WO2023045425A1 - 一种透析中低血压事件风险的评估方法和系统 - Google Patents

一种透析中低血压事件风险的评估方法和系统 Download PDF

Info

Publication number
WO2023045425A1
WO2023045425A1 PCT/CN2022/099129 CN2022099129W WO2023045425A1 WO 2023045425 A1 WO2023045425 A1 WO 2023045425A1 CN 2022099129 W CN2022099129 W CN 2022099129W WO 2023045425 A1 WO2023045425 A1 WO 2023045425A1
Authority
WO
WIPO (PCT)
Prior art keywords
rbv
data
risk
threshold
rate
Prior art date
Application number
PCT/CN2022/099129
Other languages
English (en)
French (fr)
Inventor
高光勇
魏志雄
童锦
周俊全
李绍林
朱平
李斌
Original Assignee
重庆山外山血液净化技术股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 重庆山外山血液净化技术股份有限公司 filed Critical 重庆山外山血液净化技术股份有限公司
Publication of WO2023045425A1 publication Critical patent/WO2023045425A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/02Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/021Measuring pressure in heart or blood vessels
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/02Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/026Measuring blood flow
    • AHUMAN NECESSITIES
    • A61MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61BDIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00Measuring for diagnostic purposes; Identification of persons
    • A61B5/72Signal processing specially adapted for physiological signals or for diagnostic purposes
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10Complex mathematical operations
    • G06F17/11Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10Complex mathematical operations
    • G06F17/16Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization

