Parameter fusion-based labor process risk identification method and system

Abstract

The invention relates to a parturition risk identification method and system based on parameter fusion. The method comprises the following steps: acquiring a first clinical index; and constructing a static prediction model. Positioning a pelvis entrance plane and a birth canal space; the position of the fetal head is positioned through symphysis pubis ultrasound; evaluating a fetal head orientation; key skull mark points are marked on an ultrasonic interface, and a third clinical index is obtained; and determining a spatial geometrical relationship between the mark point and the medial axis of the birth canal, and obtaining a second clinical index. Pre-constructing a dynamic prediction model based on the first clinical index, the second clinical index and the third clinical index; obtaining a dynamic prediction probability by using the clinical indexes of the patient and the dynamic prediction model; and performing auxiliary early warning based on the dynamic prediction probability. According to the method, the static indexes and the dynamic indexes are fused, the method is suitable for continuous dynamic judgment, the method can be highly fused with the progress of the production process, combined use of a static early warning model and a dynamic early warning model is supported, and the sudden change trend of the risk can be captured.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of intelligent medical treatment, and particularly relates to a method and system for labor risk identification based on parameter fusion. BACKGROUND

[0002] At present, the judgment of labor risk still largely depends on the manual experience evaluation of medical staff, such as the comprehensive judgment by the subjective methods of palpation examination of fetal head position, observation of uterine contraction frequency, and combination of maternal clinical manifestations. This traditional mode has obvious limitations: the evaluation standard is not unified, and the experience difference of different doctors may lead to judgment deviation; the real-time performance is insufficient, the manual inspection cannot continuously monitor the dynamic changes, and it is difficult to timely capture the subtle deterioration trend of risk; the quantitative basis is missing, and the lack of objective data support makes the selection of early warning opportunity and intervention measures have certain randomness.

[0003] With the development of perinatal medicine, the demand for precise and standardized labor management in the clinic is increasing. Especially in complex cases, it is difficult to accurately predict the risk of dystocia by manual judgment, which may lead to delayed intervention or excessive medical treatment. Therefore, it is urgent to introduce a quantitative calculation model to assist clinical decision-making through objective data. For example, based on the dynamic parameters of fetal head descent speed and rotation angle measured by ultrasound, combined with the static characteristics of the mother, a real-time risk scoring system can be constructed to provide data-driven early warning support based on manual judgment. The rapid development and wide application of artificial intelligence technology are promoting society into a new era of intelligent and digital. With the continuous progress of technology and the strong support of policy, artificial intelligence is expected to play a key role in more fields and promote the high-quality development of the economy and society. The vertical application of artificial intelligence technology brings development opportunities to the medical field; with the development of intelligent medical technology, the obstetric field has higher requirements for accurate prediction and dynamic management of delivery risk. In addition, the development of modern intelligent medical technology provides new possibilities for labor monitoring. Through Internet of Things technologies such as wearable devices and intelligent ultrasound probes, data such as fetal heart rate, uterine contraction pressure, and fetal head movement trajectory can be collected in real time, and machine learning algorithms can be used to analyze risk trends. This human-machine collaborative mode can make up for the shortcomings of manual judgment, timely issue early warnings in the early stages of latent period extension and fetal head descent stagnation, and assist medical staff in developing more accurate intervention strategies, thereby optimizing the delivery outcome and improving maternal and infant safety.

[0004] Traditional labor risk assessment mainly relies on manual judgment or independent judgment of static clinical indicators, lacks multi-parameter fusion capability, and is difficult to integrate ultrasound, fetal heart monitoring, electronic medical records, and other multi-source heterogeneous data, limiting the accuracy and clinical applicability of the prediction model. Based on the above problems, the present application fuses static indicators with dynamic indicators, is suitable for continuous dynamic judgment, can be highly integrated with the progress of labor, supports the joint use of static early warning models and dynamic early warning models, and can capture the mutation trend of risk. SUMMARY

