CN112397176A - Intelligent oxytocin dose regulation and control method and system based on uterine contraction signals and LightGBM - Google Patents
Intelligent oxytocin dose regulation and control method and system based on uterine contraction signals and LightGBM Download PDFInfo
- Publication number
- CN112397176A CN112397176A CN202011110757.8A CN202011110757A CN112397176A CN 112397176 A CN112397176 A CN 112397176A CN 202011110757 A CN202011110757 A CN 202011110757A CN 112397176 A CN112397176 A CN 112397176A
- Authority
- CN
- China
- Prior art keywords
- uterine contraction
- data
- oxytocin
- uterine
- lightgbm
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
- 208000036029 Uterine contractions during pregnancy Diseases 0.000 title claims abstract description 175
- 101800000989 Oxytocin Proteins 0.000 title claims abstract description 149
- 102100031951 Oxytocin-neurophysin 1 Human genes 0.000 title claims abstract description 149
- XNOPRXBHLZRZKH-UHFFFAOYSA-N Oxytocin Natural products N1C(=O)C(N)CSSCC(C(=O)N2C(CCC2)C(=O)NC(CC(C)C)C(=O)NCC(N)=O)NC(=O)C(CC(N)=O)NC(=O)C(CCC(N)=O)NC(=O)C(C(C)CC)NC(=O)C1CC1=CC=C(O)C=C1 XNOPRXBHLZRZKH-UHFFFAOYSA-N 0.000 title claims abstract description 148
- XNOPRXBHLZRZKH-DSZYJQQASA-N oxytocin Chemical compound C([C@H]1C(=O)N[C@H](C(N[C@@H](CCC(N)=O)C(=O)N[C@@H](CC(N)=O)C(=O)N[C@@H](CSSC[C@H](N)C(=O)N1)C(=O)N1[C@@H](CCC1)C(=O)N[C@@H](CC(C)C)C(=O)NCC(N)=O)=O)[C@@H](C)CC)C1=CC=C(O)C=C1 XNOPRXBHLZRZKH-DSZYJQQASA-N 0.000 title claims abstract description 148
- 229960001723 oxytocin Drugs 0.000 title claims abstract description 148
- 238000000034 method Methods 0.000 title claims abstract description 47
- 230000033228 biological regulation Effects 0.000 title claims abstract description 39
- 230000008859 change Effects 0.000 claims abstract description 29
- 238000004519 manufacturing process Methods 0.000 claims abstract description 29
- 238000000605 extraction Methods 0.000 claims abstract description 24
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 21
- 238000012216 screening Methods 0.000 claims abstract description 20
- 238000005457 optimization Methods 0.000 claims abstract description 16
- 238000004364 calculation method Methods 0.000 claims abstract description 8
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 210000002458 fetal heart Anatomy 0.000 claims description 43
- 238000001914 filtration Methods 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 20
- 230000001605 fetal effect Effects 0.000 claims description 12
- 239000012530 fluid Substances 0.000 claims description 7
- 238000009499 grossing Methods 0.000 claims description 7
- 238000003064 k means clustering Methods 0.000 claims description 7
- 238000000556 factor analysis Methods 0.000 claims description 6
- 230000007246 mechanism Effects 0.000 claims description 6
- 230000032696 parturition Effects 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 claims description 3
- 238000001802 infusion Methods 0.000 abstract description 33
- 230000000875 corresponding effect Effects 0.000 description 27
- 230000006870 function Effects 0.000 description 25
- 239000003814 drug Substances 0.000 description 18
- 229940079593 drug Drugs 0.000 description 17
- 239000000523 sample Substances 0.000 description 15
- 230000008569 process Effects 0.000 description 13
- 238000009826 distribution Methods 0.000 description 8
- 238000012544 monitoring process Methods 0.000 description 8
- 238000012549 training Methods 0.000 description 8
- 238000005070 sampling Methods 0.000 description 7
- 239000011159 matrix material Substances 0.000 description 5
- 238000002790 cross-validation Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000001105 regulatory effect Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 239000000243 solution Substances 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000035606 childbirth Effects 0.000 description 2
- 230000001276 controlling effect Effects 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000006698 induction Effects 0.000 description 2
- 238000002347 injection Methods 0.000 description 2
- 239000007924 injection Substances 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 210000000754 myometrium Anatomy 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 210000004291 uterus Anatomy 0.000 description 2
- 206010067484 Adverse reaction Diseases 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 208000001951 Fetal Death Diseases 0.000 description 1
- 208000032943 Fetal Distress Diseases 0.000 description 1
- 206010055690 Foetal death Diseases 0.000 description 1
- 206010016855 Foetal distress syndrome Diseases 0.000 description 1
- 206010021718 Induced labour Diseases 0.000 description 1
- 210000001015 abdomen Anatomy 0.000 description 1
- 210000003815 abdominal wall Anatomy 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000006838 adverse reaction Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 230000036592 analgesia Effects 0.000 description 1
- 230000010455 autoregulation Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000036772 blood pressure Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000010339 dilation Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000002196 ecbolic effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000009123 feedback regulation Effects 0.000 description 1
- 210000003754 fetus Anatomy 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 208000004104 gestational diabetes Diseases 0.000 description 1
- 229940090044 injection Drugs 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000001990 intravenous administration Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000008774 maternal effect Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- XNOPRXBHLZRZKH-MQYCRUOZSA-N oxytocin Chemical compound C([C@H]1C(=O)N[C@H](C(N[C@@H](CCC(N)=O)C(=O)N[C@@H](CC(N)=O)C(=O)N[C@@H](CSSC[C@H](N)C(=O)N1)C(=O)N1C(CCC1)C(=O)N[C@@H](CC(C)C)C(=O)NCC(N)=O)=O)[C@@H](C)CC)C1=CC=C(O)C=C1 XNOPRXBHLZRZKH-MQYCRUOZSA-N 0.000 description 1
- 229940082408 oxytocin injection Drugs 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000035935 pregnancy Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000035485 pulse pressure Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000000284 resting effect Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 210000004003 subcutaneous fat Anatomy 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
- G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
- G16H20/17—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients delivered via infusion or injection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24317—Piecewise classification, i.e. whereby each classification requires several discriminant rules
Abstract
The invention provides an intelligent oxytocin dose regulation and control method based on a uterine contraction signal and a LightGBM, which comprises the steps of obtaining and preprocessing the uterine contraction signal, and further extracting features based on uterine contraction strength; obtaining characteristic variables with statistical significance and intrapartum physiological parameters to form data characteristic items; according to the data characteristic items, combining the demographic data and the intrapartum physiological parameter data to obtain initial sample data, and screening out model input data in the initial sample data by combining the uterine contraction signals after characteristic extraction; constructing a oxytocin dose prediction model during production based on a Bayesian optimization LightGBM algorithm; and importing the model input data into a oxytocin dose prediction model for calculation to obtain a numerical value corresponding to the final oxytocin dropping speed change condition. By implementing the invention, the contradiction between dependence of the traditional infusion scheme on personal experience and shortage of human resources of medical staff can be overcome, the labor cost is reduced, and the intelligent regulation and control of the accurate oxytocin dosage are realized.