Definitions

  • the invention relates to the technical field of medical management, in particular to a method and system for assessing the risk of hypotension events in dialysis.
  • Hypotension during dialysis is the most common complication of dialysis treatment in patients with chronic kidney disease.
  • One of the main causes of this symptom is the imbalance between the ultrafiltration rate during dialysis and the reinfusion rate in the patient's body, resulting in rapid blood volume in the core area of the patient's body. decrease beyond the tolerance of the patient.
  • RBV relative blood volume
  • This risk value can be applied to the biofeedback closed-loop control during dialysis, which can intervene in advance to the imbalance in the patient's body and reduce the probability of hypotension events in the patient.
  • whether the absolute value, amount of change, rate of change (first derivative), and second derivative of RBV exceed one or more thresholds is usually used to judge whether the patient has a hypotensive event or the risk degree of the occurrence.
  • CN110151153A discloses a prior art solution, which collects patient RBV data every 5 minutes, performs threshold judgment on the second derivative SDRBV of RBV, and optionally combines blood pressure data to judge whether a patient will have a hypotensive event.
  • another fuzzy logic technique is adopted, using at least two fuzzy modules to receive hemodynamic parameters (such as relative blood volume and blood pressure), and by weighting the output of each fuzzy module, the final output is at least A variable used to assess a patient's risk of hypotensive events.
  • CN104346521A discloses another prior art, which uses a learning algorithm or a neural network to learn the individual dialysis parameters of the patient and then predicts the hemodynamic parameters of the patient during treatment. Device and method.
  • Non-real-time may miss the best opportunity for intervention.
  • the judgment result is a logic value, which is not conducive to subsequent accurate feedback control.
  • the purpose of the present invention is to provide a method for assessing the risk of hypotensive events in dialysis and a system for assessing the risk of hypotensive events in dialysis for all or part of the above-mentioned problems, so as to improve the use of RBV to evaluate the patient's dialysis process.
  • Hypotensive Event Risk Accuracy is to provide a method for assessing the risk of hypotensive events in dialysis and a system for assessing the risk of hypotensive events in dialysis for all or part of the above-mentioned problems, so as to improve the use of RBV to evaluate the patient's dialysis process.
  • a method for assessing the risk of hypotensive events in dialysis comprising:
  • the RBV data during the dialysis process is collected at predetermined intervals, and for each collected RBV data, the following process is performed:
  • the filtered descent rate is evaluated based on the evaluation threshold set at the current moment.
  • the evaluation threshold set at the current moment includes a first threshold and a second threshold higher than the first threshold.
  • the evaluation of the filtered rate of decline based on the evaluation threshold set at the current moment is to calculate the risk of the filtered rate of decrease based on the first threshold and the second threshold. coefficient.
  • the evaluation of the filtered descent rate based on the evaluation threshold set at the current moment includes:
  • the filtered decreasing rate is evaluated by using a linear interpolation method or a monotonically increasing curve.
  • the first threshold is the value of the first derivative of the set ideal RBV curve at the current moment
  • the second threshold is N times the first threshold
  • N is a constant greater than 1.
  • DRBV is the expected decrease in RBV
  • T is the total treatment time
  • n is the magnification factor
  • a is the attenuation parameter.
  • X k is the system state at the current moment
  • X k-1 is the system state at the previous moment
  • Z k is the system state observed at the current moment
  • A is the state transition matrix
  • H is the observation matrix
  • Q and R are respectively Indicates the noise covariance matrix of the system prediction model and sensor
  • P k is the covariance matrix of X k
  • P k-1 is the covariance matrix of X k-1
  • I is the identity matrix
  • K k is the Kalman gain matrix
  • calculation method of the state transition matrix A includes:
  • the state transition matrix is obtained.
  • calculation method of the observation matrix H includes:
  • the observed system state is represented by using the discretized calculation model and the drop rate, and an observation matrix is obtained by analogy with the standard representation of the observed system state.
  • the present invention provides a system for assessing the risk of hypotension events in dialysis, including a data collection unit, a data evaluation unit and a data output unit.
  • the data evaluation unit is pre-built with a Kalman filter, and is set to evaluate corresponding to each moment threshold;
  • the data collection unit collects RBV data during the dialysis process according to the configured collection cycle
  • the data evaluation unit uses the Kalman filter to filter the RBV data, calculates the observed decline rate of the RBV data, and uses the Kalman filter to filter the calculated decline rate; based on the current The evaluation threshold of the time is used to evaluate the filtered rate of decline;
  • the data output unit outputs the evaluation result of the data evaluation unit.
  • the present invention uses Kalman filtering to preprocess the relative blood volume data, and uses a simple algorithm to filter out the influence of interference signals, so that it can better reflect the actual relative blood volume changes of patients, and improves the use of RBV to evaluate patients. Accuracy of risk of hypotensive events during dialysis. Moreover, because the Kalman filter algorithm is simple and the processing speed is fast, it can realize real-time processing of RBV data and reduce the response time of correction measures.
  • the present invention uses dynamic thresholds to calculate the risk of hypotension events, so that it can better adapt to the changes in patients' tolerance to imbalance in different time periods.
  • the final evaluation result of the method and system of the present invention is the normalized risk coefficient obtained from the evaluation of the decline rate, and the judgment of the risk level is more refined, which is convenient for subsequent feedback control or parameter determination of corrective measures.
  • Figure 1 is a flowchart of one embodiment of the evaluation method of the present invention.
  • the method for assessing the risk of hypotensive events in dialysis includes the following process:
  • the first step is to determine the first derivative (FDRBV) curve of the ideal RBV curve of the dialysis process
  • DRBV expected RBV decrease
  • T total treatment duration
  • FDRBV i is calculated using the following formula:
  • the second step is to establish the Kalman filter equation and determine the filtering parameters
  • Kalman filtering is an algorithm for optimal estimation of the system state through the input and output observation data of the system.
  • the observation data includes the influence of noise and interference in the system, so the optimal estimation can also be regarded as a filtering process.
