CN114068026A

CN114068026A - Prediction device for early metabolic related fatty liver disease

Info

Publication number: CN114068026A
Application number: CN202210052302.8A
Authority: CN
Inventors: 陈菲; 王莹; 霍如松
Original assignee: Suzhou Herui Biotechnology Co ltd
Current assignee: Suzhou Herui Biotechnology Co ltd
Priority date: 2022-01-17
Filing date: 2022-01-17
Publication date: 2022-02-18
Anticipated expiration: 2042-01-17
Also published as: CN114068026B

Abstract

The invention discloses a prediction device for a super-early metabolism-related fatty liver disease, and relates to the technical field of biomedicine.

Description

Prediction device for early metabolic related fatty liver disease

Technical Field

The invention relates to the technical field of biomedicine, in particular to a prediction device for ultra-early metabolic related fatty liver disease.

Background

The MAFLD has replaced viral liver disease and becomes the first liver disease in China. According to the consensus of the international experts in MAFLD 2020 edition, the diagnosis standard of the metabolism-related fatty liver disease is as follows: based on pathological or imaging fatty liver evidence, there is one of three conditions of overweight/obesity, type 2 diabetes or metabolic dysfunction.

Wherein, the pathological diagnosis of liver tissue biopsy is still the gold standard for the diagnosis of the disease, the pathological indicators used for the MAFLD comprise steatosis (S0-S4), ballooning (B0-B4), lobular inflammation (L0-L4) and fibrosis grading (F0-F4), the sum of the four grades is less than or equal to 2, the MAFLD is definitely excluded, the MAFLD is determined by more than or equal to 5, and most of the areas between more than 2 and less than 5 are considered as the ultra-early stage or suspected MAFLD. The pathological differentiation value set to 2 means negative exclusion, and is significant in health screening. On the other hand, when 5 is set, it is more suitable for positive confirmation. The prior art (application No. 201911424602.9) discloses a diagnosis model of non-alcoholic fatty liver disease, the set discrimination value is total score 5, which can only diagnose early non-alcoholic fatty liver disease, but can not effectively predict people in latent state (ultra-early stage). In fact, the area under the highest curve of the prior published metabolic-related fatty liver disease for evaluating medium and severe metabolic-related fatty liver diseases or combined fibrosis can reach about 0.7, but the evaluation effect for early metabolic-related fatty liver disease is limited, and the area of the curve under ROC is only about 0.6. There is a clinical need for more effective methods for early assessment, particularly for more definitive negative exclusion during health examinations.

In view of this, the invention is particularly proposed.

Disclosure of Invention

The invention aims to provide a prediction device for ultra-early metabolism-related fatty liver disease.

The invention is realized by the following steps:

in a first aspect, an embodiment of the present invention provides a training method for a prediction model of a very early metabolic-related fatty liver disease, including: obtaining the individual indexes of the training samples, the detection results of the marker levels and the corresponding labeling results thereof; the markers comprise total serum protein TP and cytokeratin 18, and the individual index comprises height-mass index BMI; inputting individual indexes of the training samples and detection results of marker levels into a pre-constructed prediction model to obtain a prediction evaluation result; the prediction model is used for predicting the disease risk or disease state of the sample ultra-early metabolism-related fatty liver disease according to the individual indexes of the sample and the detection result of the marker level; and updating parameters of the prediction model based on the labeling result and the evaluation result.

In a second aspect, an embodiment of the present invention provides a prediction apparatus for ultra-early metabolic-related fatty liver disease, including: an obtaining module, configured to obtain individual indexes of a sample to be detected and detection results of levels of markers, where the markers and the individual indexes are the same as those in the foregoing embodiment; and the prediction module is used for inputting the individual indexes and the detection results into the prediction model obtained by training the training method of the ultra-early metabolic related fatty liver disease prediction model in the embodiment to obtain the prediction results of the sample to be detected.

In a third aspect, an embodiment of the present invention provides a training device for ultra-early metabolic-related fatty liver disease, including: the acquisition module is used for acquiring the detection results of the individual indexes and the marker levels of the training samples and the corresponding labeling results; the markers and the individual indexes are both the markers and the individual indexes described in the previous embodiment; the prediction module is used for inputting the individual indexes of the training samples and the detection results of the marker levels into a pre-constructed prediction model to obtain a prediction evaluation result; the prediction model is used for predicting the disease risk or disease state of the sample ultra-early metabolism-related fatty liver disease according to the individual indexes of the sample and the detection result of the marker level; and the parameter updating module is used for updating parameters of the prediction model according to the labeling result and the evaluation result.

In a fourth aspect, an embodiment of the present invention provides an electronic device, where the electronic device includes a processor and a memory; the memory is used for storing a program which, when executed by the processor, causes the processor to implement the training method of the ultra-early metabolic related fatty liver disease prediction model according to the previous embodiment.

