JP6524196B2 - System and method for transitioning patient care from signal based monitoring to risk based monitoring - Google Patents

Description

BACKGROUND The present invention relates to a risk based patient monitoring system and method. More particularly, the invention relates to a method of combining patient data from different sources to assess the current and future risks of a patient.

  Medical care is becoming increasingly complex with the introduction of new sensors and treatments. As a result, while the risk of the patient is reduced, the clinician is faced with a vast amount of patient data that needs to be evaluated and must be fully understood in order to prescribe the optimal treatment from the many options available. It will be done. One environment where such massive information becomes more problematic is the intensive care unit (ICU). In the intensive care unit, the physician's experience as well as the physician's ability to absorb available physiological information has a strong impact on clinical outcome. Hospitals that do not continuously secure trained intensive care physicians on a 24-hour basis have a 14.4% mortality rate compared to a 6.0% mortality rate for a fully staffed center. It became clear that it became. It is estimated that raising the level of care to the level of the average trained physician for all ICU can save 160,000 lives a year and save $ 4.3 billion. As of 2012, there is a shortage of intensive care physicians, and it is predicted that this shortage will only worsen, reaching 35% by 2020.

  The value of experience in critical care can be explained by the fact that the clinical data in the ICU is provided at a much higher rate than can be absorbed by the most capable physician, and the results have revealed That is, under the condition of excessive information, the probability of occurrence of error increases six times, and when there is a serious lack of time, it increases eleven times. Moreover, treatment decisions in the ICU are largely dependent on clinical signs that can not be measured directly but estimated from other physiological information. Thus, the clinician's expertise as well as the background plays an even more important role in the minute decision making process. Of course, this results in large variability in the estimation of hidden parameters. As an example, in severe patient care, although a number of parameters that substitute for cardiac output are monitored continuously, the results of the survey revealed subjective assessment by clinicians and thermal dilution It means that the correlation with the objective measurement by law is poor. Experienced intensive care physicians have effectively implemented risk management to incorporate such inherent uncertainty into their decision-making processes, ie treatment based on the most likely patient's condition Not only prescribe, but also weigh the patient's risk, which is otherwise the opposite. From this point of view, experienced intensive care physicians face excessive data in intensive care when they divert a large number of foreign signals from patient observation to risk assessment.

  Thus, what is clearly desired as a need in the ICU is to achieve a paradigm shift from signal-based patient monitoring to risk-based patient monitoring, thereby assisting and overcoming the physician in the face of massive data in the ICU It is a decision support system that makes it possible.

SUMMARY OF THE INVENTION According to the present invention, a risk based patient monitoring system for critical patient care is disclosed. The system combines data from any bedside monitor, electronic medical records or records, and other patient-specific information to assess current and future risks to the patient. Furthermore, this system can also be implemented as a decision support system that prompts the user to take a specific action according to a standardized medical plan when the patient's specific risk passes a predetermined threshold. Yet another implementation of the technology described herein is an outpatient monitoring system. According to this system, patient and family assessments are combined with information about drug dosing regimens and physician assessments to generate patient risk profiles, continuously track patient clinical trajectories, visit or additional patients Doctors are provided with decision support as to when to schedule a specific test.

  According to one embodiment, a risk based monitoring application is implemented in a system processor, which includes a data receiving module, a physiology observation module, a clinical trajectory interpreter module, and a visualization and user interaction module. According to one embodiment, the data receiving module may be configured to receive data from a bedside monitor, an electronic medical record, a treatment device, and to form an a priori information based assessment of the patient's clinical risk. It can be configured to receive any other information that may be relevant, as well as any combination of these.

  The physiology observation module utilizes a plurality of measurements to estimate a probability density function (PDF) of internal state variables (ISVs) that represent physiological function components associated with the patient's treatment and condition. The clinical trajectory module can be configured by multiple possible patient states, and given the estimated probability density function of the internal state variables, the probability that any of those patient states will occur with any probability It can be determined whether there is

  In various embodiments, the clinical trajectory interpreter module determines the condition of the patient capable of classifying the patient, and may further assume that the patient may be classified currently given the estimated probability density of the internal state variable. The state can also be determined. In this way, each possible patient condition is assigned a probability value of 0-1. The combination of patient status and their probability is defined as the clinical risk to the patient.

  The visualization and user interaction module receives: i) from the data reception module, continuously or intermittently acquired physiological measurements and patient specific identifiers such as conditions, demographics, visual inspection time series, ii) From the physiological function observation module, receive a time series of the probability density function of the estimated internal state variables, and from the clinical trajectory interpreter module, receive a time series of the probability that the patient is in a specific state and hazard levels of individual risks. The Visualization and User Interaction module then visualizes their data in a graph that shows the dependence of variables and time. The graphs are visualized by plotting the data directly on the screen, or, in the case of a probability density function, the likelihoods at specific times and specific values are encoded with color patterns and their data Visualize by plotting. The Visualization and User Interaction module can visualize the current risks to the patient by expressing them with boxes of different sizes and colors, respectively. In this case, the size of the box corresponds to the probability of patient condition at a particular point in time, and the color of the box corresponds to their hazard level. In addition to these, the Visualization and User Interaction module allows the user to set alarms based on patient state probabilities, share those alarms with other users, create notes related to patient risk, and their notes Can be shared with other users, as well as viewing other requirements in the patient's medical history.

According to one aspect of the invention, a computer implemented medium and method for risk based monitoring of a patient includes: That is,
A) acquiring by computer the data associated with a plurality of internal state variables each representing a parameter physiologically relevant to one of the patient's treatment and the condition;
B) storing the data associated with the plurality of acquired internal state variables in a computer-accessible storage device;
C) generating by computer a probability density function estimated for the plurality of internal state variables;
D) Based on the probability density function generated for the internal state variable, it is identified by the computer to which of the plurality of assumed patient states the patient can currently be classified and identified by the computer Generating by computer a probability value associated with each patient condition.

  According to one embodiment, the probability value associated with the identified assumed patient state is 0-1. According to another embodiment, the method according to the invention further comprises the step of: E) displaying on the screen the probability value and the identified individual possible patient states associated with the probability value. In this case, a combination of the identified assumed patient state and an individual probability value associated with the assumed patient state is defined as a clinical risk for the patient.

  According to another aspect of the present invention, a risk based monitoring system for monitoring a patient is provided with the following configuration. A processor, a storage device connected to the processor, a data receiving module connected to cooperate with a plurality of information sources associated with the patient, and a physiological function observation module communicating with the data receiving module; A clinical trajectory interpreter module in communication with the physiological function observation module is provided. The data receiving module acquires data associated with a plurality of internal state variables respectively representing parameters physiologically relevant to one of the patient's treatment and the condition. The physiological function observation module generates a probability density function of the internal state variable, and the clinical trajectory interpreter module is capable of classifying a patient at any of a plurality of possible patient states into which possible patient state. Identify and generate probability values associated with each of the identified putative patient states.

  According to one embodiment, the invention is further configured as follows. That is, the clinical trajectory interpreter module, the data receiving module, and a user interaction module communicating with the storage device are provided, and the user interaction module is identified corresponding to the probability value and the probability value. The individual contingent patient states are displayed, and the combination of the identified contingent patient states and the associated individual probability values is defined as a clinical risk for the patient.

According to yet another aspect of the invention, the system can utilize certain measurements, such as hemoglobin, with an unknown amount of latency, that is, these measurements can be made via the current point and the data communication link. It is valid in the past relative to the time of arrival. The physiology observation module can use backpropagation to handle such measurements out of sequence. Back propagation allows the current estimate of the ISV to be projected back to the valid point of the measurement so that information from the potential measurement can be properly incorporated. Thus, according to another aspect of the present invention, a computer implemented method for risk based monitoring of a patient comprises the following steps. That is,
A) computer-acquired data associated with a plurality of internal state variables each representing a parameter that is physiologically relevant to one of the patient's treatment and the condition, wherein the plurality of internal state variables are mapped Obtaining all but not all of the stored data in the same cycle;
B) storing the data associated with the plurality of acquired internal state variables in a computer-accessible storage device;
C) generating by computer a probability density function estimated for the plurality of internal state variables;
D) From the probability density function generated for the internal state variable, a first plurality of assumed patient states P (S ₁ ), P (S ₂ ), P (S ₃ ),. . . , P (S _n ) to which assumed patient state the patient could be classified at the preceding time point is identified by the computer, and the probability value associated with each identified leading assumed patient state is Computer generated steps.

According to one embodiment, the step of generating an estimated probability density function comprises: C1) at the current time step t _k , estimate the probability density function for the first plurality of internal state variables and generating, C2) using the defined transition probability kernel, said by progress in the opposite direction from the time step t _k definitive probability estimation until another time step t _kN, said at time step t _kN said another Generating a probability density function for a plurality of internal state variables. However, N is an integer greater than one.

  It should be understood at the outset that although exemplary approaches for one or more embodiments of the disclosure of the present application are described below, the systems and / or methods disclosed herein will now be known. However, although it is extant, it can be implemented using any number of technologies. In addition, the disclosure of the present application is not limited to the technology to be described below, including the exemplified implementation means, the drawings, and the exemplary designs and methods illustrated and described below, and is within the scope of the appended claims. Can be changed, with the corresponding maximum range.