  • the work of the Kalman filter (hereinafter referred to as the filter) is based on two equations: the system state transition equation and the system observation equation.
  • A is the state transition matrix
  • X k-1 represents the system state at the last moment
  • B is the control matrix
  • u k represents the control vector
  • H is called the observation matrix
  • v k ⁇ N(0, R k ) represents the observation noise, which is Gaussian white noise with a mean value of 0 and a variance of R k .
  • the patient's RBV is constantly changing (decreasing) due to ultrafiltration, which is denoted as rbv, and the rate of decline is denoted as m.
  • rbv the rate of decline due to ultrafiltration
  • ⁇ t represents the change duration
  • rbv t- ⁇ t represents the rbv at time t- ⁇ t
  • m t- ⁇ t represents the rate of change of t- ⁇ t.
  • the value of a is 0.985 in this embodiment, which is the same as the parameter in the FDRBV i formula calculated in the first step.
  • C is a constant
  • linearFit(C) represents the slope obtained by performing linear fitting on the latest C rbv' data.
  • the value of C determines the estimation of the short-term or long-term change of rbv'.
  • short-term change needs to be estimated, so C takes a smaller value, and the specific value depends on the collection period of rbv' data. When the period is 1min, C takes 5.
  • Z k can be expressed as:
  • P k is the covariance matrix of X k
  • P k-1 is the covariance matrix of X k-1
  • Q and R respectively represent the noise covariance matrix of the system prediction model and the sensor, which need to be determined according to the actual situation
  • K k is the Kalman gain matrix, which is the amount of intermediate calculations
  • I is the identity matrix.
  • the preferred values are:
  • the third step is to collect RBV data and perform filtering
  • RBV data are collected at predetermined time intervals (eg, every minute).
  • the fourth step is to generate the hypotension risk assessment coefficient
  • the fifth step is to judge whether the treatment process is over, if so, end the evaluation, otherwise, jump to the third step, and continue to evaluate the RBV data collected next time.
  • This embodiment discloses a system for assessing the risk of hypotension events in dialysis, which includes a data acquisition unit, a data evaluation unit and a data output unit.
  • the data evaluation part is pre-built with a Kalman filter, and an evaluation threshold corresponding to each moment is set.
  • the evaluation threshold is set based on the first derivative curve (FDRBV) of the ideal RBV curve. Firstly, determine the expected RBV decrease (DRBV) for this treatment and the total treatment duration T. The default DRBV setting is 24%, and the total treatment duration T is 4 hours. Of course, the two data can be determined according to the actual situation of the patient. Then the average decline rate of RBV (EDRBV) during the whole treatment process is:
  • FDRBV i n times of EDRBV as the initial FDRBV. Taking n times of EDRBV as the initial FDRBV, the FDRBV at the i-th moment in the treatment is expressed as FDRBV i , then FDRBV i is calculated using the following formula:
  • n is the magnification factor
  • a is the attenuation parameter
  • a ⁇ 1 in this example, n is 5, and a is 0.985.
  • the evaluation threshold includes a safety line and a prohibition line.
  • the FDRBV i at the current moment is used as the safety line of the RBV slope, and it is amplified to a certain extent (for example, N ⁇ FDRBV i , N is a constant greater than 1), as the prohibition of the RBV slope Wire.
  • the pre-built Kalman filter is as follows:
  • the work of the Kalman filter (hereinafter referred to as the filter) is based on two equations: the system state transition equation and the system observation equation.
  • A is the state transition matrix
  • X k-1 represents the system state at the last moment
  • B is the control matrix
  • u k represents the control vector
  • Z k the current state of the system observed using the measurement system.
  • H is called the observation matrix
  • v k ⁇ N(0, R k ) represents the observation noise, which is Gaussian white noise with a mean value of 0 and a variance of R k .
  • the patient's RBV is constantly changing (decreasing) due to ultrafiltration, which is denoted as rbv, and the rate of decline is denoted as m.
  • rbv the rate of decline due to ultrafiltration
  • ⁇ t represents the change duration
  • rbv t- ⁇ t represents the rbv at time t- ⁇ t
  • m t- ⁇ t represents the rate of change of t- ⁇ t.
  • the value of a is 0.985 in this scheme, which is the same as the parameter in the FDRBV i formula calculated in the first step.
  • C is a constant
  • linearFit(C) represents the slope obtained by performing linear fitting on the latest C rbv' data.
  • the value of C determines the estimation of the short-term or long-term change of rbv'.
  • short-term change needs to be estimated, so C takes a smaller value, and the specific value depends on the collection period of rbv' data. When the period is 1min, C takes 5.
  • Z k can be expressed as:
  • Q and R represent the noise covariance matrix of the system prediction model and the sensor respectively, which need to be determined according to the actual situation
  • K k is the Kalman gain matrix, which is the amount of intermediate calculation
  • I is the identity matrix.
  • the preferred values are:
  • the data acquisition unit collects RBV data once at predetermined time intervals (eg, every minute), and the RBV data collected each time are transmitted to the data evaluation unit.
  • the data evaluation part takes the received RBV data as the first input of the filter; and obtains the slope (linearFit(C)) according to the correlation algorithm in the second step as the second input of the filter, and the RBV data and its The slope is filtered to remove noise, and the filtered RBV and slope are output. Of course, it is possible to filter only the slope.
  • the risk assessment is carried out according to its size relationship with the safe line and forbidden line:
  • the data output part is connected with the data evaluation part, and outputs the evaluation result (ie, the risk coefficient) of the RBV data collected by the data collection part each time by the data evaluation part.
  • the data output by the data output unit can be used as parameters for feedback control or corrective measures.
  • the present invention is not limited to the foregoing specific embodiments.
  • the present invention extends to any new feature or any new combination disclosed in this specification, and any new method or process step or any new combination disclosed.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Biomedical Technology (AREA)
  • Pathology (AREA)
  • Theoretical Computer Science (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Molecular Biology (AREA)
  • Databases & Information Systems (AREA)
  • Surgery (AREA)
  • Cardiology (AREA)
  • Veterinary Medicine (AREA)
  • Biophysics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Physiology (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Algebra (AREA)
  • External Artificial Organs (AREA)
  • Computing Systems (AREA)
  • Psychiatry (AREA)
  • Artificial Intelligence (AREA)
  • Hematology (AREA)
  • Signal Processing (AREA)
  • Operations Research (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Vascular Medicine (AREA)
  • Computer Vision & Pattern Recognition (AREA)