[0005] To solve the above problems in the prior art, the present application provides a labor risk identification method and system based on parameter fusion, which comprises: Step S1: obtaining first clinical indicators; the first clinical indicators are static indicators; screening significant indicators based on the first clinical indicators, and constructing a static prediction model; using patient clinical indicators and the static prediction model for auxiliary early warning to determine whether the dynamic prediction condition is met, yes, then entering step S2; the first clinical indicators include age, height, body weight at delivery, parity, mode of labor and fetal clinical estimated weight; determining whether the dynamic prediction condition is met, specifically: if the static prediction probability output by the static prediction model is greater than or equal to the preset first warning threshold, it is determined that the dynamic prediction condition is met; Step S2: positioning the pelvic inlet plane and the birth canal space; taking the fifth lumbar spine as the reference point, a three-dimensional pelvic coordinate system is established; using ultrasonic assisted spatial positioning technology, the specific anatomical landmark points of the pelvis are identified through a micro site sensor; Step S3: ultrasound positioning of fetal head position at pubic symphysis; determining the third clinical indicators; real-time measurement and display of the descent degree of the fetal head relative to the pelvic inlet plane and the birth canal, construction of the fetal head descent curve, and determination of the third clinical indicators; the third clinical indicators include fetal head-pubic symphysis distance, fetal head-sciatic spine relative position, pubic arch angle, presentation height change speed and pubic arch-fetal head angle; Step S4: assessing the fetal head orientation; marking key skull landmarks on the ultrasound interface; determining the spatial geometric relationship between the landmarks and the mid-axis of the birth canal to obtain the second clinical indicators; the second clinical indicators include fetal head rotation angle and fetal head entry into the pelvis position; Step S5: pre-constructing a dynamic prediction model based on the first clinical indicators, the second clinical indicators and the third clinical indicators; using patient clinical indicators and the dynamic prediction model to obtain a dynamic prediction probability ; based on the dynamic prediction probability for auxiliary early warning.

[0006] Further, the significant indicators are screened based on the first clinical indicators, and the static prediction model is constructed; specifically: single factor attribution analysis is directly performed on the first clinical indicators, and the first clinical indicators with P<0.05 are selected as significant indicators; a prediction model is constructed based on the logistic regression coefficients of R language, the discrimination of the prediction model is verified by area under the curve evaluation, the stability of the prediction model is verified by resampling, the net benefit threshold of the prediction model is determined by decision curve, and the baseline prediction model satisfying the discrimination, stability and net benefit threshold is obtained as shown in the following formulas (1) and (2); wherein: P represents the static prediction probability of cesarean section in pregnant women; is age, years, is parity, primipara = 0; multipara = 1, is interspinous diameter cm, is clinical estimated weight g, is height cm, is mode of labor, spontaneous labor = 0; induced labor = 1; the static probability can be obtained by inputting the significant indicators into the baseline prediction model ; (1); (2).

[0007] Further, the static prediction model is represented as a nomogram prediction model; specifically, the static prediction model is represented as a nomogram prediction model by R language to visualize display, the influence of each prediction variable on the prediction result is intuitively displayed in a graphical manner, so that rapid estimation is performed.

[0008] Further, the second clinical indicator is intercepted, specifically: for the fetal head position in the pelvis, the vertical distance between the lowest point of the fetal head and the horizontal level of the ischial spine is used as a continuous value representing the fetal head position in the pelvis, and the numerical range between -2 and 3 cm is intercepted; a negative value indicates that the fetal head has not reached the ischial spine, and a positive value indicates that it has passed; the fetal head position is set to be valued between -2 and 3 cm, when less than -2 cm, the value is set to -2 cm, and when greater than or equal to +3 cm, the value is set to +3 cm; the fetal head rotation angle is segmented and weighted; the fetal head rotation angle is segmented and weighted by the following formula (3); (3).

[0009] Further, the dynamic prediction model is pre-constructed based on the first clinical indicator, the second clinical indicator and the third clinical indicator; specifically: analyzing historical clinical data to obtain significant indicators in the first clinical indicator, the third clinical indicator and the second clinical indicator; pre-constructing a dynamic prediction model.

