Objective: 1. Explore the differential metabolites and metabolic pathways between different interstitial lung disease (ILD) and healthy controls, so as to find out new biomarkers and metabolic pathways of ILD and provide new biological targets for IPF diagnosis and treatment. 2. Detect the levels of serum related fibrosis and metabolic indexes of ILD, explore the relationship between these indexes and the types and general parameters of ILD diseases, and focus on GDF-15 on the basis of metabonomics, explore new metabolic pathways related to ILD, and further expound the pathogenesis of idiopathic pulmonary fibrosis. Method: 1. According to the inclusion criteria and exclusion criteria, the ILD patients admitted to our hospital from January 1, 2021 to January 1, 2022 were signed with informed consent, and all of them had clear multidisciplinary diagnosis. At the same time, the physical examination in the health management center of our hospital was selected as the healthy control group. General clinical data of ILD patients and healthy controls were collected, including gender, age, onset, clinical manifestations, clinical biochemistry, lung function and other related data. Collect patients' peripheral venous blood, keep serum samples, and analyze the serum of ILD patients and healthy controls by liquid chromatography-mass spectrometry (LC-MS) to obtain differential metabolites, metabolic maps and related data. According to the diagnosis, the patients were divided into Idiopathic pulmonary fibrosis (IPF), connective tissue disease-associated interstitial lung disease (CTD-ILD), other interstitial lung disease (ILD) patients and healthy control group. R statistical software was used to observe whether there are differential metabolites in IPF group and healthy control group, CTD-ILD group and healthy control group, other ILD groups and healthy control group, IPF group and CTD-ILD group, IPF group and other ILD groups, CTD-ILD group and other ILD groups by using positive and negative ion flow charts and score charts of orthogonal partial least squares-discriminant analysis. And the potential biomarkers were found through the S-plot diagram of OPLS-DA and the variable weight value VIP>1. ROC curve analysis of differential metabolites was carried out by SPSS21.0 software to analyze the value of differential diagnosis and find out new biomarkers with differential diagnosis significance. Based on KEGG database, the metabolic pathway enrichment analysis of differential metabolites was carried out to find out the main metabolic pathways. 2. Serum levels of type I collagen, KL-6, IL-1β, TNF-α, IGF-1 and GDF-15 were detected by Enzyme linked immunosorbent assay (ELISA) in all patients and healthy controls. Using SPSS21.0 statistical software and R statistical software, using T-test, variance analysis or nonparametric test method, according to different diagnoses (IPF/CTD-ILD/ other ILD/ healthy control group, ILD/ healthy control group, IPF/ other), the data were analyzed and compared in subgroups. Pearson correlation analysis was used to analyze the correlation between serum biomarkers and other factors such as lung function, metabolomics and so on. ROC curve was used to analyze the diagnostic efficiency of each index, and the best critical value, sensitivity and specificity were found out. Results: 1. Analysis of general clinical data and clinical indicators: 26 cases of IPF, 21cases of CTD-ILD, 23 cases of others ILD and 20 cases of healthy control group were included, respectively. There were significant differences in age, sex and smoking history among the four groups (P < 0.05), and the differences were statistically significant. There were statistically significant differences among IPF group, CTD-ILD group and other ILD groups in cough and expectoration symptoms, eosinophil count percentage, creatine kinase, creatine kinase isoenzyme, C-reactive protein and carbon monoxide diffusion capacity (DLco) (P < 0.05). 2. Metabonomics analysis: LC-MC total ion flow chart of serum samples and orthogonal partial least squares discriminant analysis (OPLS-DA) of serum samples showed the difference of ion peak intensity, and there were different metabolites among each group. The OPLS-DA model has good quality evaluation and reliable data analysis. There are 193, 115, 101, 188, 148 and 169 differential metabolites in IPF group and healthy control group, IPF group and CTD-ILD group, CTD-ILD group and other ILD groups, others ILD groups and healthy control group respectively. There is a correlation between the differential metabolites of each ILD subgroup and the healthy control group. The most relevant metabolic pathways in IPF mainly include choline metabolic pathway, metabolic pathway and linoleic acid metabolic pathway in cancer. The metabolic pathways most related to CTD-ILD and other ILD include choline metabolic pathway and retrograde neural signal metabolic pathway in cancer. Metabolic pathway and linoleic acid metabolic pathway may distinguish IPF from other ILD. Metabolites involved in metabolic pathway are L-Glutamate, PE(18:3(6Z,9Z,12Z)/P-18:0) and PC (18: 2 (9z, 12z)/20: 2 (11z, 14z)), respectively. There are three differential metabolites of linoleic acid pathway, namely 12,13-EpOME, 13S-HODE and PC(18:2(9Z,12Z)/20:2(11Z,14Z)). There are 35 kinds of differential metabolites in IPF group and healthy control group, and the top three places of AUC are 3- sulfodeoxycholic acid, 20,22-dihydrodigitalis glycoside, (±14) (15)-EET-SI, with AUC values of 0.985, sensitivity of 100% and specificity of 92.3%. L- acetylcarnitine, 5-hydroxydodecanoate and L-Glutamate are involved in significant enrichment pathways, among which the ROC curve area of L-Glutamate is the largest, with AUC value of 0.921, cutoff value of 40145.490, sensitivity of 80%, specificity of 96%, and participation in metabolic pathways. 3. Serum metabolism-related detection indexes: The six serum detection indexes of IPF group and CTD-ILD group are higher than those of healthy control group, and the difference is statistically significant (P < 0.05), but the difference between IPF group and CTD-ILD group is not statistically significant (P>0.05). The expression of GDF-15 in IPF group and non-IPF ILD group was higher than that in healthy control group, with significant difference, but there was no significant difference between IPF group and non-IPF ILD group. Type I collagen and IL-1β; were negatively correlated with DLco. KL-6 is negatively correlated with the measured values of VC and DLco; IGF-1 was negatively correlated with FVC, VC and DLco (all P