Abstract

本发明公开了一种透析中低血压事件风险的评估方法和系统。确定理想RBV曲线的一阶导数曲线。以预定间隔时长采集透析过程中的RBV数据,每次采集的数据使用卡尔曼滤波器进行滤波,计算RBV数据的下降率,使用卡尔曼滤波器对该下降率进行滤波。通过一阶导数曲线计算当前时刻的安全线和禁止线,根据比对下降率与安全线和禁止线的大小,确定风险系数。本发明能滤除采集的RBV数据中的干扰,提高使用RBV评估患者透析过程中低血压事件风险的准确性。同时,使用动态阈值进行低血压事件风险的计算,使其能够更好地适应不同时间段患者对失衡程度的耐受度变化。滤波算法简单,处理速度快,可以实现对RBV数据的实时处理。

Description

一种透析中低血压事件风险的评估方法和系统 技术领域
本发明涉及医疗管理技术领域,尤其是一种透析中低血压事件风险的评估方法和系统。
背景技术
透析中低血压是慢性肾脏病患者透析治疗中最常见的并发症,出现此症状的主要原因之一是透析中的超滤速率与患者体内的再注入速率失衡导致患者体内核心区域的血容量迅速减少,超出患者的耐受能力。通过分析患者在治疗中的相对血容量(RBV)变化情况,可以推断出患者体内的失衡情况,从而评估患者发生低血压事件的风险。此风险值可应用于透析过程中的生物反馈闭环控制,可以提前干预患者体内的失衡情况,降低患者出现低血压事件的几率。具体地,通常通过RBV的绝对值、变化量、变化率(一阶导数)、二阶导数是否超过某一个或多个阈值来判断患者低血压事件是否或发生者发生的风险程度。
CN110151153A公开了一种现有技术方案,其通过每5分钟采集患者RBV数据,对RBV的二阶导数SDRBV进行阈值判断,并可选地结合血压数据,从而判断患者是否会发生低血压事件。另外,在该文献中,还采用了另一种模糊逻辑技术,使用至少两个模糊模块接收血液动力学参数(例如相对血容量和血压),通过对各个模糊模块的输出进行加权,最终输出至少一个变量,用以评价患者发生低血压事件的危险程度。
CN104346521A公开了另一种现有技术,其采用利用学习算法或神经网络对患者个体透析参数进行学习后对治疗中患者的血液动力学参数进行预测的 装置和方法。
现有技术存在以下的不足:
1.RBV的噪声会很大程度干扰RBV的导数变化情况,从而降低对低血压事件判断的准确性。
2.未考虑患者在治疗中对失衡程度的耐受度变化。
3.非实时,可能错过最佳干预时机。
4.判断结果为逻辑值,不利于后续精确的反馈控制。
发明内容
本发明的发明目的在于:针对上述存在的全部或部分问题,提供一种透析中低血压事件风险的评估方法,以及一种透析中低血压事件风险的评估系统,以提高使用RBV评估患者透析过程中低血压事件风险准确性。
本发明采用的技术方案如下:
一种透析中低血压事件风险的评估方法,包括:
以预定间隔时长采集透析过程中的RBV数据,对于每次采集的RBV数据,执行以下流程:
使用预构建的卡尔曼滤波器对所述RBV数据进行滤波;计算观测到的RBV数据的下降率,使用所述卡尔曼滤波器对计算的所述下降率进行滤波;
基于对当前时刻设定的评估门限,对经滤波后的所述下降率进行评估。
进一步的,所述对当前时刻设定的评估门限,包括第一门限、以及高于所述第一门限的第二门限。
进一步的,所述基于对当前时刻设定的评估门限,对经滤波后的所述下降率进行评估,为基于所述第一门限和第二门限,计算经滤波后的所述下降率的风险系数。
进一步的,所述基于对当前时刻设定的评估门限,对经滤波后的所述下降率进行评估,包括:
在经滤波后的所述下降率处于所述第一门限与第二门限之间时,采用线性插值法或单调递增曲线对经滤波后的所述下降率进行评估。
进一步的,所述第一门限为设定的理想RBV曲线的一阶导数在当前时刻的值,所述第二门限为所述第一门限的N倍,N为大于1的常数。
进一步的,所述理想RBV曲线的一阶导数在第i时刻的值FDRBV i的计算方法为:
FDRBV i=n×EDRBV×a i
Figure PCTCN2022099129-appb-000001
式中,DRBV为预期RBV下降量,T为治疗总时间,n为放大倍数,a为衰减参数。
进一步的,所述卡尔曼滤波器的系统状态预测方程为:
Figure PCTCN2022099129-appb-000002
更新方程为:
Figure PCTCN2022099129-appb-000003
式中,X k为当前时刻的系统状态,X k-1为上一时刻的系统状态,Z k为当前时刻观测到的系统状态,A为状态转移矩阵,H为观测矩阵,Q、R分别表示系统预测模型和传感器的噪声协方差矩阵,P k为X k的协方差矩阵,P k-1为X k-1的协方差矩阵,I为单位矩阵,K k为卡尔曼增益矩阵,是一个中间计算量,带有“^”表示是估计值,带有“ˉ”表示是预测值。
进一步的,所述状态转移矩阵A的计算方法包括:
构建RBV在t时刻的计算模型;
将所述计算模型离散化表示;
计算RBV的变化速率;
使用离散化的计算模型和所述变化速率表示系统状态,与系统状态标准表 示方式类比,得到状态转移矩阵。
进一步的,所述观测矩阵H的计算方法包括:
计算观测到的RBV数据的下降率;
使用离散化的计算模型和所述下降率表示观测到的系统状态,与观测到的系统状态标准表示方式类比,得到观测矩阵。
本发明提供的一种透析中低血压事件风险的评估系统,包括数据采集部、数据评估部和数据输出部,所述数据评估部预构建有卡尔曼滤波器,设定有对应各时刻的评估门限;
所述数据采集部,根据配置的采集周期,采集透析过程中的RBV数据;
所述数据评估部,使用所述卡尔曼滤波器对所述RBV数据进行滤波,计算观测到的RBV数据的下降率,使用所述卡尔曼滤波器对计算的所述下降率进行滤波;基于当前时刻的评估门限,对经滤波后的所述下降率进行评估;
所述数据输出部,输出所述数据评估部的评估结果。
综上所述,由于采用了上述技术方案,本发明的有益效果是:
1、本发明采用卡尔曼滤波对相对血容量数据进行预处理,利用简单的算法滤除干扰信号的影响,使其更好地反应出患者实际的相对血容量变化情况,提高了使用RBV评估患者透析过程中低血压事件风险的准确性。并且,由于卡尔曼滤波算法简单,处理速度快,可以实现对RBV数据的实时处理,减少修正措施的反应时间。