In a fifth aspect, the present invention provides a computer readable medium, on which a computer program is stored, and the computer program, when executed by a processor, implements the training method for the ultra-early metabolic-related fatty liver disease prediction model according to the foregoing embodiments.

The invention has the following beneficial effects:

the embodiment of the invention provides a prediction device for the ultra-early metabolic related fatty liver disease, which realizes effective prediction of the early metabolic related fatty liver disease through the detection results of markers such as total serum protein, cytokeratin 18 level and BMI, provides more time for the prevention and treatment of the metabolic related fatty liver disease, and is beneficial to improving the treatment effect of a patient.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.

FIGS. 1 to 23 are ROC graphs corresponding to the prediction methods of examples 1 to 23, respectively;

FIG. 24 is a ROC graph corresponding to the prediction method of comparative example 1.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. The examples, in which specific conditions are not specified, were conducted under conventional conditions or conditions recommended by the manufacturer. The reagents or instruments used are not indicated by the manufacturer, and are all conventional products available commercially.

The embodiment of the invention provides an application of a reagent for detecting a marker level in preparing a reagent for detecting the marker level in a screening kit for a very early metabolic-related fatty liver disease, wherein the marker comprises total serum protein TP and cytokeratin 18.

Through a series of creative capabilities, the invention discovers that the effective prediction of the metabolism-related fatty liver disease can be realized by detecting the level of the marker, more time is provided for the prevention and treatment of the metabolism-related fatty liver disease, and the treatment effect of a patient is improved.

The ultra-early metabolism-related fatty liver disease described herein specifically refers to: the proportion of liver fat cells is at least 5%, the liver histopathology detects steatosis (S0-S4), ballooning (B0-B4), lobular inflammation (L0-L4) and fibrosis grading (F0-F4), the sum of four scores is 2, a distinguishing value is obtained, the sum is less than or equal to 2, MAFLD is excluded, the diagnosis of MAFLD is confirmed to be greater than or equal to 5, and the range of more than 2 and less than 5 is considered to be the super-early MAFLD or the suspected MAFLD. The invention refers to various accepted pathological grouping methods for judging the ultra-early metabolic related fatty liver disease at home and abroad, for example, the sum of four pathological scores is less than or equal to 2 and is a non-MAFLD group, and more than 2 is an ultra-early MAFLD group; the sum of the four pathological scores is less than or equal to 2, the fibrosis grade F is less than or equal to 1 and is a non-MAFLD group, and the others are an ultra-early MAFLD group; alternatively, the sum of the four terms is less than 4, and each term is less than 1 and is non-MAFLD, the others are MAFLD.

(the value of the prior art with the application number of 201911424602.9 for distinguishing early non-NASH from NASH is set as a total pathological score of 5, and the threshold value of the invention for distinguishing MAFLD from non-MAFLD is 2. the invention focuses more on screening and predicting latent or suspected conditions in advance, and compared with the early stage (score of 5), the ultra-early stage has more hidden symptoms, so that the prediction difficulty is higher, but has important clinical significance, can effectively predict the disease risk in advance, provides more sufficient time for treatment and intervention, and improves the survival efficiency while reducing the treatment cost.

In a preferred embodiment, the cytokeratin 18 is selected from the group consisting of: at least one of keratin 18-M30 (M30) and keratin 18-M65 (M65).

In a preferred embodiment, the marker further comprises: alanine aminotransferase ALT, aspartate aminotransferase AST, glutamyltransferase GGT, hemoglobin HB, total cholesterol TC, golgi transmembrane glycoprotein 73 GP73, low density lipoprotein LDL, and at least one of fasting glucose and fasting insulin.

The type of the sample may be a serum sample, a whole blood sample or a tissue sample, and is not particularly limited as long as the detection result of the marker is obtained.

The method for detecting the marker by the reagent is not particularly limited in the present invention, and the method includes, but is not limited to, immunological methods and chemical methods. The immunological methods include, but are not limited to, enzyme-linked immunosorbent assay, immunochromatography, immunoturbidimetry, and chemiluminescence.

The present invention does not specifically limit the reagent for detecting a marker to the reagent, and any known reagent for detecting a corresponding marker may be used.

The embodiment of the invention also provides a super-early metabolism related fatty liver disease screening kit, which comprises the reagent for detecting the level of the marker in any embodiment.

Preferably, the kit further comprises a detector for detecting the height mass fraction.

The embodiment of the invention also provides a training method of the ultra-early metabolism related fatty liver disease prediction model, which comprises the following steps:

obtaining the individual indexes of the training samples, the detection results of the marker levels and the corresponding labeling results thereof; the markers comprise total serum protein TP and cytokeratin 18, and the individual index comprises height-mass index BMI;

inputting individual indexes of the training samples and detection results of marker levels into a pre-constructed prediction model to obtain a prediction evaluation result; the prediction model is used for predicting the disease risk or disease state of the sample ultra-early metabolism-related fatty liver disease according to the individual indexes of the sample and the detection result of the marker level;

and updating parameters of the prediction model based on the labeling result and the evaluation result.