Description
Technical Field
The invention relates to the technical field of computer intelligence and the technical field of medicines, in particular to an oxytocin dose intelligent regulation and control method and system based on uterine contraction signals and LightGBM.
Background
Oxytocin, also known as Oxytocin (OT), is a first-line drug for induction of labor and labor in obstetrics, and has the functions of starting labor, promoting uterine contraction and accelerating labor. Clinically, the OT oxytocic and induced labor is intravenous drip, medical staff are required to observe uterine contraction and fetal heart rate change conditions in real time, the infusion speed of the OT micropump is adjusted every 15-20min by combining factors such as labor progress, pulse and blood pressure of a parturient, liquid inflow and outflow amount balance and the like, and the speed can be increased or reduced by 5 drops every time according to an arithmetic difference method (the intelligent rate of oxytocin dosage is constant). According to the actual clinical medication situation, the dropping speed regulation situation can be divided into: maintaining the original speed, increasing the dropping speed and reducing the dropping speed indicate that the OT oxytocin dose intelligently regulates the trend of the dose. However, due to different sensitivities of the puerperae in the end period of pregnancy to OT, the infusion dosage of the puerperae also has difference, and if the puerperae is excessively taken, the puerperae can cause hyperuterine contraction, excessive density, fetal heart deceleration, even ankylosing uterine contraction, fetal distress, death and the like. Therefore, the key to the safety and the effectiveness of oxytocin administration lies in accurately adjusting the use amount according to individual reaction.
The birth guide of the obstetrics and sciences group of the Chinese medical society suggests that continuous Electronic Fetal heart Monitoring (EFM) is adopted during OT labor induction, the intrauterine condition of a fetus is monitored, the uterine contraction pressure change is indirectly measured through an external uterine contraction probe, and two important indexes of the Fetal heart rate and uterine contraction are reflected. However, the phenomenon that one midwife monitors multiple puerperae at the same time often occurs in the mode of observing uterine contraction and adjusting OT dripping speed according to the guide, OT input speed cannot be adjusted reasonably in time, medication effect is affected, and even the safety of the mothers and the children is threatened, and the mode has subjective errors of medical staff and high labor cost.
Currently, there is also a study of applying computers to OT auto-regulation. For example, zhengyuan et al (CN206434657U) developed an oxytocin injection automatic regulating device, which connects the fetal monitor and the infusion pump through a single chip microcomputer to realize that the infusion speed of the infusion pump is regulated according to the uterine contraction strength, and can give an alarm when the fetal heart rate exceeds the normal range or abnormal uterine contraction occurs. For another example, wai xiao yi et al (CN203802891U) designed an oxytocin automatic drip regulator for obstetrical department, which outputs control signals to the infusion pump through the CPU unit of the DSP chip and the signal amplification and driving circuit by the uterine contraction frequency and intensity signals collected by the pressure sensor. However, the above studies do not perform detailed data analysis and establish a relevant prediction model, and only determine and control the OT dose according to preset rules, thereby playing a role of simple early warning. As another example, silverano et al, based on OT pharmacokinetics, simulated the interaction between uterine contraction frequency, uterine cavity pressure, ostial dilation, fetal presenting fluid and OT concentration, and automatically adjusted OT dose according to real-time changes in input variables, but this study only performed a preliminary feasibility analysis of model effects, and did not mention a clear evaluation or validation scheme.
Therefore, an intelligent oxytocin dose regulation and control method is urgently needed, the contradiction that the traditional infusion scheme depends on personal experience and human resource of medical staff is in short supply can be overcome, an OT dose prediction model is built, auxiliary decision support is provided for the OT infusion scheme of obstetrical medical staff, labor cost is reduced, and accurate oxytocin dose intelligent regulation and control are achieved.
Disclosure of Invention
The technical problem to be solved by the embodiment of the invention is to provide an intelligent oxytocin dose regulation and control method and system based on a uterine contraction signal and a LightGBM, which can overcome the contradiction between dependence of a traditional infusion scheme on personal experience and shortage of human resources of medical staff, provide assistant decision support for an OT infusion scheme of obstetrical medical staff by constructing an OT dose prediction model, reduce labor cost and realize accurate intelligent oxytocin dose regulation and control.
In order to solve the technical problem, an embodiment of the present invention provides an intelligent oxytocin dose regulation and control method based on a uterine contraction signal and a LightGBM, including the following steps:
s1, acquiring a uterine contraction signal detected by the fetal heart monitor, preprocessing the uterine contraction signal, and further extracting the characteristics of the preprocessed uterine contraction signal based on the uterine contraction strength;
s2, acquiring demographics of oxytocin dose adjustment, screening out characteristic variables with statistical significance, acquiring parturition physiological parameters output by a fetal heart monitor, and forming data characteristic items by the characteristic variables with statistical significance and the parturition physiological parameters;
s3, according to the data feature items, combining the data corresponding to the demographic data of oxytocin dose adjustment and the time production physiological parameters to obtain initial sample data with the data feature items, and screening model input data in the initial sample data by combining the uterine contraction signals after feature extraction;
s4, constructing a oxytocin dose prediction model during production based on a Bayesian optimization LightGBM algorithm; the oxytocin dose prediction model during production takes the model input data as data input and takes the oxytocin dripping speed change condition as target output;
and S5, importing the model input data into the intermediate oxytocin dose prediction model for calculation to obtain a numerical value corresponding to the final oxytocin dropping speed change condition.
Wherein, the step S1 specifically includes:
acquiring a uterine contraction signal detected by the fetal heart monitor through a pressure sensor;
based on a signal smoothing mechanism of wavelet filtering, filtering the uterine contraction signal;
and after determining uterine cavity pressure, uterine contraction frequency and uterine contraction duration in the uterine contraction signals after filtering, distinguishing uterine contraction strength by using a K-means clustering method to obtain uterine contraction signals after feature extraction.