2、本发明使用动态阈值进行低血压事件风险的计算,使其能够更好地适应不同时间段患者对失衡程度的耐受度变化。
3、本发明方法和系统最终的评估结果为对下降率评估得到的归一化风险系数,对风险等级判断更加精细,方便后续的反馈控制或修正措施的参数确定。
附图说明
本发明将通过例子并参照附图的方式说明,其中:
图1是本发评估方法的一个实施方式的流程图。
具体实施方式
本说明书中公开的所有特征,或公开的所有方法或过程中的步骤,除了互相排斥的特征和/或步骤以外,均可以以任何方式组合。
本说明书(包括任何附加权利要求、摘要)中公开的任一特征,除非特别叙述,均可被其他等效或具有类似目的的替代特征加以替换。即,除非特别叙述,每个特征只是一系列等效或类似特征中的一个例子而已。
实施例一
如图1所示,透析中低血压事件风险的评估方法包括以下流程:
第一步、确定透析过程的理想RBV曲线的一阶导数(FDRBV)曲线
首先确定本次治疗预期RBV下降量(DRBV),以及治疗总时长T,默认DRBV设置为24%,治疗总时长T为4小时。当然,两个数据可根据患者实际情况确定。则整个治疗过程中的RBV平均下降率(EDRBV)为:
Figure PCTCN2022099129-appb-000004
取EDRBV的5倍作为初始FDRBV,治疗中第i时刻的FDRBV表示为FDRBV i,则FDRBV i使用下式计算:
FDRBV i=5×EDRBV×0.985 i
第二步、建立卡尔曼滤波器方程和确定滤波参数
卡尔曼滤波是一种通过系统输入输出观测数据对系统状态进行最优估计的算法,观测数据中包括系统中的噪声和干扰的影响,所以最优估计也可看作是滤波过程。卡尔曼滤波器(下文简称滤波器)的工作依据两个方程:系统状态转移方程和系统观测方程。
以X k表示当前时刻的系统状态,系统的状态转移方程可以表示为:
X k=AX k-1+Bu kk
式中,A为状态转移矩阵;X k-1表示上一时刻的系统状态;B为控制矩阵;u k表示控制向量;ω k~N(0,Q k)表示过程激励噪声,是均值为0,方差为Q k的高斯白噪声。
同时,以Z k表示使用测量系统观测到的当前时刻的系统状态,其可表示为:
Z k=HX k+v k
式中,H称为观测矩阵;v k~N(0,R k),表示观测噪声,是均值为0,方差为R k的高斯白噪声。
透析治疗期间患者的RBV由于超滤是不断变化(下降)的,记为rbv,下降速率记为m。那么t时刻的rbv可以表示为:
rbv t=rbv t-Δt+m t-Δt×Δt
式中,Δt表示变化时长,rbv t-Δt表示t-Δt时刻的rbv,m t-Δt表示t-Δt的变化速率。
将上式的rbv离散化表示为:
rbv k=rbv k-1+m k-1
从收集的rbv数据分析得出,在没有意外干扰(患者稳定治疗)的情况下,下降速率m在治疗期间是不断减小的,经验公式为:
m k=a×m k-1
则系统的预测方程可以归纳为:
Figure PCTCN2022099129-appb-000005
将系统的状态统一表示为X k,即:
Figure PCTCN2022099129-appb-000006
那么有:
Figure PCTCN2022099129-appb-000007
即:
Figure PCTCN2022099129-appb-000008
B=0
a的取值在本实施例中为0.985,同第一步中计算FDRBV i公式中的参数。
对于直接观测到的治疗中的rbv,记为rbv′。定义rbv′的下降率m′为:
Figure PCTCN2022099129-appb-000009
其中,C为一常数,linearFit(C)表示对最近的C个rbv′数据进行线性拟合得出的斜率。C的取值决定了对rbv′的短期还是长期变化的估计,在本实施例中,需要对短期变化进行估计,则C取较小值,具体取值取决于rbv′数据的采集周期,在周期为1min的情况下,C取5。
则观测到的系统状态统一表示为Z k
Figure PCTCN2022099129-appb-000010
由于观测到的参数rbv′、m′与系统状态rbv、m是相同的量,无需进行转化,所以Z k又可以表示为:
Figure PCTCN2022099129-appb-000011
也即:
Figure PCTCN2022099129-appb-000012
滤波需要用到以下5个公式:
系统状态预测方程:
Figure PCTCN2022099129-appb-000013
Figure PCTCN2022099129-appb-000014
式中,带有“^”表示是估计值,带有“ˉ”表示是预测值(下同)。
更新方程:
Figure PCTCN2022099129-appb-000015
Figure PCTCN2022099129-appb-000016
Figure PCTCN2022099129-appb-000017
式中,P k为X k的协方差矩阵,P k-1为X k-1的协方差矩阵,Q、R分别表示系统预测模型和传感器的噪声协方差矩阵,需要根据实际情况确定,K k为卡尔曼增益矩阵,是中间计算量,I为单位矩阵。在本实施例中,优选取值为:
Figure PCTCN2022099129-appb-000018
Figure PCTCN2022099129-appb-000019
第三步、采集RBV数据,进行滤波
以预定时间间隔(如每分钟)采集一次RBV数据。
将采集的RBV数据作为滤波器的第一个输入;并依照第二步中的相关算法得出斜率(linearFit(C)),作为滤波器的第二个输入,对RBV数据及其斜率进行滤波,去除噪声,输出滤波后的RBV和斜率。当然,可以仅对斜率进行滤波。
第四步、生成低血压风险评估系数
根据第一步中的方法计算出当前时刻的FDRBV i,将其作为RBV斜率的安全线,并将其进行一定程度的放大(例如n×FDRBV i,n为大于1的正整数),作为RBV斜率的禁止线。基于设定的安全线及禁止线,对采集的RBV数据进行风险评估。具体的,为根据滤波后的RVB的斜率与安全线和禁止线间的大小关系进行风险评估。
1.如果滤波后的RVB的斜率超过FDRBV i(即安全线),则认为极有可能已(或将)发生低血压事件,其风险系数记为1。
