CN-122025163-A - Auxiliary reproduction pregnancy-assisting period-oriented blood fat-clinical outcome association analysis system and decision-making method

CN122025163ACN 122025163 ACN122025163 ACN 122025163ACN-122025163-A

Abstract

The invention discloses a blood fat-clinical outcome association analysis system and a decision-making method for auxiliary reproduction and pregnancy-assisting period, which belong to the fields of medical information technology and intelligent diagnosis and treatment, and comprise the steps of acquiring blood fat data and clinical baseline characteristics of a patient on ovulation-promoting starting days and trigger days; based on blood fat data, the clinical baseline characteristic is taken as an adjusting parameter to adjust a preset standard metabolism response curve, constraint fitting is carried out on the adjusted curve through parameter inversion to generate a personalized metabolism response curve and extract dynamic response characteristics, the dynamic response characteristics and the clinical baseline characteristics are input into a risk-outcome prediction model to obtain a prediction result, and clinical decision suggestions are generated according to the prediction result and a decision rule. Discrete blood lipid data is constructed into a continuous personalized metabolic response curve by adopting a method based on parameter inversion and constraint fitting, and dynamic response characteristics are extracted from the continuous personalized metabolic response curve to conduct outcome prediction, so that clinical outcomes can be predicted more accurately, and a reliable basis is provided for personalized clinical decisions.

Inventors

ZHANG YUAN
LIU CHONGYUAN
Lv Huiyun
SUN QIUJU
SUN XIAOYA
XIA XINRU
YUAN CHUN
MA XIANG

Assignees

江苏省人民医院（南京医科大学第一附属医院）

Dates

Publication Date: 20260512
Application Date: 20251229

Claims (10)