Wherein, the step S2 specifically includes:
acquiring demographic data of oxytocin dose adjustment from a preset electronic medical record system, and screening out characteristic variables with statistical significance through single factor analysis and multiple linear stepwise regression; wherein said statistically significant characteristic variables include age, BMI, gestational week, fetal presenting fluid, and uterine height;
acquiring physiological parameters during production from the data port of the fetal heart monitor; wherein the intrapartum physiological parameters comprise the fetal heart, the uterine contraction frequency, the uterine contraction duration, the uterine contraction intensity and the average value of the uterine cavity pressure peak value in a specified time;
and forming the characteristic variables with statistical significance and the labor-hour physiological parameters into data characteristic items.
In step S5, the oxytocin dropping speed change condition includes maintaining the dropping speed at the original speed, slowing down the dropping speed, and speeding up the dropping speed; the numerical values corresponding to the oxytocin dripping speed change conditions comprise a numerical value 3 corresponding to the dripping speed maintaining original speed, a numerical value 2 corresponding to the dripping speed slowing down and a numerical value 1 corresponding to the dripping speed speeding up.
The embodiment of the invention also provides an oxytocin dose intelligent regulation and control system based on a uterine contraction signal and a LightGBM, which comprises the following steps:
the signal processing and extracting unit is used for acquiring a uterine contraction signal detected by the fetal heart monitor, preprocessing the uterine contraction signal and further extracting the characteristics of the preprocessed uterine contraction signal based on the uterine contraction strength;
the data characteristic item forming unit is used for acquiring demographics data of oxytocin dose adjustment, screening out characteristic variables with statistical significance, acquiring intrapartum physiological parameters output by a fetal heart monitor, and forming the characteristic variables with the statistical significance and the intrapartum physiological parameters into data characteristic items;
the model input data screening unit is used for combining the data corresponding to the adjusted demographics data of the oxytocin dose and the time production physiological parameters to obtain initial sample data with the data characteristic items according to the data characteristic items, and screening model input data from the initial sample data by combining the uterine contraction signals after characteristic extraction;
the prediction model construction unit is used for constructing a oxytocin dose prediction model during production based on a Bayesian optimization LightGBM algorithm; the oxytocin dose prediction model during production takes the model input data as data input and takes the oxytocin dripping speed change condition as target output;
and the prediction model result output unit is used for importing the model input data into the oxytocin dose prediction model during production to calculate, and obtaining a numerical value corresponding to the final oxytocin dropping speed change condition.
Wherein the signal processing and extracting unit includes:
the signal acquisition module is used for acquiring a uterine contraction signal detected by the fetal heart monitor through the pressure sensor;
the signal filtering processing module is used for carrying out filtering processing on the uterine contraction signal based on a signal smoothing processing mechanism of wavelet filtering;
and the clustering feature extraction module is used for distinguishing the uterine contraction strength by using a K-means clustering method after determining the uterine cavity pressure, the uterine contraction frequency and the uterine contraction duration time in the uterine contraction signals after the filtering processing so as to obtain the uterine contraction signals after feature extraction.
Wherein the data feature item forming unit includes:
the first variable acquisition module is used for acquiring demographics data of oxytocin dose adjustment from a preset electronic medical record system and screening out characteristic variables with statistical significance through single factor analysis and multiple linear stepwise regression; wherein said statistically significant characteristic variables include age, BMI, gestational week, fetal presenting fluid, and uterine height;
the second variable acquisition module is used for acquiring physiological parameters during production from the data port of the fetal heart monitor; wherein the intrapartum physiological parameters comprise the fetal heart, the uterine contraction frequency, the uterine contraction duration, the uterine contraction intensity and the average value of the uterine cavity pressure peak value in a specified time;
and the data feature item combination module is used for combining the feature variable with statistical significance and the labor-producing physiological parameter into a data feature item.
Wherein the oxytocin dripping speed change condition comprises that the dripping speed maintains the original speed, the dripping speed is slowed down and the dripping speed is accelerated; the numerical values corresponding to the oxytocin dripping speed change conditions comprise a numerical value 3 corresponding to the dripping speed maintaining original speed, a numerical value 2 corresponding to the dripping speed slowing down and a numerical value 1 corresponding to the dripping speed speeding up.
The embodiment of the invention has the following beneficial effects:
1. the invention provides a Bayesian optimization-based LightGBM to establish an OT dripping speed prediction model, which has high training speed and high model precision and overcomes the contradiction between dependence of a traditional infusion scheme on personal experience and shortage of human resources of medical staff;
2. compared with the traditional OT infusion scheme, the LightGBM model based on Bayesian optimization has the advantages of accurate regulation and control, safe medication and higher accuracy, is used as an auxiliary decision system for obstetrical medical personnel, innovates the current domestic mode for regulating and controlling OT infusion, and saves a large amount of resources for hospitals in future research or replacement of obstetrical departments for OT infusion.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is within the scope of the present invention for those skilled in the art to obtain other drawings based on the drawings without inventive exercise.
Fig. 1 is a flowchart of an oxytocin dose intelligent regulation and control method based on a uterine contraction signal and a LightGBM according to an embodiment of the present invention;
fig. 2 is a schematic diagram of uterine contraction signals after denoising of small waves in step S1 in fig. 1;
FIG. 3 is a graph of feature extraction of the uterine contraction signal in step S1 of FIG. 1;
FIG. 4 is a diagram illustrating the result of the K-means uterine contraction strength feature extraction in step S1 of FIG. 1;
fig. 5 is a graph comparing a uterine contraction signal and OT medication in an application scenario of the oxytocin dose intelligent regulation and control method based on the uterine contraction signal and the LightGBM according to the embodiment of the present invention;
fig. 6 is an importance distribution diagram of variables included in a data feature item in an application scenario of the oxytocin dose intelligent regulation and control method based on a uterine contraction signal and a LightGBM according to the embodiment of the present invention;
fig. 7 is a schematic structural diagram of an oxytocin dose intelligent regulation and control system based on a uterine contraction signal and a LightGBM according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings.
As shown in fig. 1, in an embodiment of the present invention, an intelligent oxytocin dose regulation and control method based on a uterine contraction signal and a LightGBM is provided, and the method may be embedded into a nurse station system as a data analysis center to realize intelligent regulation and control of an infusion micropump, so as to form a set of regulation system capable of facilitating monitoring of uterine contraction and fetal heart rate variation, intelligent regulation of OT infusion, and artificial assisted monitoring, and specifically includes the following steps:
step S1, acquiring a uterine contraction signal detected by a fetal heart monitor, preprocessing the uterine contraction signal, and further extracting the characteristics of the preprocessed uterine contraction signal based on the uterine contraction strength; firstly, acquiring a uterine contraction signal detected by a fetal heart monitor (such as a model M2702A) through a pressure sensor;
at the moment, the collection of the uterine contraction pressure data comes from the fetal heart monitor, the pressure sensor is used as a uterine contraction detector, and the uterine contraction condition is displayed according to the uterine contraction pressure. During monitoring, the pressure sensor is placed at the bottom of the uterus of the pregnant woman, when the uterine muscle contracts, the tension of the abdominal wall is placed on the pressure sensor, and the uterine contraction pressure signal is converted into an electric signal.