Schematic showing the risk based medical monitoring environment according to the present invention Schematic showing the basic configuration of the risk-based physiological function observation module according to the present invention Schematic showing an exemplary graph of the probability density function for a selected ISV generated by the physiological function monitoring module according to the invention Schematic showing an exemplary graph of the probability density function for a selected ISV generated by the physiological function observation module according to the invention Schematic showing an exemplary graph of the probability density function for a selected ISV generated by the physiological function observation module according to the invention Schematic showing a non-limiting example of a physiological function observation process according to the invention Schematic showing a non-limiting example of a physiological function observation process according to the invention Schematic showing a timeline in which back propagation is used to incorporate information according to the invention Schematic showing an example of the process according to the invention for mean arterial pressure (ABPm) Schematic showing an example of resampling according to the invention Schematic showing a clinical trajectory interpreter module that performs state probability estimation using the combined probability density function of multiple ISVs to calculate the probability of various patient states according to the present invention A schematic diagram showing a non-limiting example of a patient condition definition adopted by a clinical trajectory interpreter module according to the present invention Provides a non-limiting example of how the clinical trajectory interpreter module can adopt the patient condition definition to assign the probability that a patient will be classified into each of the four possible patient conditions at a particular point in time Schematic A schematic diagram illustrating an alternative approach to estimating the probabilities of different patient states according to the present invention A schematic diagram showing a non-limiting example of the definition of patient status assigned hazard levels by the clinical trajectory interpreter module according to the present invention Schematic showing patient states and their individual probabilities incorporated into three graphs, referred to as pathogenesis according to the present invention Schematic showing an example of a pathogenesis tree for a given set of patient states and physiological variables according to the invention Schematic showing the method for calculating the utility of various measurements according to the invention Schematic showing one possible approach to incorporate external calculations generated from third party algorithms in accordance with the present invention A schematic diagram showing an example of an external calculation incorporation instruction according to the present invention Schematic showing additional examples of external calculation embedded instructions according to the invention Schematic showing an example of the functionality of the visualization and user interaction module according to the invention A schematic diagram showing an example of a summary view that can transmit the risk profile for each patient on a single screen in a specific hospital unit according to the present invention A schematic diagram showing an example of a method that can realize a view representing a patient's ongoing risk according to the present invention A schematic diagram showing how the slider at the top of the patient view can be used to browse patient risk history according to the present invention Diagram showing how the user can manipulate the pathogenesis tree by clicking on the synthesized patient state and viewing the component patient state according to the present invention Summary of the present invention by showing the expected risk for the patient as it can be viewed by the user within the same framework by sliding the slider to advance beyond the current point in time Figure A schematic diagram showing how the user can select and set an alarm for a particular risk according to the present invention Schematic representation of the patient risk trajectory according to the invention, ie a further possibility of visualizing the progression of patient condition probabilities associated with a particular patient risk Schematic representation of how the system can directly visualize the probability density functions of various internal state variables in accordance with the present invention Schematic showing an example of a tagging function that can be implemented by a user interface according to the present invention Schematic showing the news distribution screen of the user interface according to the present invention Schematic showing the condition summary screen of the user interface according to the present invention Schematic showing the ability of the user interface to include and display reference material accessible through the internet or storable in the system according to the invention Schematic showing a general purpose Dynamic Bayesian Network (DBN) that can be employed to capture a physiological model of a patient in HLHS Stage 1 palliative state according to the present invention Schematic showing several equations available to model physiological dynamics in HLHS stage 1 according to the present invention A schematic diagram illustrating an equation that can be used to extract relationships between dynamic variables and derived variables in modeling according to the present invention Schematic showing the possibilities of observation models that can be used to correlate derived variables with available sensor data according to the invention In accordance with the present invention, a schematic diagram showing the attributes, patient status, and possible pathogenesis trees available to the clinical trajectory interpreter module in the HLHS Stage 1 population. In accordance with the present invention, a schematic diagram showing one possibility for an environment in which a risk based monitoring system can be employed to assist the clinician in determining whether a particular treatment should be applied. Schematic showing a non-limiting example of a set of patient status related to blood transfusion that can be used to notify the determination of blood transfusion according to the present invention Schematic showing another application of the risk based monitoring system applied to a standardized medical plan according to the present invention Schematic showing one application case for a risk based monitoring system combined with a specific type of standardized clinical plan according to the present invention Schematic showing an example of risk stratification that can be adopted by the system in the context of nitric oxide therapy according to the present invention In accordance with the present invention, a schematic diagram illustrating one possible patient condition that can describe the clinical trajectory of an ADHD patient Charts listing available patient assessment modalities as M1, M2, M3 In accordance with the present invention, a schematic diagram showing a dynamic model of the progression of a patient from one state to another, extracted by a dynamic Bayesian network Schematic diagram showing alternative embodiments with regard to two predictions of how patient condition may shift in a single month when changing medication or dosage according to the present invention Schematic representation of one possible embodiment of a visualization and a scenario showing the patient's clinical trajectories and risks according to the invention A schematic illustrating the assessment of patients and patient trajectories at week 9, at which point a risk-based patient monitoring system determines the probability distribution function for each patient's condition for the past 9 weeks according to the present invention Figure Diagram showing follow-up evaluation based on Vanderbilt diagnosis of teacher and parent according to the present invention In accordance with the present invention, a diagram showing all the available measurements that are likely to produce significant improvements, ie the resulting assessment based on the visit of the work, the parent and the teacher's assessment Diagram showing yet another follow-up according to the present invention, when it is ensured that the patient is steadily improved through nine minutes and stable improvement is observed between the two evaluations. Figure shows a follow-up assessment according to the invention on patients and patient trajectories in the absence of measured values, showing a state of increased uncertainty due to the lack of recent observations Diagram of the above mentioned uncertainty when complete patient assessment (all measurement modalities) is performed according to the present invention It shows the possibility of further visualization from the already described system output, and shows, according to the present invention, the possibility of the transition of the patient state when the treatment plan is changed, for example, the drug is changed. Figure

DETAILED DESCRIPTION Disclosed herein are techniques for providing clinical staff with risk-based patient monitoring for individual patients. The techniques described herein can be implemented as a monitoring system for severe patient care. The system combines data from various bedside monitors, electronic medical records or records, and other patient-specific information to assess current and future risks to the patient. Furthermore, the technology can also be implemented as a decision support system, which prompts the user to perform specific actions according to a standardized medical plan when the specific risk of the patient passes a predetermined threshold. Yet another embodiment of the already described technology is an outpatient monitoring system. According to this system, patient and family assessments are combined with information about drug dosing regimens and physician assessments to generate patient risk profiles, continuously track patient clinical trajectories, visit or additional patients Doctors are provided with decision support as to when to schedule a specific test.

System Modules and Interactions Referring now to the drawings, FIG. 1 illustrates a risk based medical monitoring environment 1010. This provides a health care provider, such as a physician, nurse or other health care provider, with risk based monitoring according to various embodiments of the present invention. In this case, the patient 101 can be connected to one or more physiological function sensors or bedside monitors 102, which allow monitoring of various physiological parameters of the patient. These physiological function sensors include, but are not limited to, blood oximeters, blood pressure measuring devices, pulse measuring devices, glucose measuring devices, one or more analytical measuring devices, electrocardiogram recording devices, etc. Can be included. In addition to these, routine testing and testing can be performed on the patient, and data can be stored in electronic medical record EMR (103). Electronic medical records 103 include, but are not limited to: hemoglobin, arterial and venous oxygen content, lactic acid, body weight, age, gender, ICD-9 code, capillary refill time, clinician's subjective Information such as observations, patient assessments, prescribed medications, medication regimens, genetics, etc. can be stored. Patient 101 may also be coupled to one or more treatment devices 104 configured to treat the patient. In various embodiments, the treatment device 104 can include an extracorporeal membrane oxygenator, a ventilator, a drug infusion pump, and the like.

  According to the disclosure of the present invention, improved risk-based monitoring can be applied to the patient 101 via existing methods. A patient specific risk based monitoring system (generally referred to herein as system 100) may be configured to receive patient related information. Examples of patient-related information include real-time information from the bedside monitor 102, EMR patient information from the electronic medical record 103, information from the treatment device 104 such as settings, infusion rate, and drug type. Furthermore, other patient related information may include patient's medical history, previous treatment plan, previous clinical test result and present clinical test result, allergy information, disease predisposition for various conditions, and further patient's occurrence. Evaluation based on prior information on the conditions and conditions to be obtained and the possibilities associated with them. For the sake of simplicity, the various types of information listed here in general will be referred to as "patient specific information" below. Additionally, the system can be configured to utilize the received information to determine a clinical risk, which can then be presented to the healthcare provider. It should be noted that health care providers include, but are not limited to, physicians, nurses or other types of clinical staff.

  In various embodiments, the system includes one or more of a processor 111, a storage device 112 coupled to the processor 111, and a network interface 113 configured to allow the system to communicate with other devices via a network. It is included. Moreover, the system can also include a risk based monitoring application 1020. The application may include computer-executable instructions that, when executed by processor 111, allow the system to perform risk-based monitoring of a patient, such as patient 101.

  The risk-based monitoring application 1020 includes, for example, a data reception module 121, a physiological function observation module 122, a clinical trajectory interpreter module 123, and a visualization and user interaction module 124. According to one embodiment, the data receiving module 121 can be configured to receive data from the bedside monitor 102, the electronic medical record 103, the treatment device 104, and further to a priori information about the patient's clinical risk. It may be configured to receive any other information that may be relevant to the formation of the rating based on it, as well as any combination of these.

The physiological function observation module 122 utilizes a plurality of measurements, and according to a predefined physiological function model, a probability density function (PDF) of internal state variables (ISVs) representing physiological function components related to patient treatment and condition Estimate). The ISV may be directly observable with noise (in non-limiting examples, the heart rate may be directly observable ISV), or it may be invisible or hidden ( By way of non-limiting example, the oxygen supply DO ₂ defined as aortic blood oxygen saturation can not be measured directly and therefore is invisible) or may be measured intermittently (non-limiting) For example, the concentration of hemoglobin measured by complete blood count is an intermittently observable ISV).

  According to one embodiment, rather than assuming that all variables can be deterministically estimated without error, according to the physiological function observation module 122 of the present invention, a probability density function is provided as an output . Further details regarding the physiological function observation module 122 will be described later.

  The clinical trajectory module 123 can, for example, be configured by multiple possible patient states, and given the estimated probability density function of the internal state variables, any of those patient states will occur with any probability It can be determined whether there is a possibility of doing. A patient's condition may be defined as a qualitative description of physiology at a particular point in a clinical trajectory that can be recognized by the practice and may have implications for making a clinical decision. Examples of specific patient conditions include, but are not limited to, hypotension with sinus tachycardia, hypoxia with myocardial depression, compensated circulatory shock, cardiac arrest, hemorrhage etc. I will mention. Further, these patient states can be specific to a particular medical condition, and the range of each patient state can be defined by various physiological variables and data thresholds. In various embodiments, the clinical trajectory interpreter module 123 can utilize the reference material, information provided by the healthcare provider, and any information collected from other sources to determine the condition of the patient. Patients can be classified according to their condition. The reference material may be stored, for example, in a database or other storage device 130 accessible to the risk based monitoring application 1020 via the network interface 113. These references should include references, medical literature, expert surveys, physician-provided information, and materials compiled from any other sources available as a reference for providing medical care to the patient. Can. According to some embodiments, the clinical trajectory interpreter module 123 can initially identify a patient population similar to the patient being monitored. In this way, the clinical trajectory interpreter module 123 can utilize relevant historical data based on the identified patient population to help determine possible patient status.

  The clinical trajectory interpreter module 123 can also determine possible patient states given the estimated patient state probability density function provided by the physiology observation module 122, and easily classifies patients into those patient states. be able to. In this way, each possible patient condition is assigned a probability value of 0-1. The combination of patient status and their probability is defined as the clinical risk to the patient. Further details regarding the clinical trajectory interpreter module 123 will be described later.