2.如果滤波后的RBV的斜率小于FDRBV i,则认为没有发生低血压事件的 风险,其风险系数记为0。
3.如果滤波后的RBV的斜率在FDRBV i与n×FDRBV i之间,则采用线性插值法或其它单调递增曲线计算风险系数。
第五部、判断治疗过程是否结束,若是,则结束评估,否则,跳转到第三步,继续对下一次采集的RBV数据进行评估。
实施例二
本实施例公开了一种透析中低血压事件风险的评估系统,包括数据采集部、数据评估部和数据输出部。数据评估部预构建有卡尔曼滤波器,并且设定有对应各时刻的评估门限。
评估门限是基于理想RBV曲线的一阶导数曲线(FDRBV)设定的。首先确定本次治疗预期RBV下降量(DRBV),以及治疗总时长T,默认DRBV设置为24%,治疗总时长T为4小时。当然,两个数据可根据患者实际情况确定。则整个治疗过程中的RBV平均下降率(EDRBV)为:
Figure PCTCN2022099129-appb-000020
取EDRBV的n倍作为初始FDRBV,治疗中第i时刻的FDRBV表示为FDRBV i,则FDRBV i使用下式计算:
FDRBV i=n×EDRBV×a i
式中,n为放大倍数,a为衰减参数,a<1;在本实例中n取5,a取0.985。
评估阈值包括安全线和禁止线,以当前时刻的FDRBV i作为RBV斜率的安全线,并将其进行一定程度的放大(例如N×FDRBV i,N为大于1的常数),作为RBV斜率的禁止线。
预构建的卡尔曼滤波器具体如下:
卡尔曼滤波器(下文简称滤波器)的工作依据两个方程:系统状态转移方 程和系统观测方程。
以X k表示系统状态,系统的状态转移方程可以表示为:
X k=AX k-1+Bu kk
式中,A为状态转移矩阵;X k-1表示上一时刻的系统状态;B为控制矩阵;u k表示控制向量;ω k~N(0,Q k)表示过程激励噪声,是均值为0,方差为Q k的高斯白噪声。
同时,以Z k表示使用测量系统观测到的系统当前状态,其可表示为:
Z k=HX k+v k
式中,H称为观测矩阵;v k~N(0,R k),表示观测噪声,是均值为0,方差为R k的高斯白噪声。
透析治疗期间患者的RBV由于超滤是不断变化(下降)的,记为rbv,下降速率记为m。那么t时刻的rbv可以表示为:
rbv t=rbv t-Δt+m t-Δt×Δt
式中,Δt表示变化时长,rbv t-Δt表示t-Δt时刻的rbv,m t-Δt表示t-Δt的变化速率。
将上式的rbv离散化表示为:
rbv k=rbv k-1+m k-1
从收集的rbv数据分析得出,在没有意外干扰(患者稳定治疗)的情况下,下降速率m在治疗期间是不断减小的,经验公式为:
m k=a×m k-1
则系统的预测方程可以归纳为:
Figure PCTCN2022099129-appb-000021
将系统的状态统一表示为X k,即:
Figure PCTCN2022099129-appb-000022
那么有:
Figure PCTCN2022099129-appb-000023
即:
Figure PCTCN2022099129-appb-000024
B=0
a的取值在本方案中为0.985,同第一步中计算FDRBV i公式中的参数。
对于直接观测到的治疗中的rbv,记为rbv′。定义rbv′的下降率m′为:
Figure PCTCN2022099129-appb-000025
其中,C为一常数,linearFit(C)表示对最近的C个rbv′数据进行线性拟合得出的斜率。C的取值决定了对rbv′的短期还是长期变化的估计,在本实施例中,需要对短期变化进行估计,则C取较小值,具体取值取决于rbv′数据的采集周期,在周期为1min的情况下,C取5。
则观测到的系统状态统一表示为Z k
Figure PCTCN2022099129-appb-000026
由于观测到的参数rbv′、m′与系统状态rbv、m是相同的量,无需进行转化,所以Z k又可以表示为:
Figure PCTCN2022099129-appb-000027
也即:
Figure PCTCN2022099129-appb-000028
滤波需要用到以下5个公式:
系统状态预测方程:
Figure PCTCN2022099129-appb-000029
Figure PCTCN2022099129-appb-000030
式中,带有“^”表示是估计值(下同);带有“ˉ”表示是预测值(下同)。
更新方程:
Figure PCTCN2022099129-appb-000031
Figure PCTCN2022099129-appb-000032
Figure PCTCN2022099129-appb-000033
式中,Q、R分别表示系统预测模型和传感器的噪声协方差矩阵,需要根据实际情况确定,K k为卡尔曼增益矩阵,是中间计算量,I为单位矩阵。在本实施例中,优选取值为:
Figure PCTCN2022099129-appb-000034
Figure PCTCN2022099129-appb-000035
数据采集部以预定时间间隔(如每分钟)采集一次RBV数据,每次采集的RBV数据均传递给数据评估部。
数据评估部将接收的RBV数据作为滤波器的第一个输入;并依照第二步中的相关算法得出斜率(linearFit(C)),作为滤波器的第二个输入,对RBV数据及其斜率进行滤波,去除噪声,输出滤波后的RBV和斜率。当然,可以仅对斜率进行滤波。
对于经滤波后的斜率,根据其与安全线和禁止线间的大小关系进行风险评估:
1.如果滤波后的RVB的斜率超过FDRBV i(即安全线),则认为极有可能已(或将)发生低血压事件,其风险系数记为1。
2.如果滤波后的RBV的斜率小于FDRBV i,则认为没有发生低血压事件的风险,其风险系数记为0。
3.如果滤波后的RBV的斜率在FDRBV i与n×FDRBV i之间,则采用线性插值法或其它单调递增曲线计算风险系数。
数据输出部连接数据评估部,输出数据评估部每次对数据采集部采集的RBV数据的评估结果(即风险系数)。数据输出部输出的数据可作为反馈控制或修正措施的参数。
本发明并不局限于前述的具体实施方式。本发明扩展到任何在本说明书中披露的新特征或任何新的组合,以及披露的任一新的方法或过程的步骤或任何新的组合。

