A method for early non-invasive screening of necrotizing enterocolitis in a neonate

Abstract

The application discloses a new-born necrotizing enterocolitis early non-invasive screening method, and belongs to the technical field of new-born intestinal health screening. The method synchronously collects a new-born fresh fecal sample and two non-invasive physiological parameters of transcutaneous oxygen saturation short-term variability and abdominal bilateral body surface temperature difference, and after standardization sample pretreatment, three types of biomarkers of short-chain fatty acid component ratio, flora abundance ratio and fecal calprotectin concentration are detected in parallel, and five multi-source characteristic sets are constructed; after Z-score standardization and min-max mapping processing based on the same gestational age baseline, a pre-trained XGBoost model is inputted to calculate a quantitative risk score, and finally, a low, medium and high three-level risk stratification result is outputted according to a threshold value. The application is non-invasive throughout and parameter standardization throughout the whole process, multi-dimensional heterogenous characteristic fusion is adopted to improve early screening sensitivity, individual physiological heterogeneity interference is reduced, and the application can be adapted to new-born bedside rapid screening scenes.

Description

Technical Field

[0001] This invention relates to the field of non-invasive screening technology for neonatal intestinal health, specifically to a method for early non-invasive screening of neonatal necrotizing enterocolitis. Background Technology

[0002] Necrotizing enterocolitis in newborns is a serious intestinal health disorder that is prevalent in premature infants. Early identification and stratified management of intestinal health risks are crucial clinical needs in neonatal intensive care. Non-invasive screening methods refer to a technical system that assesses and stratifies the health risks of a target population through non-invasive means such as in vitro biological sample testing and collection of surface physiological parameters. Its role is to provide medical staff with objective and quantitative reference data to assist in the daily monitoring and risk warning of neonatal intestinal health.

[0003] Existing screening technologies for this type of intestinal health risk primarily rely on single fecal protein biomarker detection and routine clinical observation. The basic process typically involves simple fecal sample processing followed by single-indicator testing, combined with clinical experience from healthcare professionals for risk assessment. This approach has several shortcomings: First, the screening indicators are limited, targeting only a single pathological pathway, failing to cover early changes in multiple pathways such as intestinal barrier damage, dysbiosis, and abnormal local perfusion, resulting in insufficient sensitivity for ultra-early screening. Second, there is a lack of standardized systems for sample preprocessing and data interpretation; operational parameters and judgment rules vary significantly among different institutions, leading to poor repeatability and consistency of screening results. Third, the physiological baseline differences among preterm infants of different gestational ages are not adequately considered; using uniform absolute thresholds for risk assessment makes them susceptible to interference from individual physiological heterogeneity, resulting in a high rate of false positives or false negatives. Fourth, a complete and systematic process from bedside sampling, sample processing, multi-indicator testing to data fusion calculation, and stratified result output has not yet been established, making it unsuitable for the application scenarios of rapid bedside screening for newborns. Summary of the Invention

[0004] This invention addresses the technical problems of existing non-invasive screening technologies for neonatal intestinal health, such as single indicator dimensions, large interference from individual physiological heterogeneity, insufficient standardization of the entire process, and poor sensitivity and consistency of early risk stratification results.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a non-invasive early screening method for necrotizing enterocolitis in newborns, comprising the following steps:

[0006] S1 Sample and Physiological Parameter Collection: Fresh fecal samples were collected from the newborns being examined, and non-invasive bedside physiological parameters were collected simultaneously. These non-invasive bedside physiological parameters included transcutaneous oxygen saturation short-term variability and the difference in abdominal surface temperature between the two sides.

[0007] S2 Standardized pretreatment of fecal samples: The collected fecal samples were mixed with phosphate buffer solution with pH 7.2~7.4 at a mass-to-volume ratio of 1:9. After thorough mixing by vortexing, the mixture was centrifuged at 4°C. The supernatant was then filtered through a filter membrane to obtain the sample solution to be tested.

[0008] S3 Multidimensional Biomarker Detection: Three parallel tests were performed on the sample solution to be tested: quantitative detection of short-chain fatty acid components to calculate the molar concentration ratio of butyric acid to the total of acetic acid and propionic acid; detection of intestinal flora abundance to calculate the relative abundance ratio of Bifidobacterium and Enterobacteriaceae; and quantitative detection of fecal calprotectin to obtain the mass concentration of calprotectin.