[0010] Further, the historical clinical data is analyzed to obtain significant indicators in the first clinical indicator, the third clinical indicator and the second clinical indicator, specifically: for the first clinical indicator and the second clinical indicator, single factor attribution analysis is directly performed, and the first clinical indicator and the second clinical indicator with P<0.05 are screened as significant indicators; for the third clinical indicator, the characteristic value of each third clinical indicator is calculated; a binary third combination is constructed by taking the characteristic value of any third clinical indicator and another arbitrary third clinical indicator; single factor attribution analysis is performed on each third clinical indicator, the characteristic value of the third clinical indicator and the third combination; P values are sorted from small to large, and P<0.05 is intercepted to form a third ordered sequence, and a significant indicator is selected based on the third ordered sequence.

[0011] A labor risk identification platform based on parameter fusion, the platform is used to realize the labor risk identification method based on parameter fusion.

[0012] A server for labor risk identification based on parameter fusion, comprising a processor, the processor and the memory are coupled, the memory stores program instructions, when the program instructions stored in the memory are executed by the processor, the labor risk identification method based on parameter fusion is realized.

[0013] A labor risk identification system based on parameter fusion, the system is used to realize the labor risk identification method based on parameter fusion.

[0014] A computer readable storage medium, comprising a program, when it runs on a computer, makes the computer execute the labor risk identification method based on parameter fusion.

[0015] The beneficial effects of the present application include: Through the static prediction model, multi-source data integration is carried out, an intuitive quantitative calculation method and a convenient nomogram visualization are provided, and clear early warning assistance can be given, compared with the risk score output by the black box AI model, the clinical interpretability is strong, and it is easy to trust and adopt; further, the static indicators and dynamic indicators are fused, which is suitable for continuous dynamic judgment, can be highly integrated with the progress of labor, supports the joint use of static early warning model and dynamic early warning model, and can capture the mutation trend of risk. (2) The dynamic prediction model supports time series analysis and dynamic probability gradient analysis, can identify early abnormal signals that are difficult to find through traditional manual observation, and the dynamic probability gradient can early warn the mutation of risk, such as the sudden drop in speed before the rotation stagnation of the fetal head in the second stage of labor, and has stronger time and data sensitivity. BRIEF DESCRIPTION OF DRAWINGS

[0016] The drawings described herein are used to provide further understanding of the present application, constitute a part of the present application, but do not constitute undue limitations on the present application, and in the drawings: Figure 1 The parameter fusion-based labor risk identification method provided by the present application is shown in the figure.

[0017] Figure 2 The nomogram prediction model provided by the present application is shown in the figure.

[0018] Figure 3 The prediction probability of the prediction model provided by the present application is shown in the figure.

[0019] Figure 4 The verification result of the prediction model provided by the present application is shown in the figure.

[0020] Figure 5 The net benefit of the prediction model provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0021] The present application will be described in detail below in conjunction with the drawings and specific embodiments, wherein the illustrative embodiments and descriptions are only used to explain the present application, but not as a limitation on the present application.