  The visualization and user interaction module 124 is configured to receive the outputs of the data receiving module 121, the physiology observation module 122, and the clinical trajectory interpreter module 123, and present them to the clinical staff. The visualization and user interaction module 124 is calculated by the two modules 122 and 123 as the by-products of the current patient, their evolution over time, the probability density function of time-dependent internal state variables, and It can show other useful features for medical practice. In addition to this, the visualization and user interaction module 124 allows the user to set alarms based on patient state probabilities, share those alarms with other users, create notes related to patient risk, Notes can be shared with other users, as well as viewing other requirements in the patient's medical history. Further details regarding the visualization and user interaction module 124 will be described later.

Physiological Function Observation FIG. 2 shows a basic schematic diagram of the physiological function observation module 122. This module uses two models of patient physiology: a dynamic model 212 and an observation model 221. Dynamic model 212 to model captures the relationship that occurs between the internal state variables at time t _{k + 1} which approaches internal state variables and another at a certain point in time t _k, thereby the patient physiology as one system The current state of the system has information about possible future developments in this system. If the patient's physiology tends to be maintained at homeostasis by autoregulation, a clear rationale to introduce this kind of memory to internal state variables representing homeostasis, such as oxygen delivery and oxygen consumption. Sex exists.

The observation model 221 can capture the relationship between measured physiological function variables and other internal variables. An example of this type of model is given below:
a) Dependence of the difference between systolic arterial pressure and diastolic arterial pressure (also referred to as pulse pressure) in stroke volume b) Relationship between heart rate and stroke volume and cardiac output c B) Relationship between hemoglobin concentration and cardiac output and oxygen delivery d) Relationship between the Vanderbilt Assessment Scale and the clinical status of patients with attention deficit / hyperactivity disorder e) Measurable and therefore observable parameters And some other relationship between internal state variables.

The physiological function observation module 122 functions as a recursive filter, wherein the information based on the previous measurements is used to generate the prediction of internal state variables and the likelihood of possible future measurements, which are It is compared to the last measured value obtained. In particular, the physiological function observation module 122 uses the dynamic model 212 in the prediction step or prediction mode 210 and uses the observation model 221 in the update step or update mode 220. During the prediction mode 210, the physiology observation module 122 creates estimated PDFs of the ISVs 213 at the current time step t _k and supplies them to the dynamic model 212, which the dynamic model 212 proceeds to the next time step T _k Generate a prediction 211 of the ISV for ₊₁ . This is achieved by using the following equation:

ただし、ISVs(t_k) = {ISV₁(t_k),ISV₂(t_k),ISV₃(t_k),...lSV_n(t_k)}およびM(t_k)は、時点ｔ_kまでのすべての測定値の集合である。確率P(ISVs(t_k+ l)|ISVs(t_k))によって、ダイナミックモデル２１２を表す遷移確率カーネルが定義され、これによれば推定されたＰＤＦが時間の経過とともにどのように進展するのかが規定される。確率P(ISVs(t_k)|M(t_k))は推論エンジン２２２によって供給され、これは先行の時間ステップにおいて測定値が取得されている場合の、ＩＳＶの事後確率である。生理機能観測モジュール１２２のアップデートモード２１０中、予測されたＩＳＶ２１１が、データ受信モジュール１２１から受け取った測定値と、観測モデル２２１を用いて比較され、その結果、ＩＳＶがアップデートされて、新たに利用可能な情報を反映させる。モジュール１２２の推論エンジン２２２は、予測されたＰＤＦを事前確率として用いることによってこのアップデートを達成し、これは測定値の統計を用いてアップデートされて、現在のＩＳＶのＰＤＦ推定値を反映する事後確率を達成する。推論エンジン２２２は、ベイズの定理である次式を用いてアップデートステップ２２０を達成する。

_{However, ISVs (t k) = {} ISV 1 (t k), ISV 2 (t k), ISV 3 (t k), ... lSV n (t k)} and M (t _k) is the time t It is the set of all measurements up to _k . Probability P | by _{(ISVs (t k + l)} ISVs (t k)), are defined transition probability kernel that represents the dynamic model 212, how to progress with the lapse estimated PDF is time, according to this Is defined. The probability P (ISVs (t _k ) | M (t _k )) is supplied by the inference engine 222, which is the ISV's a posteriori probability when measurements have been taken at prior time steps. During the update mode 210 of the physiological function observation module 122, the predicted ISV 211 is compared with the measured value received from the data reception module 121 using the observation model 221, as a result, the ISV is updated and newly available. Information is reflected. The inference engine 222 of module 122 achieves this update by using the predicted PDF as a priori probability, which is updated with the statistics of the measurements to a posteriori probability to reflect the current ISV's PDF estimate Achieve. The inference engine 222 accomplishes the update step 220 using the following equation, which is Bayesian theorem.

ただしP(m₁(t_k+1), m₂(t_k+1),... m_n(t_k+1)|ISVs(t_k+1))は、観測モデル２２１によって供給される条件付き尤度カーネルであり、これによれば現時点で予測されているＩＳＶに対し、現時点で受け取った測定値がどの程度の尤度であるかが求められる。

Where P (m ₁ (t _{k + 1} ), m ₂ (t _{k + 1} ),... M _n (t _{k + 1} ) | ISVs (t _{k + 1} )) is supplied by the observation model 221 A conditional likelihood kernel, which determines the degree of likelihood of the currently received measurement value for the currently predicted ISV.

  If the current estimate for the ISV's PDF is not available at an initial time point, for example t = 0, the physiology observation module 122 can use the initial estimate 250, which has experience with possible values for the ISV. Or from statistical analysis of previously collected patient data.

A non-limiting example of a model that enables physiological function observation in accordance with the present invention is shown in FIG. Management of the oxygen delivery DO ₂ is an important part of patient care, although it is not directly observable. Therefore, accurate estimation of DO ₂ can lead to improved medical care. In the example shown, this estimate hemoglobin concentration (Hg), heart rate (HR), arterial pressure diastolic and systolic, and is achieved by SpO _2. The dynamic model 212 assumes that the oxygen delivery is driven by a feedback process that stabilizes against stochastic disturbances. Similarly, the hemoglobin concentration is controlled around the nominal value of 15 mg / dL.

The observation model 221 is the arterial oxygen saturation SpO ₂ , the hemoglobin concentration and arterial oxygen concentration CaO ₂ , the dependence of the difference between the systole ABP _s and the diastole ABP _d , the arterial pressure (stroke) Consider also the relationship between the heart rate HR, the stroke volume SV and the cardiac output). These two models are summarized as a dynamic Bayesian network (DBN), and the physiological function observation module 122 utilizes this DBN to continuously track the oxygen transport amount. A dynamic Bayesian network is a systematic approach that graphically represents statistical dependencies, where the vertices of the graph represent variables (observable and unobservable), and the edges represent causality. DO ₂ is a detailed description of the DBN, which forms one example for the estimation, United States Provisional Application Serial No. 61 / 699,492, 9 May 11, 2012 application, entitled "SYSTEMS AND METHODS FOR EVALUATING CLINICAL TRAJECTORIES AND TREATMENT STRATEGIES FOR OUTPATIENT CARE ", Attorney Docket No. 44429-00107 PROV, and U.S. Provisional Application No. No. 61 / 684,241, filed August 17, 2012, entitled" SYSTEM AND METHODS FOR PROVIDING RISK ASSESSMENT IN ASSISTING CLINICIANS WITH EFFICIENT AND EFFECTIVE BLOOD MANAGEMENT ", Attorney Docket No. 44429-00106 PROV, and the present application claims priority to these, the disclosures of which are incorporated herein by reference. Do.

A non-limiting example of the above physiological observation of tracking DO ₂ is shown in FIG. 4, but is shown here for tracking over a longer time interval, ie tracking over four steps. In this observation, the main hidden ISV is the oxygen delivery variable (DO ₂ ). Within the dashed circle in FIG. 4, two types of measurements are depicted: hemoglobin (Hg) and oxygen (SpO ₂ ). SpO ₂ is a continuous or periodic measurement, and the physiological function observation module 122 is connected to the patient 101, for example, a sensor such as a bedside monitor 102 and a treatment device 104 that continuously reports information. Receive from Hemoglobin (Hg) is an example of intermittent or aperiodic measurement extracted from a patient's clinical examination, based on sporadic and irregular bases, and sometimes potentially relative to the current system time Can be used by the observer. The physiology observation module 122 can handle both types of measurements. The reason is that module 122, with tracking of hidden ISVs such as DO ₂ , continues to estimate observations continuously for all types of measurements, even if no measurements are present. FIG. 4 depicts these estimates for the SpO ₂ and Hg cases. As can be seen from this figure, SpO ₂ measurements are regularly available at each time step, while Hg is only available at two time steps.

  As mentioned above, the system can use certain measurements, such as hemoglobin, with an unknown amount of latency, that is to say, these measurements are for the current moment, ie for the moment of arrival via the data communication link. It is effective in relative past time. The physiology observation module 122 can use backpropagation to handle such measurements out of sequence. Back propagation allows the current estimate of the ISV to be projected back to the valid point of measurement so that information from the potential measurement can be properly incorporated.

Such a timeline is depicted in FIG. As shown in FIG. 5, hemoglobin arrives at the current system time point T _k , which is valid at time point T _k−2 and is mapped to the current ISV (DO ₂ ). Back propagation is a method of updating the probability estimate P (ISVs (t _k ) | M (t _k )) of the current ISV with the potential measurement value m (t _kn ) relative to the current point in time . Back propagation is realized by the same method as the above-described prediction method. In this case, there is a transition probability kernel P (ISVs (t _kn ) | ISVs (t _k )), which defines how the current probability evolves in the reverse direction in time. If there is a current set of measurements excluding potential measurements, this can be used as follows to calculate the probability of ISV at time t _kn :

これらの確率が計算されると、ベイズのルールを用いることで、潜在的な測定情報が標準的なアップデータに組み込まれる。

Once these probabilities are calculated, potential measurement information is incorporated into standard updates using Bayesian rules.

その後、前述の予測ステップを用いて、アップデートされた確率が現在時点ｔ_kにバックプロパゲートされる。情報を組み込むために、バックプロパゲーションを利用することができる。

Thereafter, the updated probability is back-propagated to the current time t _k using the aforementioned prediction step. Back propagation can be used to incorporate the information.