Claims (10)

  1. 一种透析中低血压事件风险的评估方法,其特征在于,包括:
    以预定间隔时长采集透析过程中的RBV数据,对于每次采集的RBV数据,执行以下流程:
    使用预构建的卡尔曼滤波器对所述RBV数据进行滤波;计算观测到的RBV数据的下降率,使用所述卡尔曼滤波器对计算的所述下降率进行滤波;
    基于对当前时刻设定的评估门限,对经滤波后的所述下降率进行评估。
  2. 如权利要求1所述的透析中低血压事件风险的评估方法,其特征在于,所述对当前时刻设定的评估门限,包括第一门限、以及高于所述第一门限的第二门限。
  3. 如权利要求2所述的透析中低血压事件风险的评估方法,其特征在于,所述基于对当前时刻设定的评估门限,对经滤波后的所述下降率进行评估,为基于所述第一门限和第二门限,计算经滤波后的所述下降率的风险系数。
  4. 如权利要求3所述的透析中低血压事件风险的评估方法,其特征在于,所述基于对当前时刻设定的评估门限,对经滤波后的所述下降率进行评估,包括:
    在经滤波后的所述下降率处于所述第一门限与第二门限之间时,采用线性插值法或单调递增曲线对经滤波后的所述下降率进行评估。
  5. 如权利要求2~4任一所述的透析中低血压事件风险的评估方法,其特征在于,所述第一门限为设定的理想RBV曲线的一阶导数在当前时刻的值,所述第二门限为所述第一门限的N倍,N为大于1的常数。
  6. 如权利要求5所述的透析中低血压事件风险的评估方法,其特征在于,所述理想RBV曲线的一阶导数FDRBV在第i时刻的值FDRBV i的计算方法为:
    FDRBV i=n×EDRBV×a i
    Figure PCTCN2022099129-appb-100001
    式中,n为放大倍数,DRBV为预期RBV下降量,T为治疗总时间,a为衰减参数。
  7. 如权利要求1所述的透析中低血压事件风险的评估方法,其特征在于,所述卡尔曼滤波器的系统状态预测方程为:
    Figure PCTCN2022099129-appb-100002
    更新方程为:
    Figure PCTCN2022099129-appb-100003
    Figure PCTCN2022099129-appb-100004
    Figure PCTCN2022099129-appb-100005
    式中,X k为当前时刻的系统状态,X k-1为上一时刻的系统状态,Z k为当前时刻观测到的系统状态,A为状态转移矩阵,H为观测矩阵,Q、R分别表示系统预测模型和传感器的噪声协方差矩阵,P k为X k的协方差矩阵,P k-1为X k-1的协方差矩阵,I为单位矩阵,K k为卡尔曼增益矩阵,是一个中间计算量,带有“^”表示是估计值,带有“ˉ”表示是预测值。
  8. 如权利要求7所述的透析中低血压事件风险的评估方法,其特征在于,所述状态转移矩阵A的计算方法包括:
    构建RBV在t时刻的计算模型;
    将所述计算模型离散化表示;
    计算RBV的变化速率;
    使用离散化的计算模型和所述变化速率表示系统状态,与系统状态标准表示方式类别,得到状态转移矩阵。
  9. 如权利要求8所述的透析中低血压事件风险的评估方法,其特征在于,所述观测矩阵H的计算方法包括:
    计算观测到的RBV数据的下降率;
    使用离散化的计算模型和所述下降率表示观测到的系统状态,与观测到的系统状态标准表示方式类比,得到观测矩阵。
  10. 一种透析中低血压事件风险的评估系统,其特征在于,包括数据采集部、数据评估部和数据输出部,所述数据评估部预构建有卡尔曼滤波器,设定有对 应各时刻的评估门限;
    所述数据采集部,根据配置的采集周期,采集透析过程中的RBV数据;
    所述数据评估部,使用所述卡尔曼滤波器对所述RBV数据进行滤波,计算观测到的RBV数据的下降率,使用所述卡尔曼滤波器对计算的所述下降率进行滤波;基于当前时刻的评估门限,对经滤波后的所述下降率进行评估;
    所述数据输出部,输出所述数据评估部的评估结果。
PCT/CN2022/099129 2021-09-23 2022-06-16 一种透析中低血压事件风险的评估方法和系统 WO2023045425A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111112325.5 2021-09-23
CN202111112325.5A CN113823409B (zh) 2021-09-23 2021-09-23 一种透析中低血压事件风险的评估方法和系统

Publications (1)

Publication Number Publication Date
WO2023045425A1 true WO2023045425A1 (zh) 2023-03-30

Family

ID=78921048

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/099129 WO2023045425A1 (zh) 2021-09-23 2022-06-16 一种透析中低血压事件风险的评估方法和系统

Country Status (2)

Country Link
CN (1) CN113823409B (zh)
WO (1) WO2023045425A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117009831B (zh) * 2023-10-07 2023-12-08 山东世纪阳光科技有限公司 一种精细化工事故风险预测评估方法
CN117798744A (zh) * 2024-02-29 2024-04-02 茌平县汇通机械制造有限公司 一种数控机床运行状态监测方法