Secondly, filtering the uterine contraction signal based on a signal smoothing mechanism of wavelet filtering;
at the moment, interference signals such as the placement position of the pressure probe, the thickness of subcutaneous fat of the abdomen, fetal movement, the tightness of the binding of the probe, the movement of a parturient and the like occur in the uterine contraction signal acquisition process, so that the interpretation of uterine contraction waveforms is influenced, the characteristics of the uterine contraction conditions in the signals are required to be kept as much as possible while noise is removed, wavelet filtering-based signal smoothing processing is adopted in the uterine contraction signal acquisition process, and the effect is shown in fig. 2. And (3) carrying out 4-scale decomposition on the actually measured signal by adopting a Sym6 wavelet packet, and taking the decomposed profile coefficient as the de-noised uterine contraction signal.
The wavelet filter is expressed as that after the wavelet function psi (t) is subjected to displacement b, the time domain graph of the uterine contraction signal is recovered after the wavelet function psi (t) is subjected to inner product with the BCG signal f (t) under different scales a.
And finally, after determining the uterine cavity pressure, the uterine contraction frequency and the uterine contraction duration in the uterine contraction signals after filtering, distinguishing the uterine contraction strength by using a K-means clustering method to obtain the uterine contraction signals after feature extraction.
During the childbirth, the uterine contractions of the lying-in woman are rapidly concentrated from the uterine corners at the two sides to the bottom of the uterus, gradually weaken at the speed of 2cm per second, gradually weaken from weak to strong (advancing stage) each time, maintain for a certain time (polar stage), then weaken from strong to weak (retreating stage) until disappearance enters the intermittent stage, the uterine muscles in the intermittent stage relax, and the uterine contractions are repeated until the childbirth is finished. The uterine contraction peak value is the strongest pressure point of the primary uterine contraction, the uterine cavity pressure peak value point needs to be positioned firstly for extracting the uterine contraction, and the algorithm process is as follows:
after confirming the peak point, determining a uterine contraction starting point and a uterine contraction ending point according to the sequential distribution of the peak point in the time domain: two minima points at the baseline level adjacent to the peak point in the time sequence are shown in fig. 3. The amplitude of the peak point is used as the pressure peak of the uterine cavity, the contrast value between unit time and the average peak value is used as the uterine contraction frequency, and the interval between the uterine contraction starting point and the uterine contraction ending point is used as the duration of uterine contraction.
After determining uterine cavity pressure, uterine contraction frequency and uterine contraction duration, distinguishing uterine contraction strength by using a K-means clustering method, wherein a K-means algorithm is firstly set as an initial central point of clustering (K is 3); then, the distances from other points to the k central points are calculated, and the points close to the other points are classified, wherein the distance calculation formula is as follows:
and finally, iteratively and successively updating the values of all the clustering centers by an averaging method until the values of all the central points are unchanged to obtain a uterine contraction signal after feature extraction, which is specifically shown in fig. 4.
Step S2, acquiring demographics of oxytocin dose adjustment, screening out characteristic variables with statistical significance, acquiring parturition physiological parameters output by a fetal heart monitor, and combining the characteristic variables with statistical significance and the parturition physiological parameters into data characteristic items;
the specific process comprises the steps of firstly, acquiring demographic data of oxytocin dose adjustment from a preset electronic medical record system, and screening out characteristic variables with statistical significance through single factor analysis and multiple linear stepwise regression; wherein the statistically significant characteristic variables include age, BMI, gestational week, fetal presenting, and uterine height;
secondly, acquiring physiological parameters during production from a data port of the fetal heart monitor; wherein the physiological parameters during labor production comprise fetal heart, uterine contraction frequency, uterine contraction duration, uterine contraction intensity, average value of uterine cavity pressure peak value in specified time (such as 15min), etc.;
finally, the characteristic variables with statistical significance and the physiological parameters during labor are combined into data characteristic items as shown in the following table 1.
TABLE 1
Step S3, according to the data feature items, combining the data corresponding to the demographic data of oxytocin dose adjustment and the time production physiological parameters to obtain initial sample data with the data feature items, and screening model input data in the initial sample data by combining the uterine contraction signals after feature extraction;
the specific process is that OT medication is adjusted once every 15min (namely the uterine contraction duration is 15min), and demographic data and intrapartum physiological parameters of the time period are integrated and taken as a recording sample to be included in the model input data. The missing value filling of human oral data such as age, BMI and the like can be realized by manually checking original data, all data are subjected to abnormal value processing by adopting a Rajda standard, namely, abnormal data larger than the value are deleted by taking a given confidence probability of 99.7% as a standard and taking 3 times of data row standard deviation as a basis.
S4, constructing a oxytocin dose prediction model during production based on a Bayesian optimization LightGBM algorithm; the oxytocin dose prediction model during production takes the model input data as data input and takes the oxytocin dripping speed change condition as target output;
the LightGBM is a fast, distributed and high-performance gradient lifting framework based on a decision tree algorithm, and is widely applied to application scenarios of classification, regression and other medical data analysis of various obstetrical departments, such as early screening of gestational diabetes and prediction of fetal birth weight.
The core idea of the gradient lifting tree is to utilize the negative gradient of the loss function in the current model Fm(x)＝Fm-1(x) The value of (d) approximately replaces the residual. Let i (i ═ 1,2,3 …, n), the number of iterations j (j ═ 1,2,3 …, m), and the loss function be L (y)i,F(xi) Then negative gradient r)ijIs calculated byThe following were used:
use of the base learning machine hj(x) Fitting the negative gradient r of the loss function, finding the best fit value that minimizes the loss function:
thus, model updating is carried out, and the strong learner of the round is as follows:
Fj(x)＝Fj-1(x)+rjhj(xi) (5)
the final gradient lifting tree is obtained by linear addition of the base learners generated in each round:
the LightGBM is used for optimizing a traditional Gradient spanning tree, the number of training samples is reduced based on a One-sided Gradient Sampling (GOSS) algorithm, the Feature dimension is reduced by adopting a complementary Feature compression (EFB) algorithm, and GPU parallel learning is supported.