[0009] S4 Multi-source feature normalization processing: The three biomarker detection results obtained in step S3 and the two physiological parameters obtained in step S1, totaling five feature indicators, are standardized and mapped to the 0~1 value range to obtain five standardized feature values.

[0010] S5 Fusion Model Risk Calculation: Input five standardized feature values ​​into a pre-trained gradient boosting tree screening model, and output a risk score for neonatal necrotizing enterocolitis through model computation.

[0011] S6 Screening Stratification Results Output: Based on the preset scoring threshold range, the risk score is divided into three screening levels: low risk, medium risk, and high risk. The standardized values ​​of each characteristic indicator and the baseline reference range for the same gestational age are output simultaneously.

[0012] Furthermore, in step S1, the newborn being examined is a premature infant with a gestational age of 28 to 36 weeks, the stool sample is fresh stool that is naturally excreted, and the amount of a single sample is 0.5 to 1g; the collection time for the physiological parameters is 30 minutes continuously, and the collection process does not require invasive operation.

[0013] Further, in step S2, the rotation speed of the vortex oscillation is 2500~3000 rpm, and the oscillation time is 5~10 min; the rotation speed of the centrifugation treatment is 12000~15000 rpm, and the centrifugation time is 10~15 min; the pore size of the filter membrane is 0.22 μm.

[0014] Further, in step S3, the quantitative detection of short-chain fatty acid components is performed using gas chromatography-mass spectrometry, and the sample solution to be tested is acidified and derivatized before detection; the abundance detection of intestinal flora is performed using 16S rRNA gene V3-V4 region sequencing, and the relative abundance of genera and families is obtained through bioinformatics analysis; the quantitative detection of fecal calprotectin is performed using enzyme-linked immunosorbent assay (ELISA).

[0015] Further, in step S1, the short-term variability of transcutaneous oxygen saturation is the coefficient of variation of the transcutaneous oxygen saturation value during the continuous collection period; the difference in abdominal bilateral body surface temperature is the absolute difference in body surface temperature at a distance of 5 cm from the umbilicus on the left and right sides at the level of the umbilicus.

[0016] Furthermore, in step S4, the standardization process adopts the Z-score standardization method, and the mean and standard deviation of each indicator are taken from a large sample baseline database of healthy preterm infants of the same gestational age. After standardization, all feature values ​​are uniformly mapped to the 0~1 interval through min-max linear transformation.

[0017] Furthermore, in step S5, the pre-trained gradient boosting tree screening model is an XGBoost model. The model training set contains five feature data of clinically confirmed preterm infants with necrotizing enterocolitis and healthy preterm infants of the same period. The training process uses five-fold cross-validation, and the feature weights of the model are automatically generated during training.

[0018] Furthermore, in step S6, the preset scoring threshold range is specifically as follows: a risk score value < 0.3 corresponds to a low-risk level, a risk score value of 0.3 to 0.7 corresponds to a medium-risk level, and a risk score value > 0.7 corresponds to a high-risk level.

[0019] Furthermore, both the sampling in step S1 and the output of the results in step S6 can be completed at the neonatal bedside, and the total time of the entire testing process does not exceed 4 hours.

[0020] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages compared with the prior art:

[0021] 1. Multi-dimensional heterogeneous feature fusion enhances the sensitivity of early screening. This invention integrates three different dimensions of feature indicators: intestinal metabolites, microbiota structure, and surface physiological signs. These indicators correspond to early changes in multiple pathways, such as intestinal barrier damage, microbiota dysbiosis, and abnormal local perfusion. Compared with single-marker screening, this invention can capture earlier and more comprehensive abnormal signals of intestinal health, effectively improving the sensitivity of ultra-early risk screening.

[0022] 2. Ratio-based indicators combined with gestational age baseline normalization reduce interference from individual heterogeneity. This invention employs two types of ratio-based indicators: the butyrate / (acetic acid + propionic acid) molar ratio and the Bifidobacterium / Enterobacterium abundance ratio. Combined with Z-score standardization based on the baseline of healthy individuals of the same gestational age, this effectively eliminates physiological baseline differences among newborns of different gestational ages and weights, reduces the interference of individual heterogeneity on screening results, and improves the accuracy of risk assessment.

[0023] 3. Standardized parameters throughout the entire process ensure consistency and reproducibility of results. This invention sets clear quantitative standards for all aspects, including sample collection volume, buffer ratio, centrifugation parameters, detection methods, normalization rules, and model calculation logic. This forms a standardized operating procedure that can be replicated and promoted, effectively reducing the differences in results between different operators and institutions, and improving the consistency of screening results.