1. The blood fat-clinical outcome association analysis decision-making method for auxiliary reproduction pregnancy-assisting period is characterized by comprising the following steps of: Acquiring blood lipid data of a patient on ovulation promoting start days and trigger days and clinical baseline characteristics containing individual physiological background information; based on the blood fat data, adjusting a preset standard metabolism response curve by taking the clinical baseline characteristic as an adjusting parameter, and performing constraint fitting on the adjusted curve through parameter inversion to generate a personalized metabolism response curve with continuous change of blood fat from a starting day to a trigger day; extracting dynamic response characteristics from the personalized metabolic response curve, wherein the dynamic response characteristics comprise at least one of metabolic track fitting degree, cooperative disturbance index, response curvature characteristics and stress area characteristics; Inputting the dynamic response characteristic and the clinical baseline characteristic into a risk-outcome prediction model to obtain a prediction result comprising embryo implantation probability and ovarian hyperstimulation syndrome risk level; and generating clinical decision suggestions according to the prediction results and the decision rules.
2. The assisted reproductive and pregnancy cycle oriented blood lipid-clinical outcome association analysis decision method of claim 1, wherein the obtaining of the patient's blood lipid data on the day of onset of ovulation and the day of triggering and clinical baseline characteristics including individual physiological background information comprises: Acquiring a first blood lipid index set of the patient on the ovulation promoting start day and a second blood lipid index set of the patient on the trigger day to jointly form blood lipid data; acquiring the age, body mass index and basic endocrine hormone level of the patient to jointly form clinical baseline characteristics; at least one of the clinical baseline characteristics is selected as a clinical regulator for adjusting the standard metabolic response curve.
3. The assisted reproductive and pregnancy cycle oriented lipid-clinical outcome association analysis decision method of claim 1, wherein the generating of the personalized metabolic response curve continuously varying from start day to trigger day lipid: constructing a parameterized function model of the standard metabolic response curve, the model comprising morphological parameters for controlling a curve baseline level and a rate of change; initializing and assigning the morphological parameters by utilizing the clinical baseline characteristics to obtain an initial response track reflecting the basic metabolic state of the patient; Setting the blood fat data of the ovulation promoting start day and the trigger day as time-value constraint anchor points of a parameterized function model; And carrying out iterative correction on the morphological parameters through a parameter inversion algorithm, so that the generated curve approaches the constraint anchor point and meets the preset smoothness constraint, and a personalized metabolic response curve is obtained.
4. The assisted reproductive and pregnancy cycle oriented lipid-clinical outcome association analysis decision method of claim 1, wherein the metabolic trace fitness comprises: calculating a residual error between a function value of the personalized metabolic response curve at a time point of the trigger day and an actual blood lipid index value of the patient at the trigger day; and calculating the metabolic track fit degree based on the residual error, wherein the metabolic track fit degree is used for representing the degree that the actual metabolic reaction of the patient under the stimulation of the ovulation promoting drug meets the standard physiological rule.
5. The assisted reproductive and pregnancy cycle oriented lipid-clinical outcome association analysis decision method of claim 1, wherein the co-perturbation index comprises: Acquiring time information of clinical intervention events in an ovulation promoting period; analyzing the trend of the personalized metabolism response curve after the time information of the clinical intervention event, and calculating the morphological deviation between the personalized metabolism response curve and a preset expected response mode; And calculating a cooperative disturbance index based on the morphological deviation, wherein the cooperative disturbance index is used for representing the disturbance intensity of exogenous hormone on lipid metabolism.
6. The assisted reproductive pregnancy cycle oriented lipid-clinical outcome association analysis decision method of claim 1, wherein the response curvature features and stress area features comprise: calculating the curvature change rate of the personalized metabolism response curve to obtain response curvature characteristics; And calculating the area of an area surrounded by the personalized metabolism response curve and a base line taking the ovulation promoting starting day blood lipid value as a constant, and obtaining the stress area characteristic.
7. The assisted reproductive and pregnancy cycle oriented lipid-clinical outcome association analysis decision method of claim 1, wherein the obtaining of the prediction result comprising embryo implantation probability and ovarian hyperstimulation syndrome risk level comprises: When the metabolic track fitting degree is lower than a first preset threshold value and the cooperative disturbance index is higher than a second preset threshold value, judging that the risk level of the ovarian hyperstimulation syndrome is high; and when the metabolic track fit degree is in a preset normal interval and the stress area characteristic is lower than a third preset threshold value, judging that the embryo implantation probability is lower than a fourth preset threshold value.
8. The assisted reproductive and pregnancy cycle oriented lipid-clinical outcome association analysis decision method of claim 7, wherein generating clinical decision advice comprises: Generating a first decision instruction suggesting cancellation of fresh embryo transfer and implementation of whole embryo freezing in response to the ovarian hyperstimulation syndrome risk level being high risk; Generating a second decision instruction suggesting preferential embryo culture optimization and endometrial preparation and temporary transplantation in response to the embryo implantation probability being below the fourth preset threshold; in response to not satisfying the above condition, a third decision instruction is generated suggesting that fresh embryo transfer may be performed.
9. The assisted reproductive and pregnancy cycle oriented blood lipid-clinical outcome association analysis decision method of claim 1, wherein the risk-outcome prediction model comprises: Constructing a training data set containing a blood fat data sequence, clinical baseline characteristics and corresponding real pregnancy-assisting ending labels of a historical patient; generating a historical personalized metabolic response curve based on the data of each historical patient by using a parameter inversion and constraint fitting method, and extracting corresponding historical dynamic response characteristics; And training a model by using the historical dynamic response characteristic and the historical clinical baseline characteristic as inputs and the real pregnancy-assisting ending label as output by using a supervised learning algorithm to establish a mapping relation from a feature space to clinical ending probability.
10. A blood lipid-clinical outcome association analysis system for assisted reproduction assisted pregnancy cycles, applied to the blood lipid-clinical outcome association analysis decision method for assisted reproduction assisted pregnancy cycles according to any one of claims 1 to 9, characterized in that the system comprises: The data acquisition module is used for acquiring blood fat data of a patient on the day of ovulation promotion starting and the day of trigger and clinical baseline characteristics containing individual physiological background information; The track simulation module is used for adjusting a preset standard metabolism response curve based on the blood fat data by taking the clinical baseline characteristic as an adjustment parameter, and performing constraint fitting on the adjusted curve through parameter inversion to generate a personalized metabolism response curve with continuous change of blood fat from a starting day to a trigger day; The characteristic extraction module is used for extracting dynamic response characteristics from the personalized metabolism response curve, wherein the dynamic response characteristics comprise at least one of metabolic track fitting degree, cooperative disturbance index, response curvature characteristics and stress area characteristics; the prediction analysis module is used for inputting the dynamic response characteristic and the clinical baseline characteristic into a risk-outcome prediction model to obtain a prediction result comprising embryo implantation probability and ovarian hyperstimulation syndrome risk level; and the decision generation module is used for generating clinical decision suggestions according to the prediction results and the decision rules.