Preferably, the cytokeratin 18 is selected from the group consisting of: at least one of keratin 18-M30 (M30) and keratin 18-M65 (M65).

Preferably, the marker further comprises: alanine aminotransferase ALT, aspartate aminotransferase AST, glutamyltransferase GGT, hemoglobin HB, total cholesterol TC, golgi transmembrane glycoprotein 73 GP73, low density lipoprotein LDL, and at least one of fasting glucose and fasting insulin.

Preferably, the individual index further includes at least one of age and gender sex.

The labeling result and the prediction result can be disease states or disease risks of the training sample, and the disease states comprise whether the training sample has metabolism-related fatty liver diseases and the degree of the metabolism-related fatty liver diseases. The training samples comprise samples of metabolic-related fatty liver patients and samples of healthy people, the types of the samples can be specifically serum samples, whole blood samples or tissue samples, and the samples are not limited as long as the detection results of the markers can be obtained correspondingly. The sample size of the training sample is more than or equal to 10; preferably ≧ 30. The pre-constructed prediction model can be Cox proportional risk regression or a partial distribution competition risk model provided by Fine and Gray, and is used as a prediction model for predicting the disease risk or the disease state of the sample ultra-early metabolic related fatty liver disease according to the individual indexes of the sample and the detection result of the marker level after training of a training sample.

After training, the accuracy of the prediction model can be detected through the test set samples, and when the accuracy reaches a set threshold value, the final prediction model is obtained.

The embodiment of the invention also provides a prediction device of the ultra-early metabolic related fatty liver disease, which comprises an acquisition module and a prediction module.

The acquisition module is used for acquiring individual indexes of a sample to be detected and detection results of the levels of the markers, wherein the markers and the individual indexes are the markers and the individual indexes in any embodiment;

and the prediction module is used for inputting the individual indexes and the detection results into a prediction model obtained by training the training method of the ultra-early metabolic related fatty liver disease prediction model in any embodiment to obtain the prediction results of the sample to be detected.

In a preferred embodiment, the prediction model obtains the prediction result of the sample through at least one of formula 1-formula 12;

formula 1: log it (P) = a + b × age + c × sex + d × BMI + e₁×M30+f×TC +g×HB + j×ALT + h×TP + i×GP73；

Formula 2: log it (P) = a + b × age + c × sex + d × BMI + e₁×M30+f×TC +g×HB + k×AST + h×TP + i×GP73；

Formula 3: log it (P) = a + d × BMI + e₁×M30+e₂×M65 + h×TP；

Formula 4: log it (P) = a + d × BMI + e₁×M30 + h×TP；

Formula 5: log it (P) = a + d × BMI + e₁×M30 + h×TP + l×LDL + n×DM3；

Formula 6: log it (P) = a + d × BMI + e₁×M30 + h×TP +n×DM3；

Formula 7: log it (P) = a + b × age + c × sex + d × BMI + e M30-M65 + l × LDL + g × HB + j × AST + h × TP + i × GP 73;

formula 8: log it (P) = a + b × age + c × sex + d × BMI + e × M30-M65 + l × LDL + n × DM3+ g × HB + k × AST + h × TP + i × GP73

Formula 9: log it (P) = a + d × BMI + e₂×M65 + h×TP；

Formula 10: log it (P) = a + d BMI + e₂×M65 + h×TP + n×DM3；

Formula 11: log it (P) = a + d × BMI + e₂×M65 + h×TP + l×LDL + n×DM3；

Formula 12: log it (P) = a + d × BMI + e × M30-M65 + h × TP;

in the formulas 1-12, BMI is height quality index; m30 is the level of keratin 18-M30 in serum, M65 is the level of keratin 18-M65 in serum, specifically, the level of M30 or M65 refers to the mass unit U of the keratin 18-M30 fragment in each liter of sample, each U corresponds to 24pmol, and can refer to the content or concentration of the marker in the sample, and also refers to the specific value of the marker; M30-M65: when M30 is more than or equal to 150U/L or M65 is more than or equal to 250U/L, M30-M65 is 0.5-1.5, and when M30 is less than 150U/L and M65 is less than 250U/L, M30-M65 is 0-0.3; TC is total cholesterol level, HB is hemoglobin concentration, TP is serum total protein concentration, GP73 is serum or plasma golgi transmembrane glycoprotein 73 concentration, ALT is serum or plasma glutamic-pyruvic transaminase concentration, AST is serum or plasma glutamic-oxaloacetic transaminase concentration, LDL is serum, plasma or tissue low density lipoprotein concentration, DM3 is diabetes index: DM3 is 0.5-1.5 if there is a history of diabetes or diabetes, or DM3 is 0-0.3 if there is no diabetes; a to n are all constants. Whether the patient has the diabetes can be obtained based on fasting blood glucose detection, the fasting blood glucose is a marker corresponding to whether the patient has the diabetes, and fasting insulin provides a basis for judging whether the patient has the diabetes and distinguishing the type of the diabetes.