Another function of the physiological function observation module 122 is smoothing. A healthcare provider utilizing system 100 may be interested in the patient's condition at some point in time. Smoothing allows the physiology observation module 122 to provide a more accurate estimate of the patient's ISV at a past time. In this case, all new measurements received by the system since that time are incorporated, resulting in a better estimate than the original filtered estimate of the overall patient condition at that time, then P ( ISVs (t _kn ) | M (t _k )) is calculated. This is achieved by using the first step of back propagation. According to this step, the estimation of the probability at this point, incorporating all the measurements up to the point t _k , is unfolded back to the point of _interest t _kn using the defined transition probability kernel . This is also illustrated in FIG. 5, according to which the user is interested in the patient condition at time T _kn and the estimation is smoothed back to this time.

  Because the physiological function observation module 122 maintains an estimate of each of the measurements available to the system 100 based on the physiological and statistical models, the module 122 measures measurements that are not related to the current information contained in the measurements. Can filter out artifacts. This is done by comparing the newly obtained measurements with the predicted likelihood of the possible measurements, if there is a previous measurement. If it is considered by the model that new measurements can not occur with considerable probability, those measurements are not incorporated into the estimation. The process of comparing the measured values with their prediction likelihood effectively filters the artifacts and reduces the noise.

  An example of this type of process involving mean arterial pressure (ABPm) is shown in FIG. Since ABPm is collected using a venous catheter, the measured signal is often degraded by artifacts when the catheter is used in medical procedures such as blood collection or line flushes. FIG. 6 shows the raw ABPM measurements before being processed by the physiology observation module, together with the artifacts of the identified measurements, and further filtered after being processed by the physiology observation module 122 Measured values are shown. As can be seen from this figure, the artifacts of the measurements have been removed, leaving the correct signal.

  According to various embodiments, the physiological function observation module 122 can utilize multiple algorithms for estimation or inference. Depending on the physiological function model used, the physiological function observation module 122 can use an appropriate inference scheme, such as a junction tree algorithm, or an approximate inference scheme using Monte Carlo sampling such as a particle filter, a Kalman filter, etc. Gaussian approximation algorithms, or any variation thereof.

  As already mentioned, the physiological function model used by the physiological function observation module 122 can be implemented using a probabilistic framework known as a dynamic Bayesian network. A dynamic Bayesian network is to treat causal relationships and probabilistic relationships between each ISV of a system as a graph, and these causal relationships and probabilistic relationships are those in only one time instance and the passage of time. As well as Because this type of model representation is flexible, the physiological function observation module 122 can utilize different inference algorithms. The choice of algorithm depends on the nature of the physiological function model used, the inference accuracy required by the application case, as well as the computational resources available in the system. As used herein, accuracy refers to whether an accurate inference scheme is used or an approximate estimation scheme is used. If the complexity of the physiological function observation model is limited, it is appropriate to use an accurate inference algorithm. In other cases, with more complex physiological observation models, there is no closed form inference solution or, if it exists, it can not be computationally processed in the available resources. In this case, an approximate inference scheme can be used.

  The simplest case where accurate reasoning can be used is where all the ISVs in the physiological function model are continuous variables and the relationships between each ISV in the model are limited to linear Gaussian relations. In this case, a standard Kalman filter algorithm can be used to perform the inference. According to this type of algorithm, the probability density function for the ISV is a multivariate Gaussian distribution, represented by the mean and covariance matrix.

  If all of the ISVs in the model are discrete variables and the structure of the graph is restricted to chains or trees, the physiological function observation module 122 may use forward-backward algorithm or probability propagation algorithm for reasoning, Each can be used. The junction tree algorithm is a generalization of these two algorithms that can be used regardless of the underlying graph structure, so the physiology observation module 122 can also use this algorithm for reasoning. The junction tree algorithm comes with additional computational costs, which may not be acceptable for the application case. In the case of discrete variables, the probability distribution function may be represented as a flat plate. It is noted that although it is possible to use this algorithm for inference if the model consists only of continuous variables with linear Gaussian relations, in this case this algorithm is equivalent to the Kalman filter. Therefore, a Kalman filter is used as an example of the algorithm.

  If the physiological function model consists of both continuous and discrete ISVs that have a non-linear relationship between variables, accurate inference solutions are not possible. In this case, the physiology module 122 can use an approximate inference scheme based on sampling techniques. The simplest version of this type of algorithm is a particle filter algorithm that uses sequential importance sampling. The Markov Chain Monte Carlo (MCMC) sampling method can also be used for more efficient sampling. For complex non-linear physiological relationships, this type of approximate inference scheme provides the highest flexibility. As would be obvious to one skilled in the art, the models and inferences employed by the physiology observation module may be any combination of the modeling and inference techniques described above, or other equivalent modeling and inferences. Techniques may also be included.

  If particle filtering is used, a resampling scheme is needed to avoid degeneration. The physiological function observation module can use an adaptive resampling scheme. As described in detail below, the region of the ISV's state space can be mapped to different patient states as well as different hazard levels for the patient. As the numbers get higher, the patient's condition becomes more dangerous to the patient's health. A sufficient number of sampled particles may be required in the region to ensure that an accurate estimate of the probability of a particular condition of the patient can be made. The most important possibility is to maintain an accurate estimate of the probability of the region with high hazard level, so the adaptive resampling approach makes enough particles in high hazard region in state space Will be sampled reliably. An example of this resampling is shown in FIG. States 1 and 2 have the highest hazard levels. The left plot depicts samples generated by standard resampling. It should be noted that, of course, more particles exist in these states, as states 1 and 2 are the most likely. The plot on the right shows the impact of adaptive resampling. Note that in this case the number of samples has increased significantly in the area of highest risk.

The Clinical Trajectory Interpreter Referring now to FIG. 8, the clinical trajectory interpreter 123 receives the joint probability density function of the ISV from the physiology function observation module 122 and performs a state probability estimation 801 to calculate the probabilities of various patient states Do. The probability density function of the ISV can be defined in closed form, for example as the multi-dimensional Gaussian curve 260, or approximated by the histogram 280 of the particle 270, as shown in FIGS. 2B-D. be able to. The probability density function of ISV in both cases can be expressed as P (ISV1 (t), ISV2 (t), .., ISVn (t)), where t is the time associated with them. Given an internal state variable, the patient state can be defined by the conditional probability density function:
P (S | ISV1, SV2, ..., ISVn) where _{S ∈ S 1, S 2,} ..., S n represents all possible patient state S _i. The determination of the probability of the patient in a particular state S _i can then be performed by

If P (ISV1 (t), ISV2 (t), .., ISVn (t)) is defined by a closed-form function such as multidimensional Gaussian curve 260, execute integration as it is Can. P (ISV1 (t), ISV2 (t), .., ISVn (t)) is approximated by a histogram 280 of the particle _{270, ISV 1, ISV 2,} ..., a space over LSV _n, in FIG. 9 In the case where P (S | ISV ₁ , ISV ₂ , ..., ISV _n ) is defined by dividing it into areas as shown, calculate the proportion of particles 270 for each area The probability P (S _i (t)) can be calculated by

  Once the patient state probability is estimated, the clinical trajectory interpreter module 123 can assign a different hazard level 802 for each patient state, or each state can be incorporated into different etiologies 803. The clinical trajectory interpreter module 123 may cooperate with the physiology observation module 122 to perform the measurement validity determination 804, whereby different observations, such as, for example, open blood pressure or open oxygen concentration monitoring, are performed. Determine the effectiveness of blood measurements. According to one embodiment, the clinical trajectory interpreter module 123 determines the probability that the patient is in a particular state, rather than the exact state the patient is in.

FIG. 9 depicts a non-limiting example of patient condition definitions that may be employed by the clinical trajectory interpreter module 123. Particularly assumed here, the internal state variables ISV _1, ISV _2, ..., by resolution of the region spanning ISV _n, the function P | definition _{(S ISV 1, ISV 2,} ..., ISV n) the It is possible. According to a particular example, the physiology of the patient is defined by two variables: pulmonary vascular resistance (PVR) and cardiac output. The specific risks and individual etiologies that can be captured by these two ISVs are based on the effects of increased pulmonary vascular resistance on the circulatory system. In particular, high PVR can cause a reduction in right heart failure and thus cardiac output. Thus, by assigning thresholds to two variables, PVR can be used to define PVR normal and PVR high attributes, and CO can be used to define CO normal and CO low attributes. By combining these attributes, four distinct states can be defined: State 1: CO Low, PVR Normal; State 2: CO Low, PVR High; State 3: CO Normal, PVR High; State 4: CO normal, PVR normal.

  Figure 10 shows how the clinical trajectory interpreter module 123 can adopt the patient condition definition to assign the probability of the patient being classified into each of the four possible patient conditions at a particular point in time. Non-limiting examples are shown for In this example, the clinical trajectory interpreter module 123 receives the joint probability density function of P (cardiac output (Tk), pulmonary vascular resistance (Tk)), integrates it over the area for each particular condition, (S1 (Tk)), P (S2 (Tk)), P (S3 (Tk)), P (S4 (Tk)) are generated. In this manner, the clinical trajectory interpreter module 123 assigns the probability that a particular patient condition will progress, given the information provided by the physiology observation module 122. Note that if the output of the physiological function observation module 122 is not the closed-form function 260 but the histogram 280 of the particle 270, the clinical interpreter does not perform integration and only calculates the individual proportions of the particle 270 in each region I do.

FIG. 11 depicts an alternative approach to estimating the probabilities for various patient conditions. According to this alternative approach, in order to calculate the probabilities P (S1), P (S2), P (S3) and P (S4), the clinical trajectory interpreter module 123 determines two successive time windows T. The moving average of the time window is calculated using the joint probability function of ISV for _k and T _{k + 1} . Note that in this example, the size of the time window is doubled for two time instances, ie the window can be of any suitable size. As a result of performing moving averaging of time windows in this manner, the clinical interpreter module 123 performs dynamic analysis of trajectories in the ISV. That is, according to this, it is possible to measure the probability that the physiological function trajectory described by the ISV exists in a specific area in a specific time frame. In other words, such probability calculations provide an estimate of the probability that a particular patient condition will progress within the selected time frame, not exactly at the selected time instance.

  The clinical trajectory interpreter module 123 also assigns hazard levels to each particular condition. FIG. 12 depicts a non-limiting example of the definition of patient status to which hazard levels have been assigned by the clinical trajectory interpreter module 123. Hazard levels can be obtained from clinician testing, references, or other clinical sources. In a special embodiment, the clinical trajectory interpreter module 123 classifies four different hazard levels: 1-minimum risk, 2- mild risk, 3- moderate risk, 4- severe risk. The combination of the probability of a given patient condition and its hazard level is hereinafter referred to as "patient risk".