[0022] The present application proposes a parameter fusion-based labor risk identification method and system, as shown in the figure. Figure 1 The method comprises the following steps: Step S1: obtaining a first clinical indicator; the first clinical indicator is a static indicator; screening a significant indicator based on the first clinical indicator, and constructing a static prediction model; using the patient's clinical indicators and the static prediction model for auxiliary early warning, and determining whether the dynamic prediction condition is met, if yes, then entering step S2; wherein: the first clinical indicator includes maternal age, height, body weight at delivery, parity, mode of labor and fetal clinical estimated weight; auxiliary early warning is performed based on the static prediction probability; The significant indicator is screened based on the first clinical indicator, and the static prediction model is constructed; specifically: single factor attribution analysis is directly performed on the first clinical indicator, and the first clinical indicator with P<0.05 is selected as the significant indicator; a prediction model is constructed based on the R language logistic regression coefficient, the discrimination of the prediction model is verified by the area under the curve (AUC) evaluation, the stability of the prediction model is verified by Bootstrap resampling, the net benefit threshold of the prediction model is determined by decision curve analysis (DCA), and the baseline prediction model meeting the discrimination, stability and net benefit threshold is obtained as shown in the following formulas (1) and (2); wherein: The value represents the probability of the current observation result under the premise that the original hypothesis is true, that is, the static prediction probability of the pregnant woman undergoing cesarean section; is age (years), is parity (primipara = 0; multipara = 1), is interspinous diameter (cm), is clinical estimated weight (g), is height (cm), is mode of labor (spontaneous labor = 0; induced labor = 1); the significant indicators are input into the baseline prediction model to obtain a static static probability ; (1); (2); The judgment whether the dynamic prediction condition is met is specifically: if the static prediction probability output by the static prediction model is , it is determined that the dynamic prediction condition is met; Alternatively: when any one of conditions 1-5 and any one of conditions 6-8 are met, it is determined that the dynamic prediction condition is met; obviously, conditions 1-8 are extensible, wherein, conditions 1-5 are static decision conditions, and conditions 6-8 are dynamic decision conditions; Condition 1: the static prediction probability predicted by the static model r>30%; Condition 2: ≥35 or ≤18; Condition 3: anteroposterior diameter of pelvic inlet <10 cm; Condition 4: g or g; Condition 5: previous cesarean section history has the risk of uterine rupture; Condition 6: for the first stage of labor, primipara >12h, multipara >8h; or for the second stage of labor, primipara >2h, multipara >1h; Condition 7: contraction frequency / intensity <3 times / 10 minutes or >7 times / 10 minutes; Condition 8: fetal heart rate variability; As shown in the accompanying Figure 2 , the step S1 further includes a step S1X: representing the static prediction model as a nomogram prediction model; specifically: representing the static prediction model as a nomogram prediction model through R language to visualize display, intuitively displaying the influence of each prediction variable on the prediction result through graphical display, so as to quickly estimate; Figure 2In each significant indicator, each significant indicator corresponds to a scale axis, and the numerical range reflects the actual distribution of the variable; the uppermost side is the score axis, and the vertical line is drawn on each variable axis according to the individual variable value, and the "contribution score" of the variable is obtained by intersecting the score axis; the lower side is the total score axis, and the scores of all variables are added to correspond to the total score axis; the total score is directly mapped to the prediction probability axis, and finally to the static prediction probability; the nomogram prediction model is convenient for clinical rapid judgment; with the diversity and popularity of detection means, more useful information can be obtained in clinical practice, so more accurate judgments can be made through dynamic factors on the basis of the static prediction model; Step S2: positioning the pelvic inlet plane and the birth canal space; Specifically: taking the fifth lumbar spine as the reference point, a three-dimensional pelvic coordinate system is established to realize the digital reconstruction of the anatomical structure; using ultrasound-assisted spatial positioning technology, specific anatomical landmark points of the pelvis are identified by a micro-site sensor; taking the fifth lumbar spine as the reference point, its anatomical position is stable and easy to identify, at this time, in the horizontal direction, connecting the two sides of the anterior superior iliac spine, pointing to the right side is positive, in the sagittal direction, pointing to the pubic symphysis is positive, and in the vertical direction, pointing to the head side is positive; at least the following markers are marked by ultrasound / sensor: marking the upper edge of the pubic symphysis, that is, the highest point of the pubic symphysis is located by the ultrasound transverse section, defining the front boundary of the pelvic inlet; the sacral promontory, that is, the most forward convex point of the sacrum is located by the ultrasound sagittal section, defining the posterior boundary of the pelvic inlet; the ischial spine, that is, the tip of the ischial spine is located by the ultrasound oblique section, serving as a reference for the narrow part of the birth canal in the pelvis; the lower edge of the pubic arch, that is, the lowest point of the pubic symphysis is located by the ultrasound coronal section, used for evaluating the outlet plane; Step S3: ultrasound positioning of the fetal head position on the pubic symphysis; determining the third clinical indicator; automatically / manually marking the upper and lower edges of the pubic symphysis and the intersection of the birth canal and the fetal head based on specific anatomical landmark points of the pelvis; real-time measurement and display of the degree of fetal head descent relative to the pelvic inlet plane and the birth canal, construction of a fetal head descent curve, and determination of the third clinical indicator; the third clinical indicator is a rapid dynamic indicator; as the