The GOSS algorithm can keep the accuracy of information gain estimation, the information gain is measured by using the variance gain after splitting, only those samples with larger contribution are kept, and the variance gain V isj(d) The following were used:
where d is the splitting point of the sample feature, n represents the number of samples, A, B represents the split large and small gradient samples, l, r represent the left and right subtrees, and g represents the sample gradient as follows:
the EFB algorithm can reduce the feature number of high-dimensional data and minimize loss, discretizes a continuous floating point type feature value by adopting a Histogram (Histogram) algorithm, accumulates statistics in the Histogram according to the discretized value as an index, and reduces the number of features by binding and combining mutually independent features, thereby reducing the memory occupation and time complexity.
Under the background of big data, obstetrical clinical data is continuously and explosively increased, the LightGBM has more excellent processing capability when facing complex and diversified big data, the information resource occupancy rate is low, and the operation is efficient, so the LightGBM is supposed to be selected to construct a prediction model.
The Bayesian optimization algorithm is an efficient parameter optimization algorithm, and can be uniformly described as a global optimal solution for solving an unknown objective function:
wherein X represents a parameter to be optimized, and X represents a parameter set; f (x) represents an objective function.
The Bayesian optimization is mainly divided into two steps. Firstly, determining a prior function to express the distribution assumption of an optimized function, and generally selecting a Gaussian process (Gaussian Processes) with flexibility and traceability; next, an extraction Function (Acquisition Function) is selected, and a utility Function is constructed in the model posterior distribution to determine the next point to be evaluated.
The gaussian process, here the extension of different parameter combinations in the LightGBM algorithm on a random process, consists of a mean function m (x) and a semi-positive covariance function k (x, x'):
m(x)＝E[f(x)] (9)
k(x,x')＝E[(f(x)-m(x))(f(x')-m(x'))] (10)
for ease of calculation, typically taking the mean function m (x) to 0, then there is a gaussian distribution satisfying:
f(x)～GP(m(x),k(x,x'))→f～N(0,K(X,X)) (11)
when a new sample X is added*A new gaussian distribution can be generated using the covariance matrix K (X, X):
f*～N(0,K(X*,X*)) (13)
the nature of the Gaussian process is available, training output f and test output f*Is jointly distributed as
By evaluating the mean and covariance matrix, the function value f can be sampled from the combined posterior distribution, and the next point to be evaluated is determined by the sampling function, so that the iteration times can be reduced, and the evaluation cost can be reduced. The sampling point extraction method comprises the following two methods: (1) and (5) expore, selecting an unexevated parameter combination (2) expoil as far as possible, and searching around the currently found optimal value to find a global optimal value according to the currently found optimal value.
Common sampling functions in Bayesian optimization include probability of improvement (POI), Expected Improvement (EI) and Upper Confidence Bound (UCB), the UCB is selected as the sampling function, and the expression is as follows:
UBX(x)＝μ(x)+βσ(x) (15)
μ (x) and σ (x) are the mean and standard deviation of the combined posterior distribution of the objective function obtained using the gaussian process, and β is a suitable constant for balancing exploration and development. The UCB sampling function directly compares the maximum value of the confidence interval, and the prediction error of the UCB sampling function is proved to have a sub-linear relationship with the iteration number, so that the effect is generally better.
And S5, importing the model input data into the intermediate oxytocin dose prediction model for calculation to obtain a numerical value corresponding to the final oxytocin dropping speed change condition.
The specific process is that model input data is used as input, and the model is imported into a oxytocin dose prediction model for calculation to obtain a numerical value corresponding to the final oxytocin dripping speed change condition; wherein, the change condition of the dropping speed of the oxytocin comprises maintaining the original dropping speed, slowing down the dropping speed and accelerating the dropping speed; the values corresponding to the change of the oxytocin dripping speed include a value 3 corresponding to the original dripping speed, a value 2 corresponding to the slow dripping speed and a value 1 corresponding to the fast dripping speed, as shown in the following table 2.
TABLE 2
It should be noted that 4G internet technology is used as a carrier, fetal heart monitor, infusion injection pump and nurse station remote monitoring are connected, the fetal heart monitor is used for collecting the uterine contraction and fetal heart change information of the puerpera, the information is remotely fed back to the nurse station through the internet of things technology for data collection, and finally the nurse station sends out a regulation and control instruction to control the infusion micropump beside the puerpera bed and regulate and control the injection speed.
As shown in fig. 5 and fig. 6, an application scenario of the oxytocin dose intelligent regulation and control method based on uterine contraction signals and LightGBM provided by the embodiment of the present invention is further explained:
fig. 5 is a case analysis of the correlation between OT dose adjustment and uterine contraction duration, uterine cavity pressure, and uterine contraction frequency, wherein 1 subject is selected to continuously infuse OT during resting state, uterine contraction signals are acquired for about 8h, the abscissa represents the dropping speed adjustment times every 15min OT during the process, and the ordinate represents uterine contraction duration, uterine cavity pressure, and uterine contraction frequency, respectively. As can be seen from the graph, the OT dose is gradually increased from 0 to 9 times, and the corresponding uterine contraction frequency, uterine cavity pressure and uterine contraction duration are all at lower levels and have a rising trend; when the dosage is reduced for the 10 th time, the uterine contraction frequency, the uterine cavity pressure and the uterine contraction duration time have obvious descending trend; the OT reaches the highest dosage after 15-20 times, and the uterine contraction frequency, the uterine cavity pressure and the uterine contraction duration also have peaks. It can be seen that the characteristics of uterine contraction signal extraction herein are highly correlated with OT medication adjustment.
10061 OT regulation records are divided into 8049 training sets and 2012 testing sets according to the ratio of 8:2, and the 5-fold cross validation method is adopted for analysis and prediction, and the best parameters are adjusted by a Bayesian Optimization (BOA) method. LightGBM has more parameters and is manually operated by a cross validation method
The parameter adjustment is complicated and affects the precision, the LightGBM mainly includes three types of parameters, i.e., a Booster parameter, a general parameter, and a learning target parameter, table 3 sets the meaning and value range of each parameter as default, the iteration number is set to 200, the LightGBM finally selects a learning rate of 0.05, the leaf number is 51, the tree maximum depth is 5, min _ data _ in _ leaf is 20, the bagging _ fraction is 0.9, and the feature _ action is 0.8.
TABLE 3
The invention selects evaluation indexes commonly used for pattern recognition: the accuracy (accuracycacy), Precision (Precision), Recall (Recall) and F1 values evaluated the performance of each model, and the formula for the 4 metric values is as follows. In addition, a confusion matrix is also used to observe the behavior of the model on each category.