  FIG. 13 shows how patient states and their respective probabilities can be incorporated into three graphs called pathogenesis. In particular, the attributes "normal" and "low" assigned to the cardiac output ISV are the base nodes of the graph. Each of these vertices has two child elements assigned to the attribute of pulmonary vascular resistance. According to this organizational structure, the patient state is a leaf (and apex) in the tree. We will call this special tree the pathogenesis tree. The visualization and user interaction module 124 may use this etiology tree, which generates hierarchical views of various patient risks, as described below.

  FIG. 14 shows that the pathogenesis tree need not be just one for a given set of patient states and physiological function variables. In particular, FIG. 14 shows an alternative pathogenesis tree for the example of FIG. The root of this alternative pathogenesis tree starts with attributes assigned to pulmonary vascular resistance rather than cardiac output. In this case, it is preferable that different rules may be used to generate the tree depending on various factors and the use situation. For example, one etiologic tree may be chosen, in preparation for different clinical situations or different realizations according to user preferences. Moreover, the tree can be changed dynamically as the risk changes and as the clinical situation develops.

Efficacy of Various Measures During hospitalization treatment, there are measures that may harm the patient or delay patient recovery. An example of this type of dangerous measurement is all measurements derived from catheters, such as invasive blood pressure and blood oxygen saturation, which have been found to significantly increase the risk of infection. Thus, it may be beneficial to provide the clinician with an assessment of the efficacy of each potentially damaging measurement during the treatment process. FIG. 15 shows a method for calculating the effectiveness of the various measurements.

Referring to FIG. 15, the risk based system 100 and the clinical trajectory interpreter module 123 can calculate the effectiveness of a particular measurement according to the illustrated procedure. In particular, in step 9001 a measurement m _i can be selected. Given the measured value m _i and the current time point (t _current ), in step 9002 the clinical trajectory interpreter module 123 issues an instruction to the physiological function observation module 122 and outputs the output of the physiological function observation module (probability of internal state variable the density function), from any given point in time prior to the current time point (t _current-T) until the present time t _current, with the exception of the measurement m _i from the output of the algorithm can be simulated. The simulated output of the physiological function observation module is generated and the patient state probability set, ie P _sim (S ₁ (t _current )), P _sim (S 2 (t _current )), ..., P _sim (S _{When n} (t _current ) comes, then the clinical trajectory interpreter module 123 can simulate an estimate of the state probability in step 9003. Then, in step 9004, the state probabilities P (S ₁ (t _current )), P (S ₂ (t _current )),..., P (S _n (t _n ) determined from all available measurements. _{Using current} )), the clinical trajectory interpreter module 123 can calculate the validity of the measurement m _{i according to} :

これは、利用可能なすべての測定値が存在するときの患者状態分布と、測定値ｍ_iがタイムインターバルＴにわたり除かれたときの患者状態分布との間の、カルバック・ライブラー・ダイバージェンスでもある。

This is also the Kalbach-Leibler divergence between the patient state distribution when there are all available measurements and the patient state distribution when the measurements m _i have been removed over the time interval T. .

In step 9005 As another alternative, the clinical trajectory interpreter module 123 provides that the use of hazard levels r _i assigned to the patient state S _i, can be calculated effectiveness against m _i according to the following equation.

  In the same manner, the clinical trajectory interpreter module 123 can calculate efficacy calculations not only for one particular measurement but also for any group of measurements. The efficacy calculation can also include components that capture potential hazards associated with a particular measurement. For example, the above-mentioned invasive catheter measurements have a high assigned hazard level. In this way, by calculation, the hazards assigned to the measurement are subject to the value of the information that supplies the measurement. An example of such a modified effectiveness calculation is:

ただしＨ（ｍ_i）は、利用可能な測定各々の害を表す関数を規定するものである。

Where H (m _i ) defines a function that represents the harm of each of the available measurements.

The risk based monitoring system 100 may also incorporate external calculations generated from third party algorithms implemented on the same computing medium as the patient based monitoring system or implemented as part of an external device. FIG. 16 illustrates one possible approach for incorporating external calculations generated from third party algorithms. In particular, the output from external calculation 9110 is provided to the clinical trajectory interpreter module 123, which executes the embedded instructions 9120. As a result, the state probability estimation 801 generates new states P (newS ₁ ), P (newS ₂ ), P (newS ₃ ),..., P (newS _{n + m} ), whereby _n states An increased number of states n + m can be generated from the original number of states. Similarly, embedded instructions 9120 can be provided to hazard level assignments 802 and pathogenesis 803.

An example of the embedded instruction is shown in FIG. In this example, information about the specific binary attribute A = a ₁ or A = a ₂ and details of how the supplied information is captured by the conditional probability P (EC | A) in the embedded instruction and Shall be supplied by the external calculation EC. Also, even if there are _four original states S ₁ , S ₂ , S ₃ and S ₄ , the state S ₃ and S ₄ are added with two additional attributes A = a ₁ and A = a ₂ by the built-in instruction. It can be specified how it can be updated and how it can be changed to the four new states newS ₃ , newS ₄ , newS ₅ and newS ₆ . To perform this update, the embedded instruction can also use the previous probabilities P (A | S ₃ ) and P (A | S ₄ ). These previous probabilities may be derived based on a retrospective survey by analyzing what proportion of patients with status S ₃ or S ₄ simultaneously shake A = a ₁ or A = a ₂ it can.

  Another way to derive previous probabilities is to solicit the clinician's opinion.

  By using built-in instructions, a state probability estimate 801 of a new state can be derived from

ただしｉは｛３，４｝にあり、ｊは｛１，２｝にあり、Ｐ（Ｓ_j）は、生理機能観測モジュール１２２の出力から導出された本来の患者状態確率である。

However, i is in {3, 4}, j is in {1, 2}, and P (S _j ) is the original patient state probability derived from the output of the physiological function observation module 122.

  FIG. 18 shows additional examples of external calculation embedded instructions. Again, the risk-based monitoring system 100 can be used in FIG. 18 in both cases where the external calculation 9110 is generated on the same calculation medium as the patient-based monitoring system or as part of the external device. Perform the embedding as shown. In this case, it is assumed that the external calculation 9110 supplies direct information on a specific internal state variable estimated by the physiological function observation module 122 (or enhanced physiological function observation module 9300). That's what it means. Thus, in order to incorporate external calculations, the physiological function observation module 122 can treat the external calculations 9110 as additional measurements, which can be incorporated directly into the observation model 221.

Visualization and user interaction [DB2]
An example of the functionality of the visualization and user interaction module 124 is shown in FIG. In particular, module 124 can receive all available patient information and data, including data from data receiving module 121, joint probability density functions generated by physiology observation module 122, clinical trajectories. The pathogenesis tree estimated by the interpreter module 123, the risk as well as the effectiveness of the invasive measurement are included. Using this information, the visualization and user interaction module 124 can generate:
1) Patient unit view 1501 describing patient risk, diagnosis etc.
2) View 1502 of the patient's electronic medical record including test results, prescribed medications, diagnoses etc .;
3) View 1503 showing the patient's ongoing risk;
4) patient risk trajectory, ie a view 1504 showing how the probability of a particular patient condition has progressed within a certain time frame;
5) View 1505 showing the effectiveness of the patient's measurement in a particular patient risk estimate;
6) Plot 1506 of the PDF against the estimated ISV of the patient, representing the temporal evolution of the PDF against ISV;
7) to enable navigation within the patient's pathogenesis tree, ie to visualize different levels of the tree 1507;
8) View 1508 showing the predicted risk of the patient;
9) View 1509 to allow clinicians to view and set alarms based on patient risk;
10) Physiological monitoring data of the patient and a view 1510 showing its temporal development;
11) Any combination of the above.
In addition to these, the visualization and user interaction module 124 can also generate patient tag views 1511, patient condition summaries 1512, reference and training materials 1513, notes and tag settings 1514.

  FIG. 20 shows an example of a summary view 2000 which can transmit the risk profile for each patient in a specific hospital unit on a single screen. The risk profile describes what the cumulative probability of a patient at a particular hazard level is. This is calculated by summing the current probabilities of all states at a particular hazard level. For example, the summed probability, the hazard level is represented by the height of four bars, each bar corresponding to a certain hazard level. Hazard levels for this particular example are: green (diagonal hatching)-minimal risk, yellow (vertical hatching)-mild risk, orange (horizontal hatching)-moderate risk, red ( Hatching by dots)-can be a serious risk.

  FIG. 21 illustrates one possible approach for a view 2100 representing the patient's ongoing risk. Each rounded box corresponds to one particular risk. Color correspondences are: hazard level of green (hatched hatching)-minimum risk, yellow (vertical hatching)-mild risk, orange (vertical hatching)-moderate risk, red (hatched hatching)-heavy And the height of the box corresponds to the probability of a particular patient condition. Risks are grouped based on their hazard levels. This screen as well as the individual risks are updated in real time as new data become available.

  Still referring to FIG. 21, in addition to ongoing patient risk visualizations, the system 100 can also provide information regarding the effectiveness of various invasive measurements in these risk determinations. In particular, according to the illustrated example, the effectiveness of the measurement of invasive arterial pressure (ABP) and invasive central venous pressure (CVP) is provided. The effectiveness can be represented by the fills of bars 2110 and 2120, and the maximum effectiveness can correspond to six filled bars. The six bars to be filled can be displayed with color gradation, 1-dark green, 2-light green, 3-yellow, 4-red, 5-purple, 6-white or no fill, It can be done. For this particular embodiment, the bars 2110 filled for ABP have all six colors, while the two bars for bars 2120 filled for CVP are each 1-dark Green, 2-light green, remaining fill bars 2110 are 6-white or no fill.

  A view 2200 is depicted in FIG. 22 and shows how sliders 2210 or other graphical elements at the top of the patient view can be used to view the history of patient risks. In particular, according to this embodiment, the slider 2210 has been moved to indicate a patient risk about four hours prior to the current time. This allows the clinician to view the continual evolution of patient risk and compare those patient risks with the applied treatment or other external sources. In various embodiments, the slider 2210 can be moved using a pointing device or command in the user interface to specify a desired time zone, or if used with a touch panel display, the slider or other graphic. It can be moved by touching and dragging the element.

  Referring now to both FIG. 21 and FIG. 22, where the pathogenesis tree is used, two states according to FIG. 21, state A2130: hypoxia with low cardiac output, state B 2140: low Qp: Combined with hypoxia with Qs, they are represented by a single patient condition (hypoxia) 2220. According to this particular embodiment, this is used to fit the text in the box of FIG. 22, which is smaller than FIG. As depicted in FIG. 23, the user may navigate the pathogenesis tree in view 2300 by clicking on the composed patient state 2220 to display the constituent states 2130 and 2140. it can.