labor progresses, the third clinical indicator changes relatively faster than the second clinical indicator; wherein: the third clinical indicator includes fetal head-pubic symphysis distance, fetal head-ischial spine relative position, pubic arch angle, speed of change of presentation height, pubic arch-fetal head angle, etc.; Preferably: using the LaborPro system to measure and display the degree of fetal head descent relative to the pelvic inlet plane and the birth canal, and constructing a fetal head descent curve with millimeter-level changes to determine the third clinical indicator; Step S4: evaluating the fetal head position; specifically: marking key skull landmarks on the ultrasound interface; analyzing the spatial geometric relationship between the landmarks and the mid-axis in the birth canal to obtain a second clinical indicator; the second clinical indicator is a slow dynamic indicator; segmenting and / or intercepting the second clinical indicator; wherein: the second clinical indicator includes fetal head rotation angle and fetal head entry into the basin position, etc.; as the labor progresses, the above-mentioned second clinical indicator and third clinical indicator need to be obtained in real time as needed, and considering that the two indicators change at different speeds, the time interval of real-time detection can be adjusted according to the situation; The intercepting processing of the second clinical indicator is specifically: for the fetal head entry into the basin position, the vertical distance between the fetal head lowest point and the ischial spine level is used as a continuous value representing the fetal head entry into the basin position, and the numerical range between [-2, 3] is intercepted; a negative value indicates that the fetal head has not reached the ischial spine, and a positive value indicates that it has passed; the fetal head entry into the basin position is set to be valued between [-2, 3] (cm), when less than -2cm, the value is set to -2cm, and when greater than or equal to +3cm, the value is set to +3cm; the fetal head rotation angle is segmented and weighted; the fetal head rotation angle is segmented and weighted by using the following formula (3) ; (3) ; The non-linear weighting method can further be used to make the dynamic prediction model capture the non-linear relationship between the entry into the basin depth and the risk, and the risk changes expressed from -2cm to 0cm and from 0cm to +2cm are different; Generally, the fetal head entry into the basin position includes anterior position, transverse position, and posterior position, and when performing quantitative calculation, this quantitative method loses the accurate data characteristics and cannot effectively express the information in subsequent prediction; and the biparietal diameter, the occipitofrontal diameter, the cervical spine midline, or the eye socket of the fetus need to be marked on the ultrasound interface; Step S5: pre-building a dynamic prediction model based on the first clinical indicator, the second clinical indicator, and the third clinical indicator; using the patient's clinical indicators and the dynamic prediction model to obtain a dynamic prediction probability ; based on the dynamic prediction probability for auxiliary early warning; in fact, after analyzing the change characteristics of the clinical indicators, it is found that through feature combination and feature value of the indicators, the accuracy of the prediction model can be further improved under certain conditions; The pre-building of the dynamic prediction model based on the first clinical indicator, the second clinical indicator, and the third clinical indicator is specifically: analyzing historical clinical data to obtain significant indicators in the first clinical indicator, the third clinical indicator, and the second clinical indicator; pre-building a dynamic prediction model; The analysis history clinical data to obtain the first clinical indicators, third clinical indicators and significant indicators in the second clinical indicators, specifically: for the first clinical indicators and the second clinical indicators, directly performing single factor attribution analysis, screening the first clinical indicators and the second clinical indicators with P<0.05 as significant indicators; for the third clinical indicators, the characteristic value of each third clinical indicator is calculated; the characteristic value of any third clinical indicator and another arbitrary third clinical indicator is constructed into a binary third combination; single factor attribution analysis is performed on each third clinical indicator, the characteristic value of the third clinical indicator and the third combination; according to the size of the P value obtained by analysis, from small to large, and the P<0.05 are intercepted to constitute a third ordered sequence, and the significant indicators are selected based on the third ordered sequence; the P value represents the probability of the current observation result under the premise that the original hypothesis is true; The third ordered sequence is selected based on the third ordered sequence, specifically: initializing the third significant indicator set to be empty; selecting an element from the head of the third ordered sequence; judging whether the element exists in the third significant indicator set or in any element of the third significant indicator set (when the element is a third combination, it may exist as an element in the combination); if yes, continue to select the next element and repeat the above steps; otherwise, put the element into the third significant indicator set; repeat the above steps until the tail of the third ordered sequence; Preferably, the characteristic value is a variable capable of indicating the change amount of the third clinical indicator, such as: mean, acceleration, change rate, gradient value, etc. The dynamic prediction model is constructed in advance; specifically: based on the R language logistic regression coefficient, the dynamic prediction model is constructed, the discrimination of the prediction model is verified by the area under the curve (AUC) evaluation, the stability of the prediction model is verified by Bootstrap resampling internal verification, the net benefit threshold of the prediction model is determined by decision curve analysis (DCA), and the dynamic prediction model satisfying the discrimination, stability and net benefit is obtained as shown in the following formula (4) (5); wherein: is age (years old), is parity (primipara = 0; multipara = 1), is the interspinous diameter (cm), is the clinical estimated weight (g), is the height (cm), is the mode of labor (natural labor = 0; induced labor = 1); is the fetal head