Preferably, the prediction model includes a step of judging whether the sample has or is at risk of the metabolism-related fatty liver disease according to the value of P, and the judgment criteria are as follows:

if P is more than 0.5, the sample is positive, and if P is less than or equal to 0.5, the sample is negative.

Note that log it (P) = ln

。

Preferably, the value of a can be-18 to-8, such as any one or a range between any two of-8, -9, -10, -11, -15 and-18;

the value of b can be-0.03 to-0.006, such as any one or a range value between-0.006, -0.008, -0.010, -0.012, -0.014, -0.016, -0.018, -0.020 and-0.030;

the value of c can be 0.20-1.50, for example, any one or a range between any two of 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, 1.00, 1.10, 1.20, 1.30, 1.40 and 1.50;

d can take the value of 0.10-0.50, such as any one or a range value between any two of 0.10, 0.20, 0.30, 0.40 and 0.50;

e₁or e₂The value of (b) can be 0.002-0.009, such as any one or a range between any two of 0.002, 0.003, 0.004, 0.005, 0.006, 0.007, 0.008 and 0.009;

e can take the value of 1.2-1.8, such as 1.2, 1.3, 1.4, 1.5, 1.6, 1.7 and 1.8;

f can be 0.05-0.50, such as any one or a range of any two of 0.05, 0.10, 0.15, 0.20, 0.25, 0.30, 0.35, 0.40, 0.45 and 0.50;

g can be-0.01-0.04, such as any one or a range of any two of-0.01, 0, 0.01, 0.02, 0.03 and 0.04;

the value of h can be 0.05-0.09, such as any one or a range between any two of 0.05, 0.06, 0.07, 0.08 and 0.09;

the value of i can be 0.002-0.010, such as any one or a range value between any two of 0.002, 0.003, 0.004, 0.005, 0.006, 0.007, 0.008, 0.009 and 0.010;

j can take the value of 0.00005-0.005, such as any one or a range between any two of 0.00005, 0.00007, 0.001, 0.0015, 0.0020, 0.0025, 0.0030, 0.0035, 0.0040, 0.0045 and 0.0050;

k can be-0.001-0.006, such as any one or a range of any two of-0.001, 0.002, 0.003, 0.004, 0.005 and 0.006;

l can be 0.05-0.15, such as any one or a range between any two of 0.05, 0.06, 0.07, 0.08, 0.09, 0.10, 0.11, 0.12, 0.13, 0.14 and 0.15;

m can be from-0.001 to-0.0005, for example, any one or a range between any two of-0.001, -0.0009, -0.0008, -0.0007, -0.0006, and-0.0005;

the value of n may be 0.05 to 0.50, for example, 0.05, 0.10, 0.15, 0.20, 0.25, 0.30, 0.35, 0.40, 0.45 and 0.50 or a range between any two thereof.

Specifically, formula 1 may correspond to subsequent embodiments 2, 3, and 5, formula 2 may correspond to embodiments 4 and 6, formula 3 may correspond to embodiments 7 and 8, formula 4 may correspond to embodiments 9 and 10, formula 5 may correspond to embodiment 11, formula 6 may correspond to embodiment 12, formula 7 may correspond to embodiments 13, 14, and 23, formula 8 may correspond to embodiments 1 and 22, formula 9 may correspond to embodiments 15 and 16, formula 10 may correspond to embodiment 17, formula 11 may correspond to embodiments 18 and 19, and formula 12 may correspond to embodiments 20 and 21.

The embodiment of the invention also provides a training device for the ultra-early metabolic related fatty liver disease, which comprises:

the acquisition module is used for acquiring the detection results of the individual indexes and the marker levels of the training samples and the corresponding labeling results; the marker and the individual index are both the marker and the individual index described in any of the preceding embodiments;

the prediction module is used for inputting the individual indexes of the training samples and the detection results of the marker levels into a pre-constructed prediction model to obtain a prediction evaluation result; the prediction model is used for predicting the disease risk or disease state of the sample ultra-early metabolism-related fatty liver disease according to the individual indexes of the sample and the detection result of the marker level;

and the parameter updating module is used for updating parameters of the prediction model according to the labeling result and the evaluation result.

The embodiment of the invention also provides the electronic equipment, which comprises a processor and a memory; the memory is used for storing a program which, when executed by the processor, causes the processor to implement the training method for the ultra-early metabolic-related fatty liver disease prediction model according to any of the preceding embodiments.