  In FIG. 24, by moving the slider 2410 to move ahead of the current point in time, it is possible to see the predicted risks for the patient, which can be viewed by the user within the same framework. [DB3] depicted in view 2400.

  In FIG. 25, an interactive dialog box 2510 is depicted in view 2500, through which the user can define conditions for setting an alarm for a particular risk. To this end, the user selects a particular risk and sets the upper and lower thresholds for the patient state probability assigned to that risk. As long as the patient state probability is between the upper threshold and the threshold, the alarm is not triggered. An alarm is triggered when the patient condition exceeds the threshold. When an alarm is triggered, system 100 notifies a selected set of people or sends a notification to other clinical systems. Any of the modules 122-124 can store their respective threshold data ranges each time, and can trigger based on particular parameters.

  In FIG. 26, a view 2600 illustrates the patient risk trajectory, or yet another possibility to visualize the progression of patient condition probabilities associated with a particular risk. In this case, the user can select which time series of patient condition probabilities to display, and the system plots those probabilities on the time axis.

FIG. 27 shows how the system 100 can directly represent the probability density functions of various internal state variables in view 2700. In particular, in this example, the estimated PDF of oxygen delivery is plotted in graph 2710 as a function of time, where the likelihood increases as the color becomes darker. Similarly, the estimated PDF of mixed venous blood oxygen saturation (SvO ₂ ) is plotted in graph 2720 while being compared to the actual measurements (dark circles).

Information Sharing Between Users FIG. 28 shows an example of a definition interface 2810 for tagging in a view 2800. This interface allows the clinician to mark or tag particular time instances or particular time periods 2820 that represent points of interest or points of importance in the clinical course. Through dialog box 2830, tags can be shared, or can be sent to designated recipients, or tags can be included in a note or any other part of the user interface. The user can tag unique comments or observations via the dialog box 2840 and can further categorize the tags based on the menu list 2850, for example by tag changes in medication dosage, intervention, It can represent notes on monitoring or measurement of equipment. Tags and their respective time series markings can be color coded to represent various attributes such as their categories. For example, a green tag mark over time can represent a change in medication, a red tag mark can represent an intervention, and a yellow tag mark can represent a growing cycle of concern. When the tag is set, the user can be prompted to define the time instance or period, the category of the tag, the annotation of the tag, and how the system should handle the tag. Furthermore, annotations can also be suggested by using natural language processing that converts the etiology of the condition into annotated form.

  In FIG. 29, a news feed view 2910 is depicted in view 2900, which includes tags, notes, or information obtained from an external source, for example, obtained from an electronic medical record (EMR). Blood collection time etc. are included. The news feed 2910 allows the clinician to view and fill in events, periods of interest, interventions, notes, tags etc written by other clinicians. The clinician may then browse through all the news feeds, or sort them based on tag categories, hazard levels, etc. In addition, the clinician can search for tags by keywords, intervention types, length of stay, sources of information, etc. Entries 2920-2926 in news feed 2010 allow any of the categories, sources of tags and overview of the patient to be displayed verbally or in an image such as a risk profile.

In FIG. 30, a condition summary view 3010 is depicted in view 3000 that allows a clinician to request a condition summary by selecting or clicking on a particular patient condition. In this case, the condition summary view 3010 can present a description of the patient's condition to the clinician, where there is a definition of the condition in the window 3020 and the system how to conclude on the probability of the patient's condition. It contains both the information of what has been reached. As shown, this view 3010 can provide patient state likelihood and hazard levels, definition of patient state by ISV thresholds, likelihood for each attribute defining the state, and further, A natural language description and evidence window 3030 can be provided for the evidence contributed, by converting the ISV's PDF into a quantitative textual description, or by revealing the numerical information associated with that evidence. To be done. As an example, FIG. 30 depicts a condition summary view 3010 for a patient condition “shock” due to low cardiac output. In this case, the condition summary view 3010 indicates the possibility of a “shock due to low cardiac output” condition and the hazard level indicated by hatching with color or dots. In addition, the definition of shock (mixed venous oxygen saturation ≦ 45%), and the low cardiac output (cardiac output <3.2 L / min / m ² ), the probability that each of these will be met (eg for shock) It is shown with 40% and 30% for low CO). Evidence window 3030 conveys information leading to an assessment of “shock due to low cardiac output”. According to this embodiment, the system converts the information on the probability of the etiology into a textual form, in particular in this case the pulse pressure below the nominal value mainly representing a reduction in stroke volume (probable probability) It is determined by the presence of systolic blood pressure-diastolic blood pressure).

  FIG. 31 illustrates the ability of the user interface to capture and display reference material accessible in the view 3100, accessible via the Internet, or storable in the system 100, or remotely accessible. By selecting the "detail" button 3110 in the condition summary view 3010, the detail view 3120 can be displayed, and this view shows reference information associated with the condition summary view 3010. As reference information, causes, interventions, general complications, anatomical structures, related documents, etc. can be mentioned. Furthermore, this feature can be used as a training tool to familiarize the clinician with the management or treatment strategies of a particular patient population.

Example of HLHS Stage 1 Next, it will be described how the system 100 according to the invention and the techniques can be applied to modeling the clinical course of a particular patient population. In this case, intensive care, that is, post-operative recovery after temporary relaxation of stage 1 of patients with hypoplastic left heart syndrome is performed.

Hypoplastic left heart syndrome is a congenital heart disease, which is manifested by poorly developed left ventricle and left atrium. As a result, patients suffering from such conditions do not have separate systemic and pulmonary blood flow, instead the right ventricle plays a role in pumping blood to both the whole body and the lungs. There is. Therefore, in order to optimize hemodynamics during intensive care, it is necessary to control the blood flow through the lungs (pulmonary blood flow Q _p ) and the ratio of blood flow throughout the body (body blood flow Q _s ). Optimal hemodynamic status is obtained when the ratio of lung blood flow to body blood flow represented by Q _p / Q _s is 1, an adequate oxygen transport amount DO ₂ to the tissue is achieved That's the case. By the time such optimum condition is reached, the patient's physiology often goes through other less beneficial conditions, by proper identification of those conditions, as well as by the application of appropriate therapeutic strategies for each of those conditions. , Determine the quality of post-operative care.

  Table 1 lists state variables, variable descriptions, units, and types of variables used in the model of HLHS physiological function after temporary relaxation of stage 1. As will be apparent to those skilled in the art, these variables include circulatory, hemodynamic and oxygen exchange components in HLHS physiology, but additional changes such as respiration and metabolism may be made without changing the premise of the present invention. The model can be modified or expanded by various physiological functional components.

  FIG. 32 illustrates a general purpose Dynamic Bayesian Network (DBN) that can be employed to capture a physiological model of a patient in HLHS Stage 1 transient relief. The graphical model depicted schematically in FIG. 32 captures causality and probability relationships between each variable of the model. According to this DBN, state variables are respectively incorporated into three groups: dynamic variables 3210, derived variables 3220, and observed variables 3230. The dynamic variable 3210 is a variable having a value that changes over time based on a dynamic probability model described below. The derived variable 3230 is an amount dependent on the dynamic variable with some functional relationship. These variables are calculated or derived from the latest dynamic variables, if required, and are therefore naturally dynamic. The observed variable 3230 is a variable directly measured by the system and one of the sensors connected with the patient. An observation variable 3230 represents an actual dynamic state variable observed while receiving noise or an entity of a derived state variable.

  A plurality of equations that can be used to model the dynamics of physiological functions in HLHS stage 1 are listed in FIG. This model consists of four major probabilistic model types. The first model type is a probabilistic feedback control model (Equations 1, 2, 7). These variables have reference values that maintain the body, but are disturbed by some random process to deviate from this reference value. The strength with which the body tries to maintain this variable is determined by the feedback constant. The second model type is the drift diffusion process (Equations 3, 4). These variables are moved over time by random white noise and drift rate processes. The third model type is a simple random walk process (Equations 5, 6, 8). The last dynamic model type is a memoryless process model, where the variables are not related to the variables in the preceding period, but follow some defined distribution that extends over the valid supporting points of the variables, ie the parameters Along with the gamma distribution over the whole positive real straight line with A and B, it is simply a random variable with values changing for each time instance (eq. 9, 10, 11). With the exception of equations 9, 10 and 11, the operating noise of each dynamic model is an independent white Gaussian noise.

  FIG. 34 shows equations that can be used to extract the relationship between dynamic variables and derived variables in modeling. While some of these functional relationships are true for general human physiology, many are the result of the parallel circulatory physiology that is unique to the HLHS population. Equations 12-15 represent the relationship for the directly measured variables. Equations 16-18 represent the functional relationships of the most important variables in managing post-surgery care of HLHS patients, in particular the functional relationship of cardiac output (CO) and lung body blood flow ratio (Qp: Qs). Represent. These variables can not be measured without complex procedures.

Given these functional relationships and definitions of dynamic states, FIG. 35 depicts one possibility of an observation model that can be used to correlate derived variables with available sensor data. Each observation model is conditional Gaussian relation. The measurements received from the sensors in this model represent direct observations of hidden state variables corrupted by additional independent Gaussian white noise with some variance. In the drawing, the quantity observed as a hidden state variable is described by adding a dashed line above the variable name. According to this embodiment, different sensors can be mapped to the same hidden state variable, although they may possibly have different noise levels. For example, SpO ₂ supplied by pulse oximetry measurement represents the potential hidden physiological function state, SaO ₂ , or non-invasively represents arterial blood oxygen saturation. A venous catheter inserted directly into the arterial blood flow also measures this amount, which is an invasive procedure. Catheter measurements should be more accurate than pulse oximeters. In this model, this is handled by the smaller measurement variance R.

  In HLHS physiology observation, inference through DBN is performed using a particle filter. As already mentioned, particle filters are an example of an approximate inference scheme that uses Monte Carlo sampling of internal state variables, whereby the probability density function of each state variable is approximated using an empirical distribution based on the number of particles. Ru. This filter utilizes a process known as Sequential Importance Sampling SIS to continuously resample particles from the most recent approximate probability distribution. In this filter, each particle is assigned a weight. When a new observation or measurement arrives, the weight of each particle is updated based on the individual particle likelihood in observation. The particles are then resampled based on their relative updated weights, where the particle with the highest weight is more likely to be resampled than the particle with lower weight.