Meanwhile, the method selects 5 machine learning algorithms including Logistic Regression (LR), Support Vector Machine (SVM), multilayer perceptron (MLP), Random Forest (RF) and gradient lifting tree (GBDT) for comparison, and all the algorithms are as follows: and 2, dividing the data set into a training set and a testing set, and comparing experimental results through 5-fold cross validation. Compared with the LightGBM adjusted by the cross validation and trial and error method, the BOA-LightGBM model optimized by the bayes has the best prediction capability on OT dropping rate regulation, the accuracy rate is 0.83, the precision rate is 0.853, the recall rate is 0.828, and the F1 value is 0.84, which are all superior to those of other comparison machine methods, as shown in the following table 4.
TABLE 4
The confusion matrix for categorizing the predicted results is shown in table 5: columns represent real categories, rows represent prediction categories, a confusion matrix indicates that the LightGBM model can effectively predict 'reduced dose', the accuracy can reach 99%, errors are concentrated in prediction of 'dripping speed acceleration' and 'dripping speed maintenance', and the main reason is that the clinical decision on 'dripping speed acceleration' of a puerpera is conservative, and the OT dosage needs to be carefully controlled to avoid potential adverse reactions; in addition, OT regulation is also affected by maternal complaints (generally, the speed of drip is not accelerated in response to the complaints) and subjective factors of medical staff (character, clinical experience, medication habits), and therefore, the distinction between "acceleration of drip rate" and "maintenance of drip rate" is weak.
TABLE 5
The LightGBM feature importance ranking is as shown in fig. 6, and by outputting the feature importance degree, it can be determined which factors have a more significant influence on OT regulation, thereby assisting medical personnel in making medication decisions. The uterine contraction frequency, uterine contraction duration, uterine cavity pressure, uterine contraction strength and fetal heart are the most important characteristics in the final model result, and are consistent with indexes of OT regulation and guidance observation of clinical practical medical staff [2], which indicates that the model highly simulates clinical thinking and enables a modeling result to have higher reliability. The medication time, the current dosage, the age of the lying-in woman, the BMI, the labor analgesia, the fetal presenting fluid, the cesarean section history and the vaginal labor history are influence factors influencing the OT infusion dosage, the data comprehensively reflect the state of the lying-in woman during the labor, and guarantee is provided for the model precision. At present, OT infusion is still manually adjusted by medical staff through electronic fetal monitoring mother and child conditions, but due to numerous influencing factors, complex relation and insufficient intuition, the OT infusion is difficult for medical staff with clinical experience, and the OT infusion is difficult to control. Therefore, the model can be used as an auxiliary decision system for an OT medication adjustment scheme during the production of medical staff, plays a role in replacing or lightening the medical staff in OT infusion, realizes accurate medication and simultaneously ensures the safety of medication.
Therefore, compared with the traditional OT infusion scheme, the LightGBM model based on bayesian optimization has both accurate regulation and control and medication safety: the OT infusion regulation and control is actually real-time feedback regulation, at present, medical staff are required to manually regulate the dosage of the medicine at intervals of 15-20min according to fetal heart monitoring information, the OT clinical use demand is large, the requirements on clinical experience of the medical staff are high, observation indexes are complex, the existing OT infusion feedback system does not solve the problems of recognition and algorithm research of fetal heart uterine contraction monitoring information, and no clinical real data are used as training samples. The model firstly extracts the characteristics of the uterine contraction signal, carries out model training based on clinical OT real medication condition, has higher accuracy, is used as an auxiliary decision-making system of obstetrical medical personnel, innovates the current domestic mode for OT infusion regulation and control, carries out OT infusion in future research or to replace midwife, and saves a large amount of resources for hospitals
As shown in fig. 7, in an embodiment of the present invention, an intelligent oxytocin dose regulation and control system based on uterine contraction signals and LightGBM includes:
the signal processing and extracting unit 110 is configured to acquire a uterine contraction signal detected by the fetal heart monitor, preprocess the uterine contraction signal, and further perform feature extraction on the preprocessed uterine contraction signal based on the uterine contraction strength;
a data feature item forming unit 120, configured to obtain demographics data of oxytocin dose adjustment, screen out a feature variable with statistical significance, obtain an intrapartum physiological parameter output by a fetal heart monitor, and form a data feature item from the feature variable with statistical significance and the intrapartum physiological parameter;
a model input data screening unit 130, configured to combine, according to the data feature item, from the data corresponding to the adjusted demographics of the oxytocin dose and the time-production physiological parameters, initial sample data having the data feature item, and screen model input data in the initial sample data in combination with a uterine contraction signal after feature extraction;
the prediction model construction unit 140 is configured to construct a oxytocin dose prediction model during production based on a bayesian-optimized LightGBM algorithm; the oxytocin dose prediction model during production takes the model input data as data input and takes the oxytocin dripping speed change condition as target output;
and the prediction model result output unit 150 is used for importing the model input data into the oxytocin dose prediction model for calculation to obtain a value corresponding to the final oxytocin dropping speed change condition.
Wherein the signal processing and extracting unit 110 includes:
a signal acquisition module 1101, configured to acquire a uterine contraction signal detected by the fetal heart monitor through the pressure sensor;
a signal filtering processing module 1102, configured to perform filtering processing on the uterine contraction signal based on a signal smoothing processing mechanism of wavelet filtering;
and a clustering feature extraction module 1103, configured to determine uterine cavity pressure, uterine contraction frequency, and uterine contraction duration in the filtered uterine contraction signal, and then distinguish uterine contraction strength by using a K-means clustering method to obtain a uterine contraction signal after feature extraction.
Wherein the data feature item forming unit 120 includes:
a first variable obtaining module 1201, configured to obtain oxytocin dose-adjusted demographic information from a preset electronic medical record system, and screen out a characteristic variable having statistical significance through single factor analysis and multiple linear stepwise regression; wherein said statistically significant characteristic variables include age, BMI, gestational week, fetal presenting fluid, and uterine height;
a second variable obtaining module 1202, configured to obtain a physiological parameter during labor production from the data port of the fetal heart monitor; wherein the intrapartum physiological parameters comprise the fetal heart, the uterine contraction frequency, the uterine contraction duration, the uterine contraction intensity and the average value of the uterine cavity pressure peak value in a specified time;
and the data feature item combination module 1203 is configured to combine the feature variable with statistical significance and the labor-hour physiological parameter into a data feature item.
Wherein the oxytocin dripping speed change condition comprises that the dripping speed maintains the original speed, the dripping speed is slowed down and the dripping speed is accelerated; the numerical values corresponding to the oxytocin dripping speed change conditions comprise a numerical value 3 corresponding to the dripping speed maintaining original speed, a numerical value 2 corresponding to the dripping speed slowing down and a numerical value 1 corresponding to the dripping speed speeding up.