rotation angle, is the fetal head position in the pelvis, is the pubic arch angle, is the change speed of the presentation height; d (4); (5); the use of patient clinical indicators and dynamic prediction model to obtain dynamic prediction probability ; Specifically: input the current patient clinical indicators into the dynamic prediction model to obtain the dynamic prediction probability corresponding to the current patient , if the dynamic prediction probability is greater than the first warning probability threshold, clinical warning is performed; Considering the continuity of the judgment, considering the time sequence information of the prediction result can consider the entire labor process as a complete conversation, and the same pregnant woman is considered as a whole and continuously, and the dynamic prediction based on the time sequence is replaced; In step S5, the dynamic prediction model is constructed as shown in formula (6) based on the time sequence dynamic prediction model; And use the dynamic prediction model as shown in formula (6) to obtain the time sequence dynamic probability ; Wherein: and 2 is the time sequence adjustment coefficient, is the dynamic prediction probability obtained by the dynamic prediction model at the t detection time; t is the detection time; T is the maximum detection time, t=1~T, which constitutes the length of the nearest T detection time; At this time, the nearest detection t=T; The farthest detection t=1; Use patient clinical indicators and time sequence based dynamic prediction model to obtain time sequence dynamic probability ; (6); Preferably: ; For example: =0.1, ; the use of patient clinical indicators and dynamic prediction model to obtain dynamic prediction probability ; Specifically: input the current patient clinical indicators into the dynamic prediction model to obtain the dynamic prediction probability corresponding to the current patient , if the dynamic prediction probability is greater than the first warning probability threshold, clinical warning is performed; Preferably: simultaneously use static prediction probability, dynamic prediction probability and / or time sequence dynamic probability to assist, when it is greater than the first warning probability threshold (for example: 50%), clinical warning is performed; Further: calculate the time sequence dynamic probability gradient value ​When the time-series dynamic probability is greater than the first warning probability threshold and the time-series dynamic probability gradient is greater than the first warning gradient threshold, a clinical warning is issued; where: the first warning probability threshold and the first warning gradient threshold are preset values; for example: the first warning gradient threshold is equal to 40%, and the first warning gradient threshold is equal to 0.05; when using the time-series dynamic probability gradient and the predicted value of the prediction model for joint warning, the threshold of the first warning gradient threshold can be appropriately reduced; The calculation of the time-series dynamic probability gradient value Specifically: ; When clinical indicators change rapidly during labor, setting a time-sensitive gradient can help detect risks more quickly. Therefore, an alternative is to calculate the time-series dynamic time gradient value. When the time-series dynamic probability is greater than the first warning probability threshold and the time-series dynamic time gradient value is greater than the first warning time gradient threshold, a clinical warning is issued; where: the first time warning gradient threshold is a preset value; for example: the first warning gradient threshold is equal to 0.05h; The calculation of the time-series dynamic time gradient value Specifically: ; It is the detection time. and detection time The length of time between; The combined time-series dynamic probability gradient and time-series dynamic temporal gradient values ​​can simultaneously provide both numerically sensitive and time-sensitive early warnings; and further reduce the threshold of the first early warning gradient. Table 1: Examples of Joint Early Warning As attached Figure 3 and 4 As shown, after testing on 1000 patients, the above predictive model was validated and showed an AUC of 0.777, indicating good predictive efficacy. Based on the evaluation of this static predictive model, as shown in the attached figure... Figure 5 The study suggests that 28% of pregnant women may benefit from optimizing their clinical decision-making process. Assuming a primiparous woman, 35 years old, 160cm tall, with an estimated fetal weight of 3500g, and an interspinous diameter of 10cm measured using a 3D obstetric navigation system, her total score is approximately 210-220 points. Based on the nomogram, the probability of conversion to cesarean section is approximately 0.4. Inputting this data into a predictive model, the calculated probability of conversion to cesarean section is 0.413 (95% confidence interval 0.315-0.519). After testing with 1000 patients, 30% of pregnant women may benefit from optimizing their clinical decision-making process using dynamic predictive models and time-series dynamic models. Based on the same inventive concept, the present application also provides a labor risk identification system based on parameter fusion, which is used to implement the labor risk identification method based on parameter fusion. Based on the same inventive concept, the present application also provides a labor risk identification server based on parameter fusion, which is used to implement the labor risk identification method based on parameter fusion. Based on the same inventive concept, the present application also provides a labor risk identification device based on parameter fusion, which is used to implement the labor risk identification method based on parameter fusion. Based on the same inventive concept, the present application also provides a labor risk identification platform based on parameter fusion, which is used to implement the labor risk identification method based on parameter fusion. A computer program, which can also be referred to or described as a program, software, a software application, an app, a script, or code, can be written in any form of programming language, including compiled or interpreted languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, object, or other unit suitable for use in a computing environment. A computer program may, but need not, correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub programs, or portions of code). A computer program can be deployed to be executed on one computer or on multiple computers that are located at one site or distributed across multiple sites and are interconnected by a communication network.

