historical data risk assessment model, a real-time data risk assessment model, and a combined risk assessment model, and the combined risk assessment model includes one or more risk assessment models. **Claim 14** The historical data risk assessment model is receiving, in a server comprising one or more processors and a memory, historical data corresponding to a plurality of patients, wherein the one or more processors include one or more of a mapping engine and a normalization engine, and the historical data of each patient among the plurality of patients corresponds to an indicator of whether the patient has encountered a complication; mapping, via the mapping engine, the historical data of each patient among the plurality of patients to a numerical value; normalizing, via the normalization engine, the historical data of each patient among the plurality of patients; performing a regression analysis to determine a relationship between the historical data and a risk value on the preprocessed data; generating the preprocessed historical data thereby; The system according to claim 13, which is trained thereby. **Claim 15** The real-time data risk assessment model is In a server comprising one or more processors and a memory, receiving time-stamped data corresponding to signal data measured by a plurality of sensors connected to a plurality of patients, wherein the one or more processors comprise one or more of a filtering and enhancement engine and a normalization engine, and the time-stamped data for each patient among the plurality of patients corresponds to an indicator of whether the patient has encountered a complication, filtering and enhancing the time-stamped data for each patient among the plurality of patients via the filtering and enhancement engine, normalizing the time-stamped data for each patient among the plurality of patients via the normalization engine, performing one or more regression analyses on the preprocessed time-stamped data to determine one or more relationships between real-time data and risk values, generating the preprocessed time-stamped data thereby, The system according to claim 13, which is trained thereby.

16. The system according to claim 11, wherein the risk value includes the probability that the target patient will develop a postoperative complication.

17. The system according to claim 11, further comprising a display system for displaying the risk value.

18. A non-transitory computer-readable storage medium that, when executed by a computer receiving target patient data via a connection mechanism, preprocessing or processing the target patient data via a processor, estimating one or more risk values associated with the target patient data using one or more risk assessment models, A computer-readable storage medium comprising instructions for causing the computer to perform the above.

19. The non-transitory computer-readable storage medium according to claim 18, wherein the target patient data includes one or more of historical data, patient population level data, and real-time data.

Images