FIG. 36 depicts in the HLHS Stage 1 population, possible attributes, patient states, and a tree of etiologies that can be utilized by the clinical trajectory interpreter module 123. The variable total cardiac output, defined as the sum of body blood flow and pulmonary blood flow, is used to define "low" and "normal" total cardiac output, and the Qp: Qs ratio is Qp The hemoglobin concentration Hgb is used to derive the “low”, “equilibrium”, “high” of the Qs ratio, and the value of the hemoglobin concentration Hgb is used to derive the “low” and “normal” of the hemoglobin, and venous blood mixing The value of oxygen saturation SvO ₂ is used to derive hemodynamic "with shock" and "without shock" attributes. This results in the eight possible states defined below:
1) Shock due to total cardiac output "low" when both the attribute "shocked" and total cardiac output "low" are present;
2) shock due to hemoglobin "low" when in the condition with the attribute "shocked", total cardiac output "normal", hemoglobin "low";
3) Attributes "shocked", total cardiac output "normal", hemoglobin "normal", circulation "balanced", shock due to unknown cause;
4) Attributes “shocked”, total stroke volume “normal”, hemoglobin “normal”, Qp: Qs “low”, Qp: shock due to Qs “low”;
5) Attributes “shocked”, total stroke volume “normal”, hemoglobin “normal”, Qp: Qs “high”, Qp: shock due to Qs “high”;
6) attribute "no shock", circulation "normal", circulation "normal";
7) Attributes "no shock", Qp: Qs: low as defined by Qs: low, Qp: low;
8) Attributes "no shock", Qp: Qs: high as defined by Qs "high".
Also shown in FIG. 35 is a possible approach to the pathogenesis tree that represents the relationship between attributes and patient status. The probability of eight states can be calculated by calculating the relative proportions of particles within each state using the particle approximation of the internal state variables.

The risk-based monitoring system associated with the assessment of the outcome of the contingency treatment Other applications of the risk-based monitoring system are used to determine whether to apply a particular treatment, for example transfusion as an example. , To help clinicians. Transfusion of blood as well as blood products is generally an in-hospital procedure. Nevertheless, transfusion indicators and policies are not well established and consistently applied among hospitals. Many studies have shown that there is a diversity of transfusion practices in different hospitals, practitioners, and procedures. Such diversity exists even when applied to a single treatment (eg, coronary artery bypass grafting).

  In addition, transfusions are increasingly recognized as an independent risk factor for morbidity and mortality. Specific events and outcomes associated with blood transfusion include sepsis, organ ischemia, increased respiratory support time, increased hospitalization, and short and long term morbidity. This relationship is proportional to blood transfusion volume and what is suggested as a sign is that high hematocrit values can be harmful. Of course, researchers routinely recommend transfusion policies to achieve information-based risk-benefit trade-offs.

  Establishing a robust and effective blood transfusion policy proved to be a difficult task. It has been agreed among healthcare professionals that simple policies such as guidelines based on the hemoglobin threshold do not provide adequate indicators. This is due to the compensating property of hemodynamic physiology, with the patient having variable receptivity to tolerate low hemoglobin. Therefore, factors such as compensatory reserve, intravascular volume, hemodynamic stability, treatment type, as well as other patient data need to be integrated to make an effective transfusion decision. Thus, there is an integral requirement for blood management policies, which make use of an entire series of relevant clinical variables, which determine the risk-benefit ratio of transfusion. This is exactly achieved by applying a risk based monitoring system.

  FIG. 37 illustrates one possible environment in which a risk based monitoring system can be employed to assist the clinician in determining whether a particular treatment should be applied. According to the invention, the patient 101 is monitored intermittently and continuously by a plurality of measuring devices 3910. As continuous measurements, mention may be made of mixed venous oxygen saturation (SvO2) 3911, systolic blood pressure, diastolic blood pressure and mean blood pressure (ABP s | d | m) 3913, heart rate 3916, Is monitored by a bedside monitor. Intermittent measurements include blood pH 3915, hemoglobin concentration (Hgb) 3912, and blood lactate concentration 3914, which are monitored by periodic blood tests. These measurements 3910 are provided to a risk based monitoring system 3940 extended by treatment assessment. This system is configured by a plurality of other modules in addition to the physiological function observation module 122 and the clinical trajectory interpreter module 123 described above. A prospective treatment complication determination module 3942 receives information from the clinical trajectory interpreter module 123 along with information regarding patient demographics 3931 and treatment types 3932. The module 3942 uses this information to query the outcome database 3943 and information is returned to module 3942 that the following a) to d) have the potential for various complications: a) Patient Is in a particular patient condition with a particular probability; b) the patient belongs to a particular demographic (age, gender etc); c) the patient is undergoing a particular type of treatment; d) above a) Some combination of b) and c). On the other hand, using retrospective survey 3991, randomized clinical trial 3992, institution-specific outcomes 3993 determined based on patient data collected in advance for a particular institution, and outcome survey 3990 derived from some combination of these requirements , An outcome database 3943 can be constructed.

  If the potential complication is determined by the contingent treatment complication determination module 3942, the module 3942 feeds this information back to the enhanced visualization and user interaction module 3941. The enhanced visualization and user interaction module 3941 combines patient-specific risk-based monitoring implemented by the physiology observation module 122 and the clinical trajectory interpreter module 123 with the assessment of possible complications. This provides the system with an excellent perspective, which allows the clinician 3920 to better recognize the risks and benefits of treatment, such as blood transfusion, and whether or not this treatment 3960 should be given. It can be judged efficiently and effectively.

A non-limiting example is shown in FIG. 38 for a set of patient conditions associated with blood transfusion that can be used to notify the determination of blood transfusion. These conditions include the dynamics of hemoglobin (decrease / stability / increase) and hemodynamic compensation when blood oxygen delivery capacity is reduced. In the case of noncompensated patients, blood flow autoregulation mechanisms can not overcome the decline in blood oxygen carrying capacity, a sign being the onset of anaerobic metabolism. These seven states can be determined by three internal state variables: oxygen delivery and hemoglobin and hemoglobin production / loss rate. Specifically, if hemoglobin is higher than 13 mg / dL, it is considered that there is no Hgb related medical condition (4001). If Hgb is lower than 13 mg / dL, there will be six remaining states, which are defined by five different attributes. From the oxygen delivery ISV the system can determine if the patient is in a compensated state, for example where it can be assumed that for a ventilated patient being ventilated, 400 ml / min. A DO2 higher than / m ² represents a compensatory state, and lower than this value represents a non-compensated patient. The other three attributes are determined from the Hgb rate ISV and are stable 4013 and 4014 (rate nearly zero), increments 4011 and 4012 (rate positive), and decrements 4015 and 4016 (rate negative).

Use of a Risk Based Monitoring System with a Standardized Clinical Plan Yet another use of a risk based monitoring system is the case with a standardized medical plan. One possible embodiment for this application is shown in FIG. Specifically, data from the clinical trajectory interpreter module 123 is provided to the treatment query module 4142. The treatment query module 4142 queries the treatment plan database 4143 based on the determined patient risk. The treatment plan database 4143 specifies the mapping between patient risk and data. Once the database 4143 feeds back any treatment plan, the advanced visualization and user interaction module 4141 presents the plan to the clinician. The clinician 3920 can make clinical decisions 4190 for the patient 101. The determination of the user, the circumstances under which the determination was made (calculated patient risk, estimated ISV, and other patient data obtained at the time of judgment) are then recorded in the determination database 4144. The decision database 4144 can then be compared to the patient outcome and used to improve the treatment plan.

  One application case is illustrated in FIG. 40 for a risk based monitoring system combined with a specific type of standardized clinical plan. In this particular example, the medical decision whether to treat the patient with nitric oxide is discussed. Nitric oxide is a pulmonary vasodilator and is used to treat high pulmonary vascular resistance and consequently pulmonary hypertension which may reduce cardiac output. In this example, the medical plan uses the risks calculated by the clinical trajectory interpreter module 123 and stratifies them into two categories, low risk and high risk (4230). If the risk is low (4201), the recommended decision is not a treatment (4202), and if the patient is classified as high risk, the recommended decision is a treatment (4203). The healthcare provider can decide whether to follow the recommendation (4250) or ignore the recommendation (4260). For high-risk patients, if the provider chooses to ignore the treatment recommendation, the provider needs to justify it (4220). Similarly, if the provider chooses a treatment for low-risk patients, the provider also needs to justify it (4210). Justifications 4210 and 4220 coupled with patient outcomes can be utilized to further refine risk stratification 4230 and risk-based monitoring system 100.

  FIG. 41 shows an example of risk stratification that can be employed by the system in the context of nitric oxide therapy. Specifically, the patient may take four different states: State 1: CO low, PVR normal; State 2: CO low, PVR high; State 3: CO normal, PVR high; State 4 : CO normal, PVR normal. Patients at low risk can be defined as P (State 1) <10% and P (State 1) + P (State 2) <30%. Similarly, patients at high risk can be defined as P (State 1)> 10% and P (State 1) + P (State 2)> 30%.

Use of a Clinical Risk Assessment System in Outpatient Care in Chronic Conditions According to yet another embodiment of the present invention, the care of an outpatient can be tracked by a clinical trajectory. Being outpatient care for chronic conditions involves sporadic patient assessment based on intermittent visits, patient self-assessment, and caregiver observation. This results in uncertainty in the determination of the patient's clinical course as well as the efficiency of the described treatment strategy. In order to achieve effective patient care, clinicians need to understand such uncertainty and have to reduce them. The clinician has two major decisions at its discretion. That is, 1) scheduling of visits, instruction for examination, request for self-evaluation (or evaluation of carers), and / or 2) further improvement of symptoms and possible side effects to improve clinician's understanding of clinical trajectories Formulation of drug or drug dosing changes to achieve a good trade-off. In order to be notified of this decision process, it is necessary to process the available patient information, in which the uncertainty of the clinical trajectory and its estimate, and the expected effect of different treatment strategies on the future development of the clinical trajectory are communicated Be done.

  As a non-limiting example of a risk-based monitoring system for clinical trajectory tracking of outpatients, here is the case where the system is applied to outpatient care of attention deficit / hyperactivity disorder (ADHD) in pediatric patients. I will consider it. FIG. 42 illustrates one possible patient condition that can describe the clinical trajectory of ADHD patients. These patient states are the same as used by the Clinical global impression-improvement scale. 1) very markedly worse; 2) significantly worse; 3) worse; 4) unchanged; 5) slightly improved; 6) significantly improved; 7) extremely significantly improved. The patient state distribution (PSD) is a collection of probabilities of which of these seven states the patient is in when all available information and observations are obtained.

  For assessment of patient status, clinicians can schedule work visits for immediate investigation, or can request a family or teacher Vanderbilt diagnostic test (this test Change according to the respondent, teacher or parent) In FIG. 43, available patient evaluation modalities are listed as M1, M2, and M3. These models can be used to map test questions and answers to patient conditions. Both clinical and test-based assessments are accompanied by uncertainties that prevent accurate determination of which of the seven conditions the patient currently belongs to.