Description

Auxiliary reproduction pregnancy-assisting period-oriented blood fat-clinical outcome association analysis system and decision-making method Technical Field The invention relates to the field of medical information technology and intelligent diagnosis and treatment, in particular to a blood fat-clinical outcome association analysis system and a decision-making method for auxiliary reproduction pregnancy-assisting period. Background At present, the auxiliary reproduction technology is an important medical means for treating infertility, and one of the key links is controlled hyperstimulation, namely, the ovarian is stimulated by exogenous drugs so as to obtain a plurality of mature oocytes. In the process, the physiological response of a patient is closely monitored, and the risks of bad outcomes such as embryo implantation success rate, ovarian hyperstimulation syndrome and the like are prejudged, so that the method is important for guiding clinical decisions, improving the success rate and guaranteeing the safety of the patient. In recent years, with the development of medical informatization, computer technology has become a research hotspot for analyzing various physiological indexes of patients in a treatment cycle to assist doctors in diagnosis and decision making. Blood lipid is an important metabolic index, and its changes in the ovulation cycle are thought to be closely related to ovarian response and pregnancy outcome. In the related technology, the Chinese patent application with the bulletin number of CN118486462A discloses a model for predicting the occurrence of preeclampsia after the PCOS patient is assisted in pregnancy, which comprises the steps of incorporating and removing a case sample, collecting clinical data of the patient, dividing a training set and a verification set, interpolating a missing value, analyzing a single factor by taking whether the preeclampsia occurs in the patient as a dependent variable and clinical characteristics of the patient as an independent variable in the training set, carrying out linear hypothesis test on a selected candidate predicted variable, adjusting the variable type, carrying out LASSO regression and Logistic regression analysis on the adjusted candidate predicted variable, constructing a candidate model, evaluating the prediction capability of the candidate model, verifying in the verification set, and screening a model with optimal prediction capability. However, the above related art and the currently mainstream prediction methods have a certain limitation in application to the fine management of ovulation-promoting cycle. In one aspect, existing models rely mostly on a single detected value for a static baseline characteristic or discrete time point of the patient for prediction. The snapshot type data processing mode omits the continuous and dynamic evolution process of the in-vivo metabolic environment, particularly the blood fat level, of a patient under the continuous stimulation of the ovulation promoting medicine. Discrete data points are difficult to capture key time sequence information such as the speed, acceleration, fluctuation form and the like of metabolic reactions, and the information often implies the tolerance degree of an organism to medicines and the stress state of an ovary. On the other hand, conventional linear regression models generally assume that there is a simple linear relationship between features and outcomes, but the metabolic regulation network of the human body is a highly complex nonlinear system, and metabolic disturbances caused by exogenous hormonal interventions tend to exhibit complex nonlinear trends. It is difficult to accurately quantify the synergistic perturbation effect caused by drug intervention and the stress load accumulated by the organism only through variable screening and linear weighting, resulting in limited prediction accuracy for high risk or low probability extremes. Furthermore, existing methods lack dynamic calibration against the physiological background of the patient individual, and it is difficult to distinguish whether the abnormality index is underlying pathological manifestations or specific overreactions. Therefore, there is a need for a feature extraction and analysis method that can convert discrete physiological monitoring data into a continuous dynamic trajectory and can deeply extract features reflecting the degree of metabolic stress and the dynamic response pattern of the body. Disclosure of Invention In order to solve the problems, the invention provides the blood fat-clinical outcome association analysis system and the decision method for the assisted reproduction pregnancy-assisting period, discrete blood fat data are constructed into a continuous personalized metabolic response curve by adopting a method based on parameter inversion and constraint fitting, dynamic response characteristics are extracted from the continuous personalized metabolic response curve to perfo