The Memory may be, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Programmable Read Only Memory (PROM), an Erasable Read Only Memory (EPROM), an electrically Erasable Read Only Memory (EEPROM), and the like.

The processor may be an integrated circuit chip having signal processing capabilities. The Processor may be a general-purpose Processor including a Central Processing Unit (CPU), a Network Processor (NP), etc.; but also Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components.

In practical applications, the electronic device may be a server, a cloud platform, a mobile phone, a tablet computer, a notebook computer, an ultra-mobile personal computer (UMPC), a handheld computer, a netbook, a Personal Digital Assistant (PDA), a wearable electronic device, a virtual reality device, and the like, and therefore, the embodiment of the present application does not limit the type of the electronic device.

Furthermore, an embodiment of the present invention further provides a computer-readable medium, where a computer program is stored on the computer-readable medium, and when the computer program is executed by a processor, the method for training a prediction model of a very early metabolic-related fatty liver disease according to any of the foregoing embodiments is implemented.

The computer readable medium may include: u disk, removable hard disk, read only memory, random access memory, magnetic or optical disk, etc. for storing program codes.

The features and properties of the present invention are described in further detail below with reference to examples.

Example 1

A construction method of a prediction model of the ultra-early metabolism-related fatty liver disease is as follows.

(1) Confirming a training set and a testing set, wherein samples in the training set are 456 cases, 380 cases of patients and 76 cases of healthy people; the test set was 354 samples, 296 patients and 58 healthy people. The healthy people refer to the population with the sum of four pathological scores I less than or equal to 2, and the patients refer to the population with the sum of I more than 2. (2) And detecting the markers and the individual indexes of the sample.

The individual indexes include: height mass index BMI, age and gender sex.

The enzymatic method is adopted to detect alanine aminotransferase ALT and aspartate aminotransferase AST. The detection range is 5-700U/L. The reference ranges are: 0-40U/L.

The substrate method was used to detect gamma-glutamyl transferase (GGT). The detection range is 0-500U/L. The reference ranges are: 0-50U/L.

HB was detected by a full-automatic blood cell analyzer by a colorimetric method. The reference ranges are: 110-160g/L for adults.

Total cholesterol TC was detected by an enzymatic method. The detection range is as follows: 1.13-12.93 mmol/L, reference range: male, 3.2-7 mmol/L; female, 3.2-6.3 mmol/L.

Golgi protein 73 (GP 73) was detected by up-conversion luminescence. The detection range of the reagent is as follows: 20-1000 ng/ml; reference ranges: GP73 is more than or equal to 85ng/ml, can be used for diagnosing significant hepatic fibrosis of HBV infected patients (moderate or severe liver injury), GP73 is more than or equal to 135ng/ml, and can be used for diagnosing liver cirrhosis.

TP levels were determined by the biuret method. The detection kit comprises: cholesterol ester hydrolase is more than or equal to 0.8KU/L, cholesterol oxidase is more than or equal to 0.5KU/L, peroxidase is 1500U/L, 4-AAP is 0.4mmol/L, DHBS is 2mmol/L, and phosphate buffer solution (pH7.6) is 400 mmol/L. The instrument comprises the following steps: hitachi model 7600 full-automatic biochemical analyzer. The main analytical parameters were: the temperature is 37 ℃, the wavelength is 540nm, and the reaction time is 600 s; and (4) calculating a result: total protein (g/l) = sample absorbance/calibration absorbance × calibrator concentration.

The enzyme-linked immunosorbent assay is adopted to detect the level of human CK18-M30 (cytokeratin 18-M30). The detection kit comprises: the kit comprises a standard substance, a quality control substance, an enzyme-labeled secondary antibody, an antibody diluent, a concentrated lotion, a display solution and a stop solution. The instrument can be a microplate reader. Linear range: 0-1000U/L. Reference ranges: 0-150U/L.

The enzyme-linked immunosorbent assay is adopted to detect the level of human CK18-M65 (cytokeratin 18-M65). The detection kit comprises: the kit comprises a standard substance, a quality control substance, an enzyme-labeled secondary antibody, an antibody diluent, a concentrated lotion, a display solution and a stop solution. The instrument comprises the following steps: kowa ST-360 microplate reader. Linear range: 0-5000U/L; reference ranges: 0-250U/L.