[0023] Those skilled in the art will appreciate that embodiments of the present application can be devised for a method, a system, or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0024] The present application is described with reference to the flowchart and / or block diagram illustrations of the methods, apparatus (systems) and computer program products according to embodiments of the present application. It will be understood that each block of the flowchart and / or block diagrams, and combinations of blocks in the flowchart and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing system or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart and / or block diagram block or blocks, and the flowchart and / or block diagram block combinations. These computer program instructions can also be stored in a computer-Figure 1 apparatuses that implement the functions specified in the flowchart Figure 1

[0025] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flowchart Figure 1 apparatuses that implement the functions specified in the flowchart Figure 1

[0026] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions that are executed on the computer or other programmable apparatus provide steps for implementing the flowchart Figure 1 apparatuses that implement the functions specified in the flowchart Figure 1

[0027] Finally, it should be noted that the above-mentioned embodiments are merely used to illustrate the technical solutions of the present application, rather than limiting the same. Even though the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced equivalently, and any modification or replacement without departing from the spirit and scope of the present application should be covered within the protection scope of the claims of the present application.​​​

Claims

1. A labor risk identification method based on parameter fusion, characterized in that, The method comprises: Step S1: obtaining a first clinical indicator; the first clinical indicator is a static indicator; screening a significant indicator based on the first clinical indicator, and constructing a static prediction model; using the patient's clinical indicators and the static prediction model for auxiliary early warning, and determining whether the dynamic prediction condition is met; if yes, go to step S2; the first clinical indicator includes age, height, body weight at delivery, parity, mode of labor and fetal clinical estimated weight; Determining whether the dynamic prediction condition is met, specifically: if the static prediction probability output by the static prediction model is greater than or equal to the preset first warning threshold, it is determined that the dynamic prediction condition is met; Step S2: positioning the pelvic inlet plane and the birth canal space; taking the fifth lumbar spine as the reference point, establishing a three-dimensional pelvic coordinate system; using ultrasonic auxiliary spatial positioning technology, identifying specific anatomical landmark points of the pelvis through a micro site sensor; Step S3: ultrasonic positioning of the fetal head position at the pubic symphysis; determining a third clinical indicator; real-time measurement and display of the degree of descent of the fetal head relative to the pelvic inlet plane and the birth canal, construction of a fetal head descent curve, and determination of the third clinical indicator; the third clinical indicator includes fetal head-pubic symphysis distance, fetal head-sciatic spine relative position, pubic arch angle, presentation height change speed and pubic arch-fetal head angle; Step S4: evaluating the fetal head position; marking key skull landmarks on the ultrasound interface; determining the spatial geometric relationship between the landmarks and the mid-axis of the birth canal to obtain a second clinical indicator; the second clinical indicator includes fetal head rotation angle and fetal head entry into the pelvis position; Step S5: Pre-constructing a dynamic prediction model based on the first clinical indicator, the second clinical indicator, and the third clinical indicator; obtaining a dynamic prediction probability using the patient clinical indicators and the dynamic prediction model ; performing auxiliary early warning based on the dynamic prediction probability. 