  As shown in FIG. 44, dynamic models or patient progress from one state to another can be extracted by a dynamic Bayesian network (DBN). According to FIG. 44, the direction of the arc-shaped arrow represents a statistical dependency, that is, the connection from “patient state @ t1” 4601 to “M1” has probability density function (PDF): P (M1 | patient It represents the state @ t1). Similarly, the DBM shown in the figure "patient status @ t2" 4602 (patient status at a certain point in time t2) depends on "patient status @ t1" 4601 (patient status at the point t1 returned before). Represents that. According to the idea of the invention, this model is able to estimate the patient's condition even if some or all of the possible measurements are not obtained, ie M1 is missing as shown. At time t2 as well as at time t3 where all measurements are missing ("patient state @ t3" 4603), it is possible to estimate the patient state.

FIG. 45 shows an alternative embodiment for the two predictions of how patient status may shift in a single month when changing medication or dosage . This prediction is performed based on a statistical model derived by the following method:
Step 1: Separate from the retrospective data the group of patients who have passed Condition A at any time of treatment and have received a change in treatment (Drug 1 Metered 1 → Drug 2 Metered 2);
Step 2: For each patient, set the time at which this particular event occurred to t0;
3) Step 3: Identify, for each patient, what the patient's condition at time t0 + 1 months (1 M) (or any desired time step unit) is;
4) Calculate the proportion of patients transitioning from state A to state i, where i represents all seven possible patient states;
5) Set rates as transition probabilities under specific treatment changes.

  FIG. 46 illustrates one possible embodiment of visualization and a scenario that displays the patient's clinical trajectory and risk. This user interface represents that the patient is in the "invariant" state, which has been established by three separate measurements: work visit, teacher-based Vanderbilt diagnosis, and parent-based Vanderbilt diagnosis. There is. The solid line on the screen represents that the patient was prescribed a drug during the first week of treatment (drug 1).

  FIG. 47 shows the evaluation of the patient and patient trajectory at week 9. At this point, the clinical risk assessment system determines the probability density function for the patient's condition for each of the last six weeks. The measurements available at this point are Vanderbilt's diagnoses of teachers and parents. In the case of the illustrated embodiment, the clinician indicates a change in drug metering due to the high probability of patient condition deterioration, which is represented by the dashed red line. In addition to this, the user interface also shows the reported side effects of a patient's headache.

  FIG. 48 shows the follow-up evaluation based on the Vanderbilt diagnosis of the teacher and the parent. According to this embodiment, the physician determines the drug change indicated by the open line.

  FIG. 49 shows the resulting assessment, based on all available measurements that are likely to cause significant improvement, ie, the visit to the work site and the parental and teacher assessment.

  FIG. 50 shows that the patient is steadily improved through nine minutes, and shows another follow-up when stable improvement is observed between the two evaluations. There is. Here, the PDF for patient trajectories is continuously estimated based on the applied inference. However, the accuracy (density of the PDF) is even higher if the measurement is present.

  FIG. 51 shows the follow-up evaluation of patients and patient trajectories in the absence of measurements. In this case, uncertainty is rising because recent observations are lacking.

  FIG. 52 shows the state of the above uncertainty when a complete patient assessment (all measurement modalities) has been performed. The estimation engine backpropagates this uncertainty with respect to time to generate a more accurate patient trajectory estimate, which assists the physician in deducing the patient to be stable.

  FIG. 53 shows the possibility of further visualization from the previously described system output. This figure shows the transition of patient states that can occur when the treatment plan is changed, for example, when the drug is changed. It also allows one to anticipate what side effects may occur. There are three expression stages for all side effects: mild, moderate and severe, which are represented in the figure by the three boxes drawn beside the side effects respectively. Each is colored corresponding to the probability of specific expression for a specific side effect, and darker color represents higher probability.

  Various embodiments and embodiments in accordance with the present invention have been described above in detail. It will be appreciated, however, that these examples and embodiments of the present invention are listed for illustrative purposes only, and those skilled in the art will deviate from the disclosed scope of the present invention as defined in the appended claims. Instead, various improvements and modifications can be made to the embodiments disclosed herein.

Claims (15)

A risk based monitoring system for converting measurement data of a severe patient including heart rate and blood oxygen content into hidden internal state data that can not be measured directly and noninvasively by a sensor, wherein the hidden internal state is Monitored by the system, the system
A processor,
A data receiving module having a set of inputs for the plurality of sensors including at least one heart rate sensor and one oxygen saturation measuring sensor;
A physiological function observation module that communicates with the data reception module;
A clinical trajectory interpreter module in communication with the physiological function observation module;
Is provided,
The plurality of sensors are connectable to the severe patient, and the plurality of sensors are connected to the data receiving module in a series of time steps t. _k Corresponding multiple internal state variables m across (k = 0, 1, ... Z) _s Provides data associated with (s = 1,... N), and each internal state variable m _s Characterizes parameters that are physiologically relevant to the treatment and / or condition of said patient,
Said physiological function observation module is configured to update data provided by said sensor to said data reception module via first and second computer processes implemented by said processor;
The first computer process is time t _k Internal state variable m at _s Generate a conditional likelihood kernel for, the conditional likelihood kernel including a set of probability density functions, each probability density function having a time t _k Separate internal state variables m at _s Is about
The second computer process uses Bayesian theory to _k Internal state variable m at _s Said conditional likelihood kernel for and time t _k Assuming a predicted conditional probability density function for each of a plurality of internal state variables for _k Multiple internal state variables m for _s Generate a posteriori prediction conditional probability density function for each of
The clinical trajectory interpreter module is configured to _k Multiple internal state variables m in _s To generate a probability value representing the probability that the patient's oxygen delivery capacity can not be measured directly and non-invasively based on the generated posterior prediction conditional probability density function of Configured to determine a set of possible states of the state variable,
Risk based monitoring system. The oxygen saturation measuring sensor is SpO ₂ The system of claim 1, wherein the system is a sensor. A display device operatively connected to the processor;
With user interaction module
And have
The user interaction module is configured to display a plurality of graphical indicators on the display device, each of the plurality of graphical indicators being one of a set of possible states of the hidden internal state variable. Corresponding to one of the states, each of the plurality of graphical indicators graphically identifying the probability that the hidden internal state variable is in the corresponding state at a point in a given time range; The system of claim 1, wherein the graphical indicator is configured to indicate a hazard level. The user interaction module is further configured to display the patient's clinical trajectory on the display device based on the conditional probability density function of the hidden internal state variable over at least a portion of the series of time steps. The system of claim 3. The physiological function observation module assumes that the previously received measurement value is time t _k One internal state variable m in _s Comparing the newly received measurement value associated with the predetermined predicted likelihood of the possible measurement value, the newly received measurement value being the associated internal state variable m _s , The associated internal state variable m, if not within the predetermined prediction likelihood of the possible measurement value for _s The system of claim 1, wherein the system does not generate a conditional probability density function for. It further comprises a timeline controller, said timeline controller allowing the user to _k Dynamically select a plurality of points in time, and the graphical index dynamically according to the specification of the user of one of the points in order to display the progress of the hidden internal state variable The system of claim 3, wherein the system is modified. The physiological function observation module uses the predicted conditional probability density function generated in the preceding time step for each of the plurality of internal state variables to calculate the time t _k Internal state variable m at _s The system of claim 1, generating the conditional likelihood kernel for. The physiological function observation module performs time step t _k At the time step t _k Multiple internal state variables m for _s Time t based on said posterior prediction conditional probability density function for each of _{k + 1} Multiple internal state variables m for _s The system of claim 1, further configured to generate a predicted conditional probability density function for each of. Said time step t _{k + 1} The predicted conditional probability density function for is the preceding time step t _k The plurality of internal state variables m obtained from _s Using the above posterior conditional probability density function for each of
P (ISVs (t _{k + 1} ) | M (t _k )) = ∫ _{ISVs ∈ ISV} P (ISVs (t _{k + 1} ) | ISVs (t _k )) P (ISVs (t _k ) | M (t _k )) dISVs
The system of claim 8, wherein the system is generated using A method for converting measurement data of a severe patient including heart rate and blood oxygen content into hidden internal state data that can not be measured directly by a sensor, the method comprising
Providing a plurality of sensors including at least one heart rate sensor and one oximetry sensor, wherein the plurality of sensors are configured to be physically attachable to the critically ill patient When,
Attaching the plurality of sensors to the patient;
A series of time steps t from the plurality of sensors connected to the patient _k Corresponding multiple internal state variables m across (k = 0, 1, ... Z) _s Acquiring substantially continuously the physiological data related to (s = 1,... N) by the computer;
Time t by the computer _k Multiple internal state variables m in _s Generating a conditional likelihood kernel for at least one of said time steps t, _k Multiple internal state variables m for _s A set of probability density functions, each of said internal state variables having a time step t _k Describing a parameter physiologically relevant to at least one of the treatment and the condition of the patient in
By the computer and using Bayesian theory, time t _k Multiple internal state variables m in _s Said conditional likelihood kernel for and time step t _k Multiple internal state variables m for _s Assuming a predicted conditional probability density function for each of _k Multiple internal state variables m for _s Generating a posteriori prediction conditional probability density function for each of
Using the computer, time step t _k Multiple internal state variables m in _s From the generated posterior prediction conditional probability density function of ₂ The patient's oxygen delivery (DO) that identifies and does not directly measure the possible set of hidden internal state variables associated with ₂ Generating a probability value that represents the likelihood that
Method, including. The oxygen saturation measuring sensor is SpO ₂ 11. The method of claim 10, wherein the method is a sensor. The method may further comprise substantially continuously displaying the patient's clinical trajectory on a graphical user interface, the user interface displaying the probability of the hidden internal state variable in response to a plurality of time steps. 11. The method of claim 10, wherein the method is configured to: 11. The method of claim 10, wherein the probability value associated with the hidden internal state variable indicates the patient's oxygen delivery is indivisible and the risk that the patient may become shocked. Time t _k Multiple internal state variables m in _s 11. The method of claim 10, wherein generating the conditional likelihood kernel for n comprises using, for each of the plurality of internal state variables, a predicted conditional probability density function generated in a preceding time step. Method described. Time step t _k At the time step t _k Of the multiple internal state variables m _s Time t based on said posterior prediction probability density function for each of _{k + 1} Of the multiple internal state variables m _s 15. The method of claim 14, further comprising generating a predicted conditional probability density function for each of.

Images