LDL levels were measured using a two-point endpoint method. The detection kit comprises: HDAOS 0.56 mmo/L, cholesterol ester hydrolase 0.8kU/L, cholesterol oxidase 507 IU/L, peroxidase 4kU/L, 4AAP 4mmol/L, Goods buffer (pH7.0) 100 mmol/L. The detection instrument can adopt a Hitachi 7600 type full-automatic biochemical analyzer. The main analytical parameters include: 3 mul of sample; reagent 1: 225 μ l; reagent 2: 75. mu.l, reaction temperature: 37 ℃; dominant wavelength 600, subwavelength 700; the reaction direction is as follows: forward direction; reading points 0/27(Point 1Fst/Lst), 0/10(Point 2Fst/Lst), Reagent Fst L-0.1 Fst H0.1, Lst L-0.5, Lst H0.1; instrumental factor a was 1, B was 0, calibration type AB. And (4) calculating a result: serum LDL-C concentration (mmol/L) specimen a; (600-700) nm multiplied by the cholesterol concentration (mmol/L) of the calibration solution/the standard solution A (600-700) nm. The instrument automatically gives the measurement result of each sample according to the formula. Linear range: 0.26 to 25.8 mmol/L. Reference ranges: adult: less than or equal to 3.12 mmol/L.

(3) Obtaining the individual indexes of the training samples, the detection results of the marker levels and the corresponding labeling results thereof;

and updating parameters of the prediction model based on the labeling result and the evaluation result to obtain a trained prediction model.

The embodiment also provides a prediction method of the ultra-early metabolism-related fatty liver disease, which comprises the following steps.

The acquisition module is used for acquiring the individual indexes of the sample to be detected and the detection result of the marker level, and the marker comprises: LDL, DM, HB, TP and GP 73;

and the prediction module is used for inputting the height quality index and the detection result into a trained prediction model to obtain a prediction result of the sample to be detected.

The formula of the prediction model is as follows:

logit（P）=-10.7775 - 0.0150×age+0.4151×sex+0.2575×BMI+1.4747×M30-M65 + 0.1160×LDL+0.2384×DM3 - 0.00408×HB+0.00282×AST + 0.0656×TP + 0.00542×GP73。

in the formula:

DM3 is an index of diabetes, DM3=1 if there is a history or presence of diabetes, and DM3=0 if there is no diabetes.

Values of the sex: setting male as 1 and female as 2; in other embodiments, the values of gender may be reversed, and the associated coefficients may differ.

M30-M65: when M30 is more than or equal to 150U/L or M65 is more than or equal to 250U/L, M30-M65=1, and when M30 is less than 150U/L and M65 is less than 250U/L, M30-M65= 0; in other embodiments, M30 is the serum keratin 18-M30 level (in U/L, 1U corresponds to 24 pmol), M65 is the serum keratin 18-M65 level (in U/L), TP is the total serum protein concentration (in g/L), LDL is the low density lipoprotein concentration (in mmol/L), AST is the glutamic-oxaloacetic transaminase concentration (in U/L), and GP73 is the serum or plasma Golgi transmembrane glycoprotein 73 concentration (in ng/ml). In other embodiments, GGT is the glutamyl transferase concentration of serum (in U/L) and TC is total cholesterol (in mmol/L).

After the P value is calculated, if P is more than 0.5, the sample is positive, and if P is less than or equal to 0.5, the sample is negative.

Example 2

A method for predicting a very early metabolic-related fatty liver disease, which is substantially the same as in example 1, except that the formula of the prediction model is different from that of example 1, and the formula is as follows:

logit（P）=-10.9251- 0.00688×age + 0.2864×sex+0.2636×BMI+0.00751×M30 +0.1384×TC -0.00958×HB + 0.00425×ALT + 0.0634×TP + 0.00344×GP73。

example 3

A method for predicting a very early metabolic-related fatty liver disease, which is substantially the same as in example 1, except that the formula of the prediction model is different from that of example 1, specifically as follows:

log it（P）= -10.9269 - 0.00711×age + 0.2843×sex + 0.2647×BMI + 0.00753×M30 + 0.1377×TC - 0.00971×HB + 0.00423×ALT + 0.0635×TP + 0.00347×GP73。

example 4

logit（P）=-10.4957 - 0.0108×age + 0.3095×sex + 0.2543×BMI + 0.00788×M30 + 0.1462×TC - 0.00840×HB + 0.00194×AST + 0.0607×TP + 0.00419×GP73。

example 5

logit（P）=-15.8634-0.0173×age +1.2801×sex +0.1388×BMI+0.00311×M30 +0.0879×TC +0.0271×HB +0.000075×ALT +0.0695×TP +0.00977×GP73。

example 6

logit（P）=-10.4946 - 0.0106×age + 0.3112×sex + 0.2534×BMI + 0.00786×M30 + 0.1468×TC - 0.00828×HB + 0.00196×AST + 0.0606×TP + 0.00415×GP73。