The embodiment of the invention has the following beneficial effects:
1. the invention provides a Bayesian optimization-based LightGBM to establish an OT dripping speed prediction model, which has high training speed and high model precision and overcomes the contradiction between dependence of a traditional infusion scheme on personal experience and shortage of human resources of medical staff;
2. compared with the traditional OT infusion scheme, the LightGBM model based on Bayesian optimization has the advantages of accurate regulation and control, safe medication and higher accuracy, is used as an auxiliary decision system for obstetrical medical personnel, innovates the current domestic mode for regulating and controlling OT infusion, and saves a large amount of resources for hospitals in future research or replacement of obstetrical departments for OT infusion.
It should be noted that, in the above system embodiment, the included components are merely divided according to functional logic, but are not limited to the above division as long as the corresponding functions can be implemented; in addition, specific names of the functional units are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present invention.
It will be understood by those skilled in the art that all or part of the steps in the method for implementing the above embodiments may be implemented by relevant hardware instructed by a program, and the program may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc.
The above disclosure is only for the purpose of illustrating the preferred embodiments of the present invention, and it is therefore to be understood that the invention is not limited by the scope of the appended claims.
Claims (8)
1. An oxytocin dose intelligent regulation and control method based on uterine contraction signals and LightGBM is characterized by comprising the following steps:
s1, acquiring a uterine contraction signal detected by the fetal heart monitor, preprocessing the uterine contraction signal, and further extracting the characteristics of the preprocessed uterine contraction signal based on the uterine contraction strength;
s2, acquiring demographics of oxytocin dose adjustment, screening out characteristic variables with statistical significance, acquiring parturition physiological parameters output by a fetal heart monitor, and forming data characteristic items by the characteristic variables with statistical significance and the parturition physiological parameters;
s3, according to the data feature items, combining the data corresponding to the demographic data of oxytocin dose adjustment and the time production physiological parameters to obtain initial sample data with the data feature items, and screening model input data in the initial sample data by combining the uterine contraction signals after feature extraction;
s4, constructing a oxytocin dose prediction model during production based on a Bayesian optimization LightGBM algorithm; the oxytocin dose prediction model during production takes the model input data as data input and takes the oxytocin dripping speed change condition as target output;
and S5, importing the model input data into the intermediate oxytocin dose prediction model for calculation to obtain a numerical value corresponding to the final oxytocin dropping speed change condition.
2. The method for intelligently controlling oxytocin dose based on uterine contraction signals and LightGBM as claimed in claim 1, wherein the step S1 specifically comprises:
acquiring a uterine contraction signal detected by the fetal heart monitor through a pressure sensor;
based on a signal smoothing mechanism of wavelet filtering, filtering the uterine contraction signal;
and after determining uterine cavity pressure, uterine contraction frequency and uterine contraction duration in the uterine contraction signals after filtering, distinguishing uterine contraction strength by using a K-means clustering method to obtain uterine contraction signals after feature extraction.
3. The method for intelligently controlling oxytocin dose based on uterine contraction signals and LightGBM as claimed in claim 1, wherein the step S2 specifically comprises:
acquiring demographic data of oxytocin dose adjustment from a preset electronic medical record system, and screening out characteristic variables with statistical significance through single factor analysis and multiple linear stepwise regression; wherein said statistically significant characteristic variables include age, BMI, gestational week, fetal presenting fluid, and uterine height;
acquiring physiological parameters during production from the data port of the fetal heart monitor; wherein the intrapartum physiological parameters comprise the fetal heart, the uterine contraction frequency, the uterine contraction duration, the uterine contraction intensity and the average value of the uterine cavity pressure peak value in a specified time;
and forming the characteristic variables with statistical significance and the labor-hour physiological parameters into data characteristic items.
4. The intelligent oxytocin dose regulation and control method based on uterine contraction signals and LightGBM as claimed in claim 1, wherein in step S5, the oxytocin drip speed change condition comprises that the drip speed maintains the original speed, the drip speed is slowed down and the drip speed is accelerated; the numerical values corresponding to the oxytocin dripping speed change conditions comprise a numerical value 3 corresponding to the dripping speed maintaining original speed, a numerical value 2 corresponding to the dripping speed slowing down and a numerical value 1 corresponding to the dripping speed speeding up.
5. An oxytocin dose intelligent regulation and control system based on uterine contraction signals and LightGBM, comprising:
the signal processing and extracting unit is used for acquiring a uterine contraction signal detected by the fetal heart monitor, preprocessing the uterine contraction signal and further extracting the characteristics of the preprocessed uterine contraction signal based on the uterine contraction strength;
the data characteristic item forming unit is used for acquiring demographics data of oxytocin dose adjustment, screening out characteristic variables with statistical significance, acquiring intrapartum physiological parameters output by a fetal heart monitor, and forming the characteristic variables with the statistical significance and the intrapartum physiological parameters into data characteristic items;
the model input data screening unit is used for combining the data corresponding to the adjusted demographics data of the oxytocin dose and the time production physiological parameters to obtain initial sample data with the data characteristic items according to the data characteristic items, and screening model input data from the initial sample data by combining the uterine contraction signals after characteristic extraction;
the prediction model construction unit is used for constructing a oxytocin dose prediction model during production based on a Bayesian optimization LightGBM algorithm; the oxytocin dose prediction model during production takes the model input data as data input and takes the oxytocin dripping speed change condition as target output;
and the prediction model result output unit is used for importing the model input data into the oxytocin dose prediction model during production to calculate, and obtaining a numerical value corresponding to the final oxytocin dropping speed change condition.
6. The intelligent oxytocin dose regulation system according to claim 5, characterized in that the signal processing and extraction unit comprises:
the signal acquisition module is used for acquiring a uterine contraction signal detected by the fetal heart monitor through the pressure sensor;
the signal filtering processing module is used for carrying out filtering processing on the uterine contraction signal based on a signal smoothing processing mechanism of wavelet filtering;
and the clustering feature extraction module is used for distinguishing the uterine contraction strength by using a K-means clustering method after determining the uterine cavity pressure, the uterine contraction frequency and the uterine contraction duration time in the uterine contraction signals after the filtering processing so as to obtain the uterine contraction signals after feature extraction.
7. The intelligent oxytocin dose regulation system based on uterine contraction signals and LightGBM as claimed in claim 5, wherein the data feature item forming unit comprises:
the first variable acquisition module is used for acquiring demographics data of oxytocin dose adjustment from a preset electronic medical record system and screening out characteristic variables with statistical significance through single factor analysis and multiple linear stepwise regression; wherein said statistically significant characteristic variables include age, BMI, gestational week, fetal presenting fluid, and uterine height;
the second variable acquisition module is used for acquiring physiological parameters during production from the data port of the fetal heart monitor; wherein the intrapartum physiological parameters comprise the fetal heart, the uterine contraction frequency, the uterine contraction duration, the uterine contraction intensity and the average value of the uterine cavity pressure peak value in a specified time;
and the data feature item combination module is used for combining the feature variable with statistical significance and the labor-producing physiological parameter into a data feature item.