[0024] 4. The entire process is non-invasive and suitable for bedside applications. All sampling and testing steps in this invention are performed in a non-invasive manner. Fecal samples are naturally excreted samples, and physiological parameters are collected through surface sensors. No additional trauma is caused to the newborn, and both sampling and result output can be completed at the bedside, making it suitable for the daily monitoring and rapid screening needs of neonatal intensive care units.

[0025] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0026] Figure 1 This is a timing diagram of the fecal multi-index detection and analysis process of the present invention;

[0027] Figure 2 This is a timeline diagram of the synchronous acquisition of bedside samples and physiological data according to the present invention;

[0028] Figure 3 This is a flowchart of the standardized pretreatment process for fecal samples according to the present invention;

[0029] Figure 4 This is a flowchart of the multi-source feature normalization and risk scoring calculation process of the present invention;

[0030] Figure 5 This is a flowchart illustrating the screening risk stratification and clinical application of the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] It should be noted that the terms vertical, horizontal, up, down, left, right, and similar expressions used in this article are for illustrative purposes only and do not represent the only possible implementation.

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention; the terminology used herein includes any and all combinations of one or more of the associated listed items.

[0034] The following is in conjunction with the appendix Figures 1-5 The present invention provides a complete and clear description of its technical solutions through specific embodiments. These described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0035] This embodiment provides a non-invasive screening method for early neonatal necrotizing enterocolitis, targeting preterm newborns with a gestational age of 28-36 weeks for bedside non-invasive screening. The complete workflow and operation steps are as follows:

[0036] 1. Sample and physiological parameter collection

[0037] Select premature newborns for examination. Within 10 minutes after their natural defecation, collect 0.5-1g of fresh stool sample using a sterile stool collection spoon, place it in a sterile sample tube and seal it, and label the sample with the corresponding newborn's gestational age and birth weight information.

[0038] Synchronous bedside physiological parameter acquisition: A transcutaneous oxygen saturation sensor was fixed to the sole of the newborn's foot, and two body surface temperature sensors were fixed to the abdominal skin surface on the left and right sides of the newborn's umbilicus, 5 cm away from the umbilicus, respectively, and physiological data were continuously collected for 30 minutes; after the collection, the coefficient of variation of the transcutaneous oxygen saturation value during this period was calculated as the short-term variability of transcutaneous oxygen saturation; the absolute difference between the values ​​measured by the two body surface temperature sensors was calculated as the bilateral abdominal body surface temperature difference.

[0039] This entire procedure is non-invasive and requires no invasive treatment of the newborn.

[0040] 2. Standardized pretreatment of fecal samples

[0041] Take out the collected fecal samples and add phosphate buffer solution at pH 7.3 at a mass-to-volume ratio of 1:9, that is, add 0.9 mL of buffer solution for every 0.1 g of feces; place the sample tube on a vortex mixer, set the speed to 2800 rpm, and vortex for 8 minutes to fully mix the fecal samples and buffer solution to form a homogenate.

[0042] The homogenized sample slurry was placed in a low-temperature centrifuge and centrifuged for 12 minutes at a temperature of 4℃ and a speed of 13000 rpm. After centrifugation, the supernatant was carefully aspirated and filtered using a 0.22 μm aqueous filter membrane. The resulting clear sample solution was then stored at 4℃ for later testing.

[0043] 3. Multidimensional biomarker detection

[0044] Take equal amounts of the sample solution to be tested, divide it into three portions, and perform the following three parallel tests on each portion:

[0045] (1) Quantitative detection of short-chain fatty acid components: Take a sample solution to be tested, add hydrochloric acid for acidification, add derivatization reagent for derivatization, and then use gas chromatography-mass spectrometry to detect the molar concentrations of acetic acid, propionic acid and butyric acid respectively. Calculate the ratio of the molar concentration of butyric acid to the sum of the molar concentrations of acetic acid and propionic acid, and record it as the short-chain fatty acid ratio index.

[0046] (2) Detection of gut microbiota abundance: Take a sample solution to be tested, extract total microbial DNA from the sample using a genomic DNA extraction kit, perform PCR amplification and library construction on the V3-V4 region of the 16S rRNA gene, and after high-throughput sequencing, obtain the relative abundance of each genera and family through bioinformatics analysis, calculate the ratio of the relative abundance of Bifidobacterium to the relative abundance of Enterobacteriaceae, and record it as the gut microbiota abundance ratio index.

[0047] (3) Quantitative detection of fecal calprotectin: Take a sample solution to be tested, use an enzyme-linked immunosorbent assay kit, and perform incubation, washing, color development and other operations according to the kit operation specifications. Read the absorbance value through an ELISA reader and calculate the mass concentration of calprotectin in the sample by referring to the standard curve.

[0048] 4. Multi-source feature normalization processing

[0049] The three biomarkers obtained from the above detection—short-chain fatty acid ratio, bacterial abundance ratio, and calcium protectant protein concentration—were normalized along with two physiological parameters—short-term variability of transcutaneous oxygen saturation and difference in bilateral abdominal body surface temperature—to a total of five characteristic indicators.

[0050] First, a large-sample baseline database of healthy preterm infants of the same gestational age (divided by gestational age) is retrieved to obtain the population mean and standard deviation for each indicator. The Z-score standardization method is used to calculate the standardized score for each raw indicator. Then, through the min-max linear transformation, all standardized scores are uniformly mapped to the numerical range of 0 to 1, and finally, 5 standardized feature values ​​with consistent value ranges are obtained for subsequent model calculations.

[0051] 5. Risk Calculation of Fusion Model

[0052] Five standardized feature values ​​are input into the pre-trained XGBoost screening model in a fixed order. The model automatically completes the weight allocation and fusion calculation, and outputs a risk score value in the range of 0 to 1.

[0053] The pre-training process of the XGBoost screening model is as follows: samples of preterm infants with clinically confirmed necrotizing enterocolitis and samples of healthy preterm infants of the same period are collected, and the above five feature data are extracted to form a training dataset. The model training and parameter tuning are completed by using five-fold cross-validation. The feature weights of the model are automatically generated during the training process and do not need to be set manually.

[0054] 6. Output of screening stratification results

[0055] Risk scores are stratified based on a preset scoring threshold range: a risk score less than 0.3 is classified as low risk, a risk score between 0.3 and 0.7 is classified as medium risk, and a risk score greater than 0.7 is classified as high risk.

[0056] The final screening report includes: overall risk level, quantitative risk score in the range of 0 to 1, raw and standardized values ​​of 5 characteristic indicators, and baseline reference ranges for each indicator at the same gestational age, providing a reference for medical staff to carry out neonatal intestinal health management.

[0057] The clinical application of this method is as follows:

[0058] (1) Sampling preparation: Medical staff prepare sterile fecal collection tubes, body surface physiological sensors, pretreatment reagents and related testing equipment in advance, and confirm that the equipment is operating normally and the reagents are within the validity period.

[0059] (2) Bedside sampling: After the newborn has defecated naturally, fecal samples were collected and packaged and numbered according to aseptic operation procedures. At the same time, the newborn was fitted with a physiological sensor on the body surface and 30 minutes of continuous data collection was started to ensure that the sample and physiological data corresponded one by one.

[0060] (3) Sample processing and testing: The collected fecal samples are transferred to the testing area and pre-processed according to the standardized procedure, such as homogenization, centrifugation, and filtration. Then, the samples are separated and tested in parallel for short-chain fatty acids, intestinal flora, and calcium protectants to obtain all the original test data.

[0061] (4) Data calculation and result output: Import all raw test data and physiological parameters into the matching screening calculation system. The system automatically retrieves the baseline database corresponding to the gestational age, completes feature normalization, model fusion calculation and risk stratification, and automatically generates and outputs a standardized screening report.

[0062] (5) Application of results: Medical staff review the screening report and, in conjunction with the newborn’s daily clinical manifestations and signs, formulate corresponding health monitoring frequency and nursing management plan to achieve early identification and early intervention of intestinal health risks.

[0063] This invention employs non-invasive operation and standardized procedures throughout. All detection and calculation steps are based on mature and universal detection technologies and statistical algorithms, requiring no special customized equipment and possessing good feasibility and clinical applicability.

[0064] Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make various modifications and alterations without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be determined by the claims.

Claims

1. A non-invasive screening method for early necrotizing enterocolitis in newborns, characterized in that, Includes the following steps: S1 Sample and Physiological Parameter Collection: Fresh fecal samples were collected from the newborns under examination, and non-invasive bedside physiological parameters were collected simultaneously. The non-invasive bedside physiological parameters included transcutaneous oxygen saturation short-term variability and the difference in abdominal bilateral body surface temperature. S2 Standardized pretreatment of fecal samples: The collected fecal samples were mixed with phosphate buffer solution with pH 7.2~7.4 at a mass-to-volume ratio of 1:

9. After thorough mixing by vortexing, the mixture was centrifuged at 4°C. The supernatant was then filtered through a filter membrane to obtain the sample solution to be tested. S3 Multidimensional Biomarker Detection: Three parallel tests were performed on the sample solution to be tested: quantitative detection of short-chain fatty acid components to calculate the molar concentration ratio of butyric acid to the total of acetic acid and propionic acid; detection of intestinal flora abundance to calculate the relative abundance ratio of Bifidobacterium and Enterobacteriaceae; and quantitative detection of fecal calprotectin to obtain the mass concentration of calprotectin. S4 Multi-source feature normalization processing: The three biomarker detection results obtained in step S3 and the two physiological parameters obtained in step S1, totaling five feature indicators, are standardized and mapped to the 0~1 value range to obtain five standardized feature values. S5 Fusion Model Risk Calculation: Input five standardized feature values ​​into a pre-trained gradient boosting tree screening model, and output a risk score for neonatal necrotizing enterocolitis through model computation; S6 Screening Stratification Results Output: Based on the preset scoring threshold range, the risk score is divided into three screening levels: low risk, medium risk, and high risk. The standardized values ​​of each characteristic indicator and the baseline reference range for the same gestational age are output simultaneously.

2. The method for early non-invasive screening of neonatal necrotizing enterocolitis according to claim 1, characterized in that, In step S1, the newborn being examined is a premature infant with a gestational age of 28 to 36 weeks, and the stool sample is fresh stool that is naturally excreted, with a single sampling amount of 0.5 to 1g; the collection time for the physiological parameters is 30 minutes continuously, and the collection process does not require invasive operation.

3. The method for early non-invasive screening of neonatal necrotizing enterocolitis according to claim 1, characterized in that, In step S2, the rotation speed of the vortex oscillation is 2500~3000 rpm, and the oscillation time is 5~10 min; the rotation speed of the centrifugation treatment is 12000~15000 rpm, and the centrifugation time is 10~15 min; the pore size of the filter membrane is 0.22 μm.

4. The method for early non-invasive screening of neonatal necrotizing enterocolitis according to claim 1, characterized in that, In step S3, the short-chain fatty acid component is quantitatively detected using gas chromatography-mass spectrometry, and the sample solution to be tested is acidified and derivatized before detection; the abundance of intestinal flora is detected using 16S rRNA gene V3-V4 region sequencing, and the relative abundance of genera and families is obtained through bioinformatics analysis; the quantitative detection of fecal calprotectin is performed using enzyme-linked immunosorbent assay (ELISA).

5. A non-invasive early screening method for neonatal necrotizing enterocolitis according to claim 1, characterized in that, In step S1, the short-term variability of transcutaneous oxygen saturation is the coefficient of variation of the transcutaneous oxygen saturation value during the continuous collection period; the difference in abdominal bilateral body surface temperature is the absolute difference in body surface temperature at a distance of 5 cm from the umbilicus on the left and right sides at the level of the umbilicus.

6. A non-invasive early screening method for neonatal necrotizing enterocolitis according to claim 1, characterized in that, In step S4, the standardization process adopts the Z-score standardization method, and the mean and standard deviation of each indicator are taken from a large sample baseline database of healthy preterm infants of the same gestational age. After standardization, all feature values ​​are uniformly mapped to the 0~1 interval through min-max linear transformation.

7. A non-invasive early screening method for neonatal necrotizing enterocolitis according to claim 1, characterized in that, In step S5, the pre-trained gradient boosting tree screening model is an XGBoost model. The model training set contains five feature data of preterm infants with necrotizing enterocolitis diagnosed by the clinical gold standard and healthy preterm infants of the same period. The training process uses five-fold cross-validation, and the feature weights of the model are automatically generated during training.

8. A non-invasive early screening method for neonatal necrotizing enterocolitis according to claim 1, characterized in that, In step S6, the preset scoring threshold range is specifically as follows: a risk score of <0.3 corresponds to a low-risk level, a risk score of 0.3 to 0.7 corresponds to a medium-risk level, and a risk score of >0.7 corresponds to a high-risk level.

9. A non-invasive early screening method for neonatal necrotizing enterocolitis according to claim 1, characterized in that, The sampling in step S1 and the output of the results in step S6 can both be completed at the neonatal bedside, and the total time of the entire testing process does not exceed 4 hours.

Images