example 7

logit（P）=-10.4753 +0.2435×BMI +0.00719×M30 +0.000908×M65 +0.0602×TP。

example 8

logit（P）=-10.4515 +0.2425×BMI +0.00718×M30 +0.000904×M65 +0.0603×TP。

example 9

logit（P）=-10.5759 +0.2442×BMI +0.00827×M30 +0.0617×TP。

example 10

logit（P）=-10.5466 + 0.2432×BMI + 0.00825×M30 + 0.0616×TP。

example 11

logit（P）=-10.9791 +0.2402×BMI +0.00822×M30 +0.1452×DM3 +0.0637×TP +0.1064×LDL。

example 12

logit（P）=-10.3734 +0.2361×BMI +0.00837×M30 +0.0699×DM3 +0.0614×TP。

example 13

logit（P）=-10.3524 -0.0138×age +0.4371×sex +0.2547×BMI +1.5034×M30-M65 +0.0864×LDL -0.00430×HB +0.00237×AST +0.0620×TP +0.00592×GP73。

example 14

logit（P）=-10.3410 - 0.0136×age + 0.4375×sex + 0.2544×BMI + 1.4982×M30-M65 + 0.0864×LDL - 0.00425×HB + 0.00239×AST + 0.0618×TP + 0.00591×GP73。

example 15

logit（P）=-10.0260 +0.2445×BMI +0.00411×M65 +0.0585×TP。

example 16

logit（P）=-10.0027 + 0.2434×BMI + 0.00411×M65 + 0.0585×TP。

example 17

logit（P）=-9.9289 +0.2378×BMI +0.00412×M65 +0.1063×DM3 +0.0591×TP。

example 18

logit（P）=-10.4786 +0.2411×BMI +0.00405×M65 +0.1803×DM3 +0.0612×TP +0.0973×LDL。

example 19

logit（P）=-10.7614 +0.2493×BMI +1.6834×M30-M65 +0.2477×DM3 +0.0648×TP +0.1059×LDL。

example 20

logit（P）=-10.3762 +0.2541×BMI +1.6675×M30-M65 +0.0631×TP。

example 21

logit（P）= -10.3471 +0.2537×BMI +1.6625×M30-M65 +0.0628×TP。

example 22

logit（P）= -17.1744 -0.0262×age+1.4266×sex+0.1485×BMI +1.3034×M30-M65 +0.271×LDL+0.6099×DM3+0.0300×HB+0.00686×AST+0.0762×TP+0.00903×GP73。

example 23

logit（P）= -16.1721 -0.0219×age+1.4268×sex+0.1477×BMI +1.3420×M30-M65 +0.1471×LDL+0.0283×HB-0.00003×AST+0.0684×TP+0.00975×GP73。

comparative example 1

Y=-12.037+1.525×M30+1.488×sex+0.701×TC-0.586×LDL+0.096×BMI-0.048×age+0.032×HB+0.004×ALT-0.004×GGT。

the formula is too large to calculate because the value of P is calculated according to logit (P). Therefore, the result value is still calculated by adopting the original formula.

Test examples

Based on the prediction methods provided in examples 1-23 and comparative example 1, the ultra-early prediction was performed on 354 samples, wherein 296 patients and 58 healthy people. Healthy people refer to people with the sum of four pathological scores I less than or equal to 2, patients refer to people with I greater than 2, ROC curves are drawn based on detection results, and prediction results of prediction methods provided in examples 1-23 and comparative example 1 are sequentially shown in figures 1-24.

The AUC results are summarized in table 1.

TABLE 1 AUC results

Examples of the experiments	Example 1	Example 2	Example 3	Example 4	Example 5	Example 6
							AUC	0.8134	0.8361	0.8370	0.8302	0.8199	0.8293
Examples of the experiments	Example 7	Example 8	Example 9	Example 10	Example 11	Example 12
							AUC	0.8221	0.8213	0.8209	0.8201	0.8203	0.8194
Examples of the experiments	Example 13	Example 14	Example 15	Example 16	Example 17	Example 18
							AUC	0.8149	0.8139	0.8115	0.8107	0.8082	0.8080
Examples of the experiments	Example 19	Example 20	Example 21	Example 22	Example 23	Comparative example 1
							AUC	0.8024	0.8046	0.8036	0.8156	0.8113	0.7242

As can be seen from the results, the area under the ROC curve (AUC) of examples 1-22 was > 0.8, while the AUC of comparative example 1 was 0.7242.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A training method of a prediction model of ultra-early metabolism-related fatty liver disease is characterized by comprising the following steps:

inputting individual indexes of the training samples and detection results of marker levels into a pre-constructed prediction model to obtain a prediction evaluation result; the prediction model is used for predicting the disease risk or disease state of the sample ultra-early metabolism-related fatty liver disease according to the individual indexes of the sample and the detection result of the marker level; and updating parameters of the prediction model based on the labeling result and the evaluation result.

2. The method for training a prediction model of ultra-early metabolic-related fatty liver disease according to claim 1, wherein the cytokeratin 18 is selected from the group consisting of: at least one of keratin 18-M30 and keratin 18-M65.