8. The intelligent oxytocin dose regulation system based on a uterine contraction signal and a LightGBM as claimed in claim 5, wherein the oxytocin drip rate change includes a drip rate maintenance original rate, a drip rate slowing down and a drip rate speeding up; the numerical values corresponding to the oxytocin dripping speed change conditions comprise a numerical value 3 corresponding to the dripping speed maintaining original speed, a numerical value 2 corresponding to the dripping speed slowing down and a numerical value 1 corresponding to the dripping speed speeding up.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011110757.8A CN112397176A (en) | 2020-10-16 | 2020-10-16 | Intelligent oxytocin dose regulation and control method and system based on uterine contraction signals and LightGBM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011110757.8A CN112397176A (en) | 2020-10-16 | 2020-10-16 | Intelligent oxytocin dose regulation and control method and system based on uterine contraction signals and LightGBM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112397176A true CN112397176A (en) | 2021-02-23 |
Family
ID=74596649
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011110757.8A Pending CN112397176A (en) | 2020-10-16 | 2020-10-16 | Intelligent oxytocin dose regulation and control method and system based on uterine contraction signals and LightGBM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112397176A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113257422A (en) * | 2021-06-04 | 2021-08-13 | 福州大学 | Method and system for constructing disease prediction model based on glucose metabolism data |
| CN113762600A (en) * | 2021-08-12 | 2021-12-07 | 北京市燃气集团有限责任公司 | LightGBM-based monthly gas consumption prediction method and device |
| CN117649950A (en) * | 2024-01-29 | 2024-03-05 | 北京大学第三医院（北京大学第三临床医学院） | Oxytocin pharmacokinetics model, and construction method and application thereof |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014035672A2 (en) * | 2012-08-30 | 2014-03-06 | Medtronic Minimed, Inc. | Safeguarding techniques for a closed-loop insulin infusion system |
| CN206434657U (en) * | 2016-11-18 | 2017-08-25 | 温州市人民医院 | Oxytocin injections self-checking device |
| CN107349489A (en) * | 2017-08-04 | 2017-11-17 | 佛山科学技术学院 | A kind of transfusion drip speed based on neutral net determines method and system |
| CN107854748A (en) * | 2017-10-18 | 2018-03-30 | 温州医科大学 | A kind of oxytocin syringe pump reponse system |
| CN110547868A (en) * | 2019-09-05 | 2019-12-10 | 栾其友 | Continuous anesthesia device |
-
2020
- 2020-10-16 CN CN202011110757.8A patent/CN112397176A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014035672A2 (en) * | 2012-08-30 | 2014-03-06 | Medtronic Minimed, Inc. | Safeguarding techniques for a closed-loop insulin infusion system |
| CN206434657U (en) * | 2016-11-18 | 2017-08-25 | 温州市人民医院 | Oxytocin injections self-checking device |
| CN107349489A (en) * | 2017-08-04 | 2017-11-17 | 佛山科学技术学院 | A kind of transfusion drip speed based on neutral net determines method and system |
| CN107854748A (en) * | 2017-10-18 | 2018-03-30 | 温州医科大学 | A kind of oxytocin syringe pump reponse system |
| CN110547868A (en) * | 2019-09-05 | 2019-12-10 | 栾其友 | Continuous anesthesia device |
Non-Patent Citations (1)
| Title |
|---|
| 邬天骥: "基于机器学习的数据驱动故障诊断方法研究", 《中国优秀博硕士学位论文全文数据库(硕士) 工程科技Ⅱ辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113257422A (en) * | 2021-06-04 | 2021-08-13 | 福州大学 | Method and system for constructing disease prediction model based on glucose metabolism data |
| CN113762600A (en) * | 2021-08-12 | 2021-12-07 | 北京市燃气集团有限责任公司 | LightGBM-based monthly gas consumption prediction method and device |
| CN117649950A (en) * | 2024-01-29 | 2024-03-05 | 北京大学第三医院（北京大学第三临床医学院） | Oxytocin pharmacokinetics model, and construction method and application thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cömert et al. | Fetal hypoxia detection based on deep convolutional neural network with transfer learning approach | |
| CN112397176A (en) | Intelligent oxytocin dose regulation and control method and system based on uterine contraction signals and LightGBM | |
| Georgieva et al. | Artificial neural networks applied to fetal monitoring in labour | |
| Spilka et al. | Sparse support vector machine for intrapartum fetal heart rate classification | |
| EP0645035B1 (en) | Medical signal analyzer | |
| US6254537B1 (en) | Fetal outcome predictor and monitoring system | |
| CN109924965A (en) | A kind of cardiovascular and cerebrovascular rehabilitation training system | |
| US20140378855A1 (en) | Apparatus and method for feature extraction and classification of fetal heart rate | |
| Arif et al. | Decision tree method using for fetal state classification from cardiotography data | |
| Liang et al. | Automatic evaluation of fetal heart rate based on deep learning | |
| Jezewski et al. | Analysis of extracted cardiotocographic signal features to improve automated prediction of fetal outcome | |
| Allahem et al. | Automated labour detection framework to monitor pregnant women with a high risk of premature labour using machine learning and deep learning | |
| Peterek et al. | Human fetus health classification on cardiotocographic data using random forests | |
| Lesmana et al. | Sleep stage identification using the combination of ELM and PSO based on ECG signal and HRV | |
| Veena et al. | An automated pre-term prediction system using EHG signal with the aid of deep learning technique | |
| Robu et al. | The analysis and classification of birth data | |
| Spilka et al. | Intrapartum fetal heart rate classification: Cross-database evaluation | |
| Nsugbe et al. | On the prediction of premature births in Hispanic labour patients using uterine contractions, heart beat signals and prediction machines | |
| Magenes et al. | Identification of fetal sufferance antepartum through a multiparametric analysis and a support vector machine | |
| Fergus et al. | Prediction of intrapartum hypoxia from cardiotocography data using machine learning | |
| Bahar et al. | Model Structure of Fetal Health Status Prediction | |
| Shruthi et al. | A Method for Predicting and Classifying Fetus Health Using Machine Learning | |
| Jansri et al. | Automatic sleep data scoring by artificial intelligence: A pilot study in Thai population | |
| CN112365952A (en) | Intelligent control method and system for oxytocin speed during production based on machine learning method | |
| Sridevi | Integrated UnderOversampling Responsive based Fetal Health Prediction using Cardiotocographic Data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210223 |