2.The labor risk identification method based on parameter fusion according to claim 1, characterized in that, The significant indicators are screened based on the first clinical indicators, and a static prediction model is constructed; specifically, single-factor attribution analysis is directly performed on the first clinical indicators, and the first clinical indicators with P<0.05 are screened as significant indicators; a prediction model is constructed based on R language logistic regression coefficients, the discrimination of the prediction model is verified through area evaluation of the prediction model curve, the stability of the prediction model is verified through resampling, the net benefit threshold of the prediction model is determined by using a decision curve, and a benchmark prediction model satisfying the discrimination, stability and net benefit threshold is obtained as shown in the following formulas (1) and (2); wherein: The value represents the static prediction probability of cesarean section for pregnant women; is age, years, is parity, primipara = 0; multipara = 1, is the interspinous diameter, cm, is the clinical estimated weight, g, is the height, cm, is the mode of labor, natural labor = 0; induced labor = 1; the static probability can be obtained by inputting the significant indicators into the benchmark prediction model. （1）； （2）。 3.The labor risk identification method based on parameter fusion according to claim 2, characterized in that, The static prediction model is represented as a nomogram prediction model; specifically: the static prediction model is represented as a nomogram prediction model by R language to visualize and display, and the influence of each prediction variable on the prediction result is intuitively displayed in a graphical manner, so as to quickly estimate. 4.The labor risk identification method based on parameter fusion according to claim 3, characterized in that, The second clinical indicator is intercepted, specifically: for the position of the fetal head in the pelvis, the vertical distance between the lowest point of the fetal head and the horizontal level of the ischial spine is used as a continuous value representing the position of the fetal head in the pelvis, and the numerical range between [-2, 3] cm is intercepted; a negative value indicates that the fetal head has not reached the ischial spine, and a positive value indicates that it has passed; the position of the fetal head in the pelvis is set to be valued between [-2, 3] cm, when less than -2 cm, the value is set to -2 cm, and when greater than or equal to +3 cm, the value is set to +3 cm; the rotation angle of the fetal head is segmented and weighted; the rotation angle of the fetal head is segmented and weighted is segmented and weighted; （3）。 5. The parameter fusion-based labor risk identification method according to claim 4, characterized in that, The dynamic prediction model is pre-constructed based on the first clinical indicator, the second clinical indicator and the third clinical indicator; specifically: analyzing historical clinical data to obtain significant indicators in the first clinical indicator, the third clinical indicator and the second clinical indicator; pre-constructing the dynamic prediction model. 6.The labor risk identification method based on parameter fusion according to claim 5, characterized in that, The analysis of the historical clinical data to obtain the significant indicators in the first clinical indicator, the third clinical indicator and the second clinical indicator is specifically: performing single-factor attribution analysis on the first clinical indicator and the second clinical indicator directly, and screening the first clinical indicator and the second clinical indicator with P<0.05 as significant indicators; for the third clinical indicator, the characteristic value of each third clinical indicator is calculated; a binary third combination is constructed from the characteristic value of any third clinical indicator and another arbitrary third clinical indicator; single-factor attribution analysis is performed on each third clinical indicator, the characteristic value of the third clinical indicator and the third combination; the P values obtained by analysis are sorted from small to large, and the P<0.05 ones are intercepted to form a third ordered sequence, and the significant indicators are selected based on the third ordered sequence.

7. A platform for labor risk identification based on parameter fusion, characterized in that, The platform is used to implement the parameter fusion-based labor risk identification method of any one of claims 1-6. 8.A server for labor risk identification based on parameter fusion, characterized in that, The system comprises a processor, the processor and the memory are coupled, the memory stores program instructions, and the program instructions stored in the memory realize the parameter fusion based labor risk identification method in any one of claims 1-6 when executed by the processor.

9. A parameter fusion-based labor risk identification system, characterized in that, The system is used to realize the parameter fusion based labor risk identification method in any one of claims 1-6.

10. A computer-readable storage medium, characterized in that, The program makes the computer execute the parameter fusion based labor risk identification method in any one of claims 1-6 when running on the computer.

Images