3. The method for training a prediction model for ultra-early metabolic-related fatty liver disease according to claim 1, wherein the markers further comprise: alanine aminotransferase ALT, aspartate aminotransferase AST, glutamyltransferase GGT, hemoglobin HB, total cholesterol TC, golgi transmembrane glycoprotein 73 GP73, low density lipoprotein LDL, fasting glucose, and fasting insulin.

4. The training method for the ultra-early metabolic-related fatty liver disease prediction model according to any one of claims 1 to 3, wherein the individual index further comprises at least one of age and gender sex.

5. A prediction device for ultra-early metabolism-related fatty liver disease, comprising:

an obtaining module, configured to obtain individual indicators of a sample to be detected and detection results of levels of markers, where the markers and the individual indicators are the markers and the individual indicators according to any one of claims 1 to 4;

a prediction module, configured to input the individual index and the detection result into a prediction model obtained by training the training method of the ultra-early metabolic related fatty liver disease prediction model according to any one of claims 1 to 4, so as to obtain a prediction result of a sample to be tested.

6. The apparatus for predicting very early metabolic-related fatty liver disease according to claim 5, wherein the prediction model obtains the prediction result of the sample by at least one of formulas 1 to 12;

formula 1: logic (p) = a + b × age + c × sex + d × BMI + e₁×M30+f×TC +g×HB + j×ALT + h×TP + i×GP73；

Formula 2: logic (p) = a + b × age + c × sex + d × BMI + e₁×M30+f×TC +g×HB + k×AST + h×TP + i×GP73；

Formula 3:logit（P） =a+ d×BMI +e₁×M30+e₂×M65 + h×TP；

formula 4: logic (p) = a + d × BMI + e₁×M30 + h×TP；

Formula 5: logic (p) = a + d × BMI + e₁×M30 + h×TP + l×LDL + n×DM3；

Formula 6: logic (p) = a + d × BMI + e₁×M30 + h×TP +n×DM3；

Formula 7: logic (p) = a + b × age + c × sex + d × BMI + e M30-M65 + l × LDL + g × HB + j × AST + h × TP + i × GP 73;

formula 8: logic (P) = a + b × age + c × sex + d × BMI + e × M30-M65 + l × LDL + n × DM3+ g × HB + k × AST + h × TP + i × GP73

Formula 9: logic (p) = a + d × BMI + e₂×M65 + h×TP；

Formula 10: logic (p) = a + d BMI + e₂×M65 + h×TP + n×DM3；

Formula 11: logic (p) = a + d × BMI + e₂×M65 + h×TP + l×LDL + n×DM3；

Formula 12: logic (p) = a + d × BMI + e × M30-M65 + h × TP;

in the formulas 1-12, BMI is height quality index; m30 is the level of keratin 18-M30 in serum, M65 is the level of keratin 18-M65 in serum, M30-M65: when M30 is more than or equal to 150U/L or M65 is more than or equal to 250U/L, M30-M65 is 0.5-1.5, and when M30 is less than 150U/L and M65 is less than 250U/L, M30-M65 is 0-0.3; TC is total cholesterol level, HB is hemoglobin concentration, TP is serum total protein concentration, GP73 is serum or plasma golgi transmembrane glycoprotein 73 concentration, ALT is serum or plasma glutamic-pyruvic transaminase concentration, AST is serum or plasma glutamic-oxaloacetic transaminase concentration, LDL is serum, plasma or tissue low density lipoprotein concentration, DM3 is diabetes index: DM3 is 0.5-1.5 if there is a history of diabetes or diabetes, or DM3 is 0-0.3 if there is no diabetes; a to n are all constants.

7. The device for predicting the very early metabolic-related fatty liver disease according to claim 6, wherein a is-18 to-8, b is-0.03 to-0.006, c is 0.20 to 1.50, and d is 0.10 to 0.50; e.g. of the type₁Or e₂0.002 to 0.009, e 1.2 to 1.8, f 0.05 to 0.50, g-0.01 to 0.04, h 0.05 to 0.09, i 0.002 to 0.010, j 0.00005 to 0.005, k-0.001 to 0.006, l 0.05 to 0.15, m-0.001 to-0.0005, n 0.05 to 0.50.

8. A training device for ultra-early metabolism-related fatty liver diseases is characterized by comprising:

the acquisition module is used for acquiring the detection results of the individual indexes and the marker levels of the training samples and the corresponding labeling results; the marker and the individual index are both the marker and the individual index described in any one of claims 1 to 4;

9. An electronic device, comprising a processor and a memory; the memory is used for storing a program, and when the program is executed by the processor, the program causes the processor to realize the training method of the ultra-early metabolism-related fatty liver disease prediction model according to any one of claims 1 to 4.

10. A computer-readable medium, wherein a computer program is stored on the computer-readable medium, and when executed by a processor, the computer program implements the method for training the ultra-early metabolic-related fatty liver disease prediction model according to any one of claims 1-4.