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113823409B (zh) * 2021-09-23 2023-12-29 重庆山外山血液净化技术股份有限公司 一种透析中低血压事件风险的评估方法和系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130267858A1 (en) * 2010-07-08 2013-10-10 Intelomed, Inc. System and method for characterizing circulatory blood flow
CN104703534A (zh) * 2012-08-28 2015-06-10 弗雷泽纽斯医疗保健控股有限公司 通过测量相对血容量、血压和心率来检测透析中病态事件的指示的方法
CN108986419A (zh) * 2018-10-17 2018-12-11 暨南大学 一种用于血液透析的数据报警方法
CN110151153A (zh) * 2013-02-28 2019-08-23 B·布莱恩·阿维图姆股份公司 模糊逻辑
CN112203578A (zh) * 2018-03-20 2021-01-08 甘布罗伦迪亚股份公司 用于确定体外血液回路中循环的血液的至少一个参数的传感器和设备
CN113823409A (zh) * 2021-09-23 2021-12-21 重庆山外山血液净化技术股份有限公司 一种透析中低血压事件风险的评估方法和系统

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10751004B2 (en) * 2016-07-08 2020-08-25 Edwards Lifesciences Corporation Predictive weighting of hypotension profiling parameters
US11076813B2 (en) * 2016-07-22 2021-08-03 Edwards Lifesciences Corporation Mean arterial pressure (MAP) derived prediction of future hypotension
CN111939353A (zh) * 2019-05-14 2020-11-17 吴元昊 血液透析中低血压事件的预测模型的构建方法
CN110648755A (zh) * 2019-09-10 2020-01-03 云南博亚医院有限公司 一种血液透析质量评估及管理系统

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130267858A1 (en) * 2010-07-08 2013-10-10 Intelomed, Inc. System and method for characterizing circulatory blood flow
CN104703534A (zh) * 2012-08-28 2015-06-10 弗雷泽纽斯医疗保健控股有限公司 通过测量相对血容量、血压和心率来检测透析中病态事件的指示的方法
CN110151153A (zh) * 2013-02-28 2019-08-23 B·布莱恩·阿维图姆股份公司 模糊逻辑
CN112203578A (zh) * 2018-03-20 2021-01-08 甘布罗伦迪亚股份公司 用于确定体外血液回路中循环的血液的至少一个参数的传感器和设备
CN108986419A (zh) * 2018-10-17 2018-12-11 暨南大学 一种用于血液透析的数据报警方法
CN113823409A (zh) * 2021-09-23 2021-12-21 重庆山外山血液净化技术股份有限公司 一种透析中低血压事件风险的评估方法和系统

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117009831B (zh) * 2023-10-07 2023-12-08 山东世纪阳光科技有限公司 一种精细化工事故风险预测评估方法
CN117798744A (zh) * 2024-02-29 2024-04-02 茌平县汇通机械制造有限公司 一种数控机床运行状态监测方法
CN117798744B (zh) * 2024-02-29 2024-05-10 茌平县汇通机械制造有限公司 一种数控机床运行状态监测方法

Also Published As

Publication number Publication date
CN113823409A (zh) 2021-12-21
CN113823409B (zh) 2023-12-29

Similar Documents

Publication Publication Date Title
WO2023045425A1 (zh) 一种透析中低血压事件风险的评估方法和系统
US10342910B2 (en) Device and method for predicting intradialytic parameters
WO2019051969A1 (zh) 心率估算方法和装置
JP2017042269A (ja) 瞬目検知装置
KR20150113700A (ko) 진단 시스템 및 방법
CN112043258A (zh) 动态心率预测方法、装置、设备及可读存储介质
CN116682566B (zh) 一种血液透析的数据处理方法及系统
Doike et al. A blood glucose level prediction system using machine learning based on recurrent neural network for hypoglycemia prevention
US11406309B2 (en) Heartbeat rate calculation device and method
CN114916940A (zh) 基于专家知识的心律失常信号检测方法、系统及终端
CN111317458A (zh) 一种基于深度学习的血压检测系统
US9265477B2 (en) Adaptive lightweight acoustic signal classification for physiological monitoring
US20180050144A1 (en) Method and system for postdialytic determination of dry weight
CN110378031B (zh) 实时呼吸信号监测方法、装置、呼吸监测器及存储介质
CN110710981B (zh) 基于二值传感器卡尔曼融合的血液氧气含量估计方法
TWI693062B (zh) 預估血壓驟降的方法及電子裝置
CN112535464A (zh) 基于pai的状态评估方法、装置、设备和存储介质
Chen et al. IoT-enabled intelligent dynamic risk assessment of acute mountain sickness based on data from wearable devices
CN112932424A (zh) 一种数据获取方法及系统
CN115422976B (zh) 一种基于人工网络的心肺耦合关系分析方法及监测系统
CN117243569B (zh) 一种基于多源信息融合的认知功能评估方法和系统
CN117679624B (zh) 基于一次性使用压力传感器的冲洗阀流量智能控制方法
CN110720929B (zh) 基于二值传感器有界递归优化融合的血液氧气含量估计方法
CN115177271A (zh) Ssvep异步系统中的检测指标的个性化阈值确定方法及应用
Shoji et al. Robust R-Wave Detection Under Long-Term Measurement Using HRV Sensor System with Automatic Gain Readjustment

Legal Events

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

Ref document number: 22871486

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE