Newborn inherited metabolic disease screening automatic interpretation system based on artificial intelligence

Abstract

The invention relates to the technical field of neonatal inherited metabolic disease screening, and discloses a neonatal inherited metabolic disease screening automatic interpretation system based on artificial intelligence, which comprises a data acquisition module, a data preprocessing module, an intelligent analysis engine module and a grading early warning module, the data acquisition module receives metabolism index data of amino acid, organic acid and fatty acid in a neonatal sample obtained through tandem mass spectrometry detection and then transmits the data to the data preprocessing module, and the data preprocessing module conducts normalization processing and denoising processing on the data. And then the processed data is transmitted to the intelligent analysis engine module. According to the invention, after the data acquisition module receives the detection data, the detection data is automatically transmitted to the data preprocessing module, so that the problems of low data quality and low processing efficiency during processing of a large amount of complex data due to the fact that existing hereditary metabolic disease screening mostly depends on manual processing and analysis are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of neonatal genetic metabolic disease screening, and particularly to an automatic interpretation system for neonatal genetic metabolic disease screening based on artificial intelligence. Background Art

[0002] Neonatal genetic metabolic disease screening is an important medical measure for early detection of potential genetic metabolic diseases in newborns. Most of these diseases are caused by genetic factors, resulting in abnormal metabolic functions in the body. They are often difficult to detect in the neonatal stage. Once the disease occurs, it will seriously affect the growth and development, intellectual development of the child, and even endanger life. The screening is generally carried out 72 hours after the birth of the newborn and after sufficient breastfeeding. The heel blood is mainly collected, and the blood is dropped on a specific filter paper to make a dried blood spot specimen. Then, advanced detection technologies, such as tandem mass spectrometry, can efficiently screen out multiple diseases at one time, such as phenylketonuria, congenital hypothyroidism, glucose-6-phosphate dehydrogenase deficiency, etc. If these diseases can be detected early, through special dietary interventions, drug treatments and other means, the deterioration of the condition can be greatly avoided, and the child can grow up healthily like a normal child. For example, children with phenylketonuria can effectively prevent mental retardation by strictly implementing a low-phenylalanine diet in the early stage. Neonatal genetic metabolic disease screening is like a "safety lock" for the health of newborns, playing a crucial role in reducing birth defects and improving the population quality, and laying a good foundation for the future of the child.

[0003] Most of the existing genetic metabolic disease screenings rely on manual processing and analysis. Therefore, when dealing with a large amount of complex data, there may be problems such as low data quality and low processing efficiency. Summary of the Invention

[0004] To make up for the above deficiencies, the present invention provides an automatic interpretation system for neonatal genetic metabolic disease screening based on artificial intelligence, aiming to improve the situation that most of the existing genetic metabolic disease screenings rely on manual processing and analysis, and thus may have problems such as low data quality and low processing efficiency when dealing with a large amount of complex data.

[0005] In a first aspect, the present invention provides the following technical solution. An automatic interpretation system for neonatal genetic metabolic disease screening based on artificial intelligence includes a data acquisition module, a data preprocessing module, an intelligent analysis engine module, and a grading and warning module. The data acquisition module receives data on amino acid, organic acid, and fatty acid metabolic indicators in neonatal samples obtained by tandem mass spectrometry, and then transports the data to the data preprocessing module. The data preprocessing module performs normalization processing and denoising processing on the data, and then transports the processed data to the intelligent analysis engine module. The intelligent analysis engine module establishes a risk assessment model for 36 genetic metabolic diseases through a random forest algorithm and a gradient boosting tree algorithm. The expert diagnosis rules and historical detection data of genetic metabolic diseases are integrated inside the risk assessment model. The intelligent analysis engine module evaluates the data after normalization processing and denoising processing, and then transports the evaluation results to the grading and warning module. The grading and warning module intelligently identifies the evaluation results and generates corresponding grading and warning information.

[0006] Preferably, the amino acid metabolic indicators include alanine, valine, glycine, ornithine, arginine, leucine, methionine, phenylalanine, and tyrosine.

[0007] Preferably, the organic acid metabolic indicators include propionylcarnitine, methylmalonylcarnitine, isovalerylcarnitine, glutarylcarnitine, and 3-hydroxy-isovalerylcarnitine.

[0008] Preferably, the fatty acid metabolic indicators include free carnitine, acetylcarnitine, octanoylcarnitine, decanoylcarnitine, hexadecanoylcarnitine, and octadecanoylcarnitine.

[0009] Preferably, the expert diagnosis rules include the abnormal threshold range of metabolic indicators corresponding to genetic metabolic diseases, the ratio relationship between indicators, and the clinical symptom association rules.

[0010] Preferably, the historical detection data includes the metabolic indicator data of historical confirmed cases and the corresponding disease labels. The metabolic indicator data includes the concentration values of amino acid, organic acid, and fatty acid metabolic indicators.

[0011] Preferably, a warning level is set inside the grading and warning module. The warning level corresponds to different risk score intervals. The grading and warning information includes the warning level, the name of the abnormal indicator, the value of the abnormal indicator, and the corresponding genetic metabolic disease prompt.

[0012] In a second aspect, the present invention provides the following technical solution. An automatic interpretation method for neonatal genetic metabolic disease screening based on artificial intelligence includes the following steps:

[0013] S1. Receive data on amino acid, organic acid, and fatty acid metabolic indicators in neonatal samples obtained by tandem mass spectrometry through the data acquisition module;

[0014] S2. Use the data preprocessing module to perform normalization processing and denoising processing on the metabolic index data;

[0015] S3. Input the preprocessed metabolic index data into the intelligent analysis engine module, and perform analysis through the built-in random forest algorithm and gradient boosting tree algorithm;

[0016] S4. Based on the risk assessment model for 36 genetic metabolic diseases constructed by the intelligent analysis engine module, combined with the corresponding expert diagnosis rules and historical detection data, calculate the risk score for each sample;

[0017] S5. Through the hierarchical warning module, intelligently identify abnormal samples according to the risk score, and generate hierarchical warning information including warning level, abnormal index name, abnormal index value, and corresponding genetic metabolic disease tips.

[0018] In the third aspect, the present invention provides the following technical solution. A computer device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned automatic interpretation method for neonatal genetic metabolic disease screening based on artificial intelligence.

[0019] In the fourth aspect, the present invention provides the following technical solution. A readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the above-mentioned automatic interpretation method for neonatal genetic metabolic disease screening based on artificial intelligence.

[0020] The present invention has the following beneficial effects:

[0021] 1. In the present invention, after receiving the detection data through the data acquisition module, it is automatically transported to the data preprocessing module. After normalization and denoising processing, it enters the intelligent analysis engine module. The random forest algorithm and gradient boosting tree algorithm are used, and at the same time, the expert diagnosis rules and historical detection data of genetic metabolic diseases are integrated for analysis and calculation. Finally, the warning is completed through the hierarchical warning module. The entire process has a high degree of automation, greatly improving the processing efficiency, and can quickly process a large number of neonatal samples to meet the needs of large-scale screening. Thus, it solves the problem that most of the existing genetic metabolic disease screenings rely on manual processing and analysis, and thus there may be problems such as low data quality and low processing efficiency when processing a large amount of complex data.

[0022] 2. In the present invention, the data preprocessing module performs normalization processing and denoising processing on the original amino acid, organic acid, and fatty acid metabolism index data, eliminates the data differences caused by different detection batches, equipment, etc., removes noise and outliers, improves the accuracy of the data, and provides a reliable data basis for subsequent accurate analysis.

[0023] 3. In the present invention, the risk assessment model for 36 genetic metabolic diseases constructed by the intelligent analysis engine module can simultaneously perform risk assessment on 36 genetic metabolic diseases, covering various diseases related to amino acid, organic acid, and fatty acid metabolism. Compared with the traditional screening technology that only targets a few common genetic metabolic diseases, it significantly expands the types of diseases screened, enabling more potential children with genetic metabolic diseases to be detected early. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a system architecture diagram of the automatic interpretation system for neonatal genetic metabolic disease screening based on artificial intelligence proposed by the present invention;

[0025] Figure 2 It is a flowchart of the automatic interpretation method for neonatal genetic metabolic disease screening based on artificial intelligence proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings. Apparently, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0027] Embodiment 1

[0028] Refer to Figure 1, in the first embodiment of the present invention, the present invention provides an automatic interpretation system for neonatal genetic metabolic disease screening based on artificial intelligence, including a data acquisition module, a data preprocessing module, an intelligent analysis engine module, and a hierarchical early warning module. The data acquisition module receives the data of amino acid, organic acid, and fatty acid metabolism indicators in the neonatal samples obtained by tandem mass spectrometry detection, and then transports the data to the data preprocessing module. The data preprocessing module performs normalization processing and denoising processing on the data, and then transports the processed data to the intelligent analysis engine module. The intelligent analysis engine module establishes a risk assessment model for 36 genetic metabolic diseases through the random forest algorithm and the gradient boosting tree algorithm. The expert diagnosis rules and historical detection data of genetic metabolic diseases are integrated inside the risk assessment model. The intelligent analysis engine module evaluates the data after normalization processing and denoising processing, and then transports the evaluation results to the hierarchical early warning module. The hierarchical early warning module performs intelligent identification on the evaluation results and generates corresponding hierarchical early warning information.

[0029] Specifically, the data acquisition module can receive data on amino acid, organic acid, and fatty acid metabolism indicators in neonatal samples from the tandem mass spectrometry detection link, providing the original data for subsequent analysis. By obtaining various types of metabolic indicator data, the system can integrate and analyze these data, comprehensively considering the relationships between indicators of different metabolic pathways. For example, when diagnosing phenylketonuria, not only the concentration of phenylalanine is considered, but also the ratio of phenylalanine to tyrosine is combined, improving the accuracy of disease diagnosis, avoiding the limitations of single-indicator analysis, and providing data input for the machine learning algorithm and risk assessment model of the intelligent analysis engine module. By training the model with a large amount of sample data, it can accurately identify abnormal samples. The risk assessment model calculates the risk score based on this, and the grading and warning module generates warning information accordingly, thus realizing the intelligent identification and grading warning of inborn errors of metabolism; the data preprocessing module can perform normalization processing on the original data of amino acid, organic acid, and fatty acid metabolism indicators, eliminating data differences caused by factors such as different detection batches and equipment, ensuring data consistency and comparability. At the same time, through denoising processing, noise and outliers generated during the detection process are removed, improving data accuracy, providing a reliable data basis for subsequent accurate analysis. The processed data better meets the requirements of the machine learning algorithm in the intelligent analysis engine module, enabling the algorithm to more effectively extract data features, improving the accuracy and stability of model training and risk assessment, and further ensuring the reliability of the risk assessment of inborn errors of metabolism. The preprocessed data enables the grading and warning module to more accurately identify abnormal samples. After removing interference factors, abnormal data is more likely to stand out, helping the grading and warning module to accurately generate corresponding warning information based on the risk score, reducing misjudgment, and improving the reliability of screening results;Through the intelligent analysis engine module, the expert diagnosis rules for 36 genetic metabolic diseases can be integrated with a large amount of historical test data. These rules contain key information such as the abnormal thresholds of metabolic indicators corresponding to various diseases and the ratio relationships between indicators. The historical data covers the relevant test values of numerous confirmed cases, providing rich references for subsequent analysis. Using built-in machine learning algorithms (such as random forest algorithm, gradient boosting tree algorithm), in-depth analysis is carried out on the metabolic indicator data processed by the data preprocessing module. The 36 genetic metabolic diseases include: 12 amino acid metabolism diseases: phenylketonuria, maple syrup urine disease, tyrosinemia, homocystinuria, non-ketotic hyperglycinemia, citrullinemia, argininemia, ornithine carbamoyltransferase deficiency, methioninemia, hyperphenylalaninemia, citrin deficiency, hyperornithinemia; 12 organic acid metabolism diseases: methylmalonic acidemia, propionic acidemia, isovaleric acidemia, glutaric acidemia type I, glutaric acidemia type II, multiple carboxylase deficiency, 3-methylcrotonyl-CoA carboxylase deficiency, 3-hydroxy-3-methylglutaric aciduria, ethylmalonic acidemia, malonic acidemia, isobutyryl-CoA dehydrogenase deficiency, β-ketothiolase deficiency; 12 fatty acid oxidation metabolism diseases: short-chain acyl-CoA dehydrogenase deficiency, medium-chain acyl-CoA dehydrogenase deficiency, very-long-chain acyl-CoA dehydrogenase deficiency, carnitine transport disorder, carnitine palmitoyltransferase I deficiency, carnitine palmitoyltransferase II deficiency, carnitine / acylcarnitine translocase deficiency, primary carnitine deficiency, long-chain hydroxyacyl-CoA dehydrogenase deficiency, mitochondrial trifunctional protein deficiency, medium-chain ketoacyl-CoA thiolase deficiency, short-chain acyl-CoA dehydrogenase deficiency. Through complex calculations, potential patterns and associations in the data are mined to construct a risk assessment model for each genetic metabolic disease, thereby calculating the disease risk score for each sample, providing decision support for the hierarchical warning module. The output risk score is the key basis for hierarchical warning. Based on the score, the hierarchical warning module can intelligently identify abnormal samples, accurately judge the severity of the abnormal situation, and generate corresponding warning information, providing important references for subsequent diagnosis and intervention;Through the hierarchical warning module, based on the risk scores given by the risk assessment module, abnormal samples can be classified and corresponding warnings can be generated, presenting the complex risk levels in different grades (such as first-level, second-level, and third-level warnings), enabling medical staff to quickly understand the severity of the sample abnormalities, clearly indicating the names and values of the abnormal indicators in the warning information, suggesting the possible corresponding inborn errors of metabolism, and providing directions for subsequent diagnosis. For example, it is prompted that "the concentration of propionylcarnitine (C3) is abnormally elevated, and methylmalonic academia may be suffered", helping doctors to arrange further examinations targeted, achieving early and precise intervention. Different grades of warnings can correspond to different processing procedures. For example, a first-level warning may only require a routine reexamination, a second-level warning requires a quick reexamination and judgment in combination with clinical symptoms, and a third-level warning requires immediately starting the confirmation process, thus optimizing the entire screening process and improving the screening efficiency and quality.

[0030] Refer to Figure 1 , amino acid metabolism indicators include alanine, valine, glycine, ornithine, arginine, leucine, methionine, phenylalanine, and tyrosine; organic acid metabolism indicators include propionylcarnitine, methylmalonylcarnitine, isovaleryl carnitine, glutaryl carnitine, and 3-hydroxyisovaleryl carnitine; fatty acid metabolism indicators include free carnitine, acetyl carnitine, octanoyl carnitine, decanoyl carnitine, hexadecanoyl carnitine, and octadecanoyl carnitine.

[0031] Specifically, through these metabolic indicators are important bases for the early screening of inborn errors of metabolism. Tandem mass spectrometry technology can simultaneously detect multiple related indicators, and can conduct a preliminary screening for multiple inborn errors of metabolism in one experiment, improving the screening efficiency, being able to detect potential disease risks at the neonatal stage, achieving early detection. Abnormal changes in different metabolic indicators correspond to different types of inborn errors of metabolism. For example, abnormal phenylalanine metabolism indicators indicate phenylketonuria, abnormal organic acid metabolism indicators such as methylmalonylcarnitine are related to methylmalonic academia, and abnormal fatty acid metabolism indicators, such as medium-chain acyl-CoA dehydrogenase deficiency, can be reflected by specific acylcarnitine indicator changes. Doctors can accurately judge the disease type based on these indicators, combined with clinical symptoms and other tests. During the disease treatment process, continuously monitoring the changes of these metabolic indicators can evaluate the treatment effect and the disease progression. If the treatment is effective, the related abnormal metabolic indicators will gradually tend to be normal. Otherwise, it indicates that the treatment plan needs to be adjusted, providing important references for clinical treatment.

[0032] Refer to Figure 1 , the expert diagnosis rules include the abnormal threshold ranges of metabolic indicators corresponding to inborn errors of metabolism, the ratio relationships between indicators, and the clinical symptom association rules.

[0033] Specifically, this can provide a unified and clear standard for the diagnosis of inborn errors of metabolism. The abnormal thresholds of metabolic indicators and the ratio relationships between indicators can convert the detection data into specific diagnostic bases, avoiding the subjectivity and uncertainty of diagnosis. For example, in phenylketonuria, when the phenylalanine concentration is higher than a specific threshold and the Phe / Tyr ratio exceeds the normal range, combined with relevant clinical symptoms, a definite diagnosis can be made, improving the diagnostic accuracy. As the core component of the intelligent analysis engine module and the risk assessment model, it provides a logical judgment basis. The intelligent system automatically identifies abnormal samples and calculates the risk score by comparing the detection data with the expert diagnosis rules, realizing automated diagnosis. The clinical symptom association rules help doctors comprehensively consider symptoms and detection indicators and formulate reasonable diagnosis and treatment plans. For patients with specific abnormal metabolic indicators accompanied by corresponding clinical symptoms, doctors can judge the severity of the disease according to these rules and select appropriate treatment methods, improving the pertinence and effectiveness of treatment.

[0034] Refer to Figure 1 , the historical detection data includes the metabolic indicator data of historical confirmed cases and the corresponding disease labels, and the metabolic indicator data includes the concentration values of amino acid, organic acid and fatty acid metabolic indicators.

[0035] Specifically, this can provide learning materials for the machine learning algorithm in the intelligent analysis engine module. By learning the metabolic indicator data (concentration values of amino acid, organic acid and fatty acid metabolic indicators) of a large number of historical confirmed cases and the corresponding disease labels, the algorithm can discover the potential associations and patterns between different diseases and metabolic indicators, and then construct an accurate risk assessment model, enhancing the prediction ability for inborn errors of metabolism. It is used to verify the accuracy and effectiveness of the expert diagnosis rules. By comparing the disease labels and the actual indicator situations in the historical data, the deficiencies of the existing diagnosis rules can be found and optimized to make them more in line with the actual clinical situation and improve the reliability of diagnosis. In actual detection, by comparing and analyzing the detection data of new samples with the historical data, the current detection results can be calibrated and the errors can be reduced. Especially when facing complex or atypical cases, the historical data can provide reference to assist in judging the abnormal situation of samples and provide a more comprehensive basis for clinical diagnosis.

[0036] Refer to Figure 1 , there is a warning level set inside the hierarchical warning module, and the warning level corresponds to different risk score intervals. The hierarchical warning information includes the warning level, the name of the abnormal indicator, the value of the abnormal indicator and the corresponding inborn error of metabolism prompt.

[0037] Specifically, by setting different warning levels and corresponding risk score ranges, the complex risk assessment results can be presented to medical staff in an intuitive and concise manner. For example, green represents low risk, yellow represents medium risk, and red represents high risk. Medical staff can quickly know the risk level of the sample, improve the screening efficiency, clearly list the names and values of abnormal indicators, so that medical staff can accurately understand where the metabolic abnormalities are, such as "propionylcarnitine (C3), abnormal value: 12.37 μmol / L", which is convenient for medical staff to focus on the problem indicators, analyze the reasons for the abnormalities, give corresponding genetic metabolic disease prompts, provide a direction for subsequent diagnosis, such as suggesting "may have methylmalonic acidemia", help medical staff quickly lock in the possible disease range, combine with clinical symptoms and other examinations for targeted diagnosis, avoid blind screening, save the diagnosis time. Different warning levels can be matched with different processing procedures. Low-risk may only require routine reexamination, while high-risk requires immediate confirmatory examinations and interventions, which helps to reasonably allocate medical resources, optimize the screening process, and improve the screening quality.

[0038] Embodiment 2:

[0039] Referring to Figure 2 , in the second embodiment of the present invention, the present invention provides an automatic interpretation method for neonatal genetic metabolic disease screening based on artificial intelligence, including the following steps:

[0040] S1. Receive data on amino acid, organic acid, and fatty acid metabolism indicators in neonatal samples obtained by tandem mass spectrometry detection through a data acquisition module;

[0041] S2. Use a data preprocessing module to perform normalization processing and denoising processing on the metabolic indicator data;

[0042] S3. Input the preprocessed metabolic indicator data into an intelligent analysis engine module, and perform analysis through the built-in random forest algorithm and gradient boosting tree algorithm;

[0043] S4. Based on the risk assessment model for 36 genetic metabolic diseases constructed by the intelligent analysis engine module, combined with the corresponding expert diagnosis rules and historical detection data, calculate the risk score for each sample;

[0044] S5. Through a hierarchical warning module, intelligently identify abnormal samples according to the risk score, and generate hierarchical warning information including warning level, abnormal indicator name, abnormal indicator value, and corresponding genetic metabolic disease prompts.

[0045] Specifically, S1 collects various metabolic index data through a data acquisition module to comprehensively reflect the metabolic status of newborns and provide rich original information for screening. The data preprocessing module in S2 normalizes and denoises the data to improve the data quality and ensure the accuracy and reliability of subsequent analysis. S3 deeply mines data features using the random forest algorithm and the gradient boosting tree algorithm. S4 constructs a risk assessment model by combining expert diagnosis rules and historical detection data, calculates the risk score, and realizes the efficient and accurate risk assessment of 36 genetic metabolic diseases. S5, based on the risk score, intelligently identifies abnormal samples through a hierarchical warning module, generates detailed hierarchical warning information, clarifies the abnormal conditions and potential diseases, provides a key basis for early intervention, improves the screening efficiency and quality, and reduces missed diagnoses and misdiagnoses.

[0046] Embodiment 3

[0047] In the third embodiment of the present invention, based on the same inventive concept, a computer-readable storage medium is proposed. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it realizes the steps of the above-mentioned embodiment of the automatic interpretation method for neonatal genetic metabolic disease screening based on artificial intelligence.

[0048] Embodiment 4

[0049] In the fourth embodiment of the present invention, based on the same inventive concept, a computer device is proposed, including: a processor and a memory; the processor and the memory communicate with each other; the memory is used to store instructions; the processor is used to execute the instructions in the memory to execute the above-mentioned embodiment of the automatic interpretation method for neonatal genetic metabolic disease screening based on artificial intelligence.

[0050] It should be understood that each part of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following well-known technologies in the art can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits with appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0051] Finally, it should be noted that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An artificial intelligence-based automatic interpretation system for screening genetic metabolic diseases in newborns, comprising a data acquisition module, a data preprocessing module, an intelligent analysis engine module and a graded warning module, characterized in that: The data acquisition module receives the amino acid, organic acid and fatty acid metabolism index data in the neonatal sample obtained by tandem mass spectrometry detection, and then transmits the data to the data preprocessing module, the data preprocessing module normalizes and denoises the data, and then transmits the processed data to the intelligent analysis engine module, the intelligent analysis engine module establishes a risk assessment model for 36 genetic metabolic diseases through random forest algorithm and gradient boosting tree algorithm, the risk assessment model internally integrates expert diagnostic rules and historical detection data of genetic metabolic diseases, the intelligent analysis engine module evaluates the normalized and denoised data, and then transmits the evaluation results to the graded warning module, the graded warning module intelligently identifies the evaluation results and generates corresponding graded warning information.

2. The artificial intelligence-based automatic interpretation system for screening genetic metabolic diseases in newborns according to claim 1, characterized in that: The amino acid metabolism indicators include alanine, valine, glycine, ornithine, arginine, leucine, methionine, phenylalanine and tyrosine.

3. The artificial intelligence-based automatic interpretation system for screening genetic metabolic diseases in newborns according to claim 2, characterized in that: The organic acid metabolism indicators include propionylcarnitine, methylmalonylcarnitine, isovalerylcarnitine, glutarylcarnitine and 3-hydroxy-isovalerylcarnitine.

4. The artificial intelligence-based automatic interpretation system for screening genetic metabolic diseases in newborns according to claim 1, characterized in that: The fatty acid metabolism indicators include free carnitine, acetyl carnitine, octanoyl carnitine, decanoyl carnitine, hexadecanoyl carnitine and octadecanoyl carnitine.

5. The artificial intelligence-based automatic interpretation system for screening genetic metabolic diseases in newborns according to claim 1 is characterized in that: The expert diagnosis rules include abnormal threshold ranges of metabolic indicators corresponding to inherited metabolic diseases, ratio relationships between indicators, and clinical symptom association rules.

6. The artificial intelligence-based automatic interpretation system for screening genetic metabolic diseases in newborns according to claim 1, characterized in that: The historical test data include metabolic index data of historically confirmed cases and corresponding disease labels, and the metabolic index data include concentration values ​​of amino acid, organic acid and fatty acid metabolic indexes.

7. The artificial intelligence-based automatic interpretation system for screening genetic metabolic diseases in newborns according to claim 1, characterized in that: The graded warning module is internally provided with warning levels, and the warning levels correspond to different risk score intervals. The graded warning information includes the warning level, the abnormal indicator name, the abnormal indicator value and the corresponding genetic metabolic disease prompt.

8. An artificial intelligence-based automatic interpretation method for screening genetic metabolic diseases in newborns, characterized in that: The use of the artificial intelligence-based automatic interpretation system for screening genetic metabolic diseases in newborns according to any one of claims 1 to 7 comprises the following steps: S1, receiving amino acid, organic acid and fatty acid metabolism index data of neonatal samples obtained by tandem mass spectrometry through a data acquisition module; S2, using a data preprocessing module to perform normalization and denoising on the metabolic index data; S3, input the pre-processed metabolic index data into the intelligent analysis engine module, and analyze it through the built-in random forest algorithm and gradient boosting tree algorithm; S4, based on the risk assessment model for 36 inherited metabolic diseases constructed by the intelligent analysis engine module, combined with the corresponding expert diagnosis rules and historical test data, calculate the risk score of each sample; S5. Intelligently identify abnormal samples according to the risk score through the graded warning module, and generate graded warning information including warning level, abnormal indicator name, abnormal indicator value and corresponding genetic metabolic disease prompts.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method for automatically interpreting genetic metabolic diseases in newborns based on artificial intelligence as described in claim 8 is implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by the processor, the method for automatically interpreting genetic metabolic diseases in newborns based on artificial intelligence as claimed in claim 8 is implemented.

Images