RETRACTED ARTICLE: An interpretable machine learning model for precise prediction of biomarkers for intermittent fasting pattern Authors: Xiaoli Hu, Qingjun Xu, Xuan Ma, Lin Li, Yongning Wu, Feifei Sun Published: 2024 ABSTRACT Intermittent fasting is currently a highly sought-after dietary pattern. To explore the potential biomarkers of intermittent fasting, untargeted metabolomics analysis of fecal metabolites in two groups of mice, intermittent fasting and normal feeding, was conducted using UPLC-HRMS. The data was further analyzed through interpretable machine learning (ML) to data mine the biomarkers for two dietary patterns. We developed five machine learning models and results showed that under three-fold cross-validation, Random Forest model was the most suitable for distinguishing the two dietary patterns. Finally, Shapely Additive exPlanations (SHAP) were explored to perform a weighted explanatory analysis on the Random Forest model, and the contribution of each metabolite to the model was calculated. Results indicated that Ganoderenic Acid C is the potential biomarkers to distinguish the two dietary patterns. Our work provides new insights for metabolic biomarker analysis and lays a theoretical foundation for the selection of a healthieir dietary lifestyle.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12986-024-00876-y. Introduction Intermittent fasting becomes quite popular and attractive in recent years, both in daily human life and in clinical research, since intermittent fasting life style not only facilitate weight control, but seems to have a close correlation with mammalian functions such as metabolism [19] and alleviate certain diseases such as diabetes [17], heart diseases [6], cancers [7], etc. Intermittent fasting, in simple terms, is the alternation of feeding and fasting periods. The common intermittent fasting ways are: alternate-day fasting (alternating fasting days with ad libitum days); 5:2 fasting (fasting on 2 days per week, alternating with 5 ad libitum days); and time-restricted feeding (TRF, including 4-TRF, 6-TRF and 8-TRF) [4]. Various studies demonstrated that intermittent fasting could affect mammalian physiology by altering and improving mammalian immune and intestinal flora [1], lifestyle and circadian rhythms [19], and blood glucose and insulin levels [17], leading to the alleviation and treatment of certain diseases. Among the three modes of light fasting, TRF patterns have been widely studied due to the fact that they do not require excessive hunger tolerance and much easier to perform. It’s been verified that the TRF had significant effects on cardio-metabolic problems caused by ad libitum feeding in rodents and has a moderate effect on cardio-metabolism and obesity in obese people. The TRF shows significant positive effects on cardio-metabolic dysfunction, weight gain and hepatic steatosis in rodents caused by ad libitum dietary, and it also exhibits moderate effects on cardio-metabolism and obesity in obese people [20]. Metabolomics is a promising technique to catalog and quantify the myriad small molecules found in biological fluids under different conditions, which is an important tool to systematically identify and quantitate all metabolites in a variety of fields, from ecology to cancer [9]. Mass spectrometry-based (MS-based) metabolomics is the important technology to identify the metabolites [2]. Metabolomics include targeted metabolomics and non-targeted metabolomics. Targeted metabolomics enables qualitative and quantitative analysis of specific compounds or pathways. Non-targeted metabolomics measures all measurable compounds in the biological samples, and imports the measured substances into a database or computer library for analysis. It is mainly used to determine the differential metabolites and possible biomarkers under different conditions [21]. However, the credibility of metabolomics is still questioned due to factors such as different sample preparation steps, different metabolite standards, and the lack of clear interpretable standards [2]. Untargeted metabolomics generates large amounts of data, making it difficult to determine the underlying associations. The vigorous development of computer science and statistics enables satisfactory alternatives to interprete the omics-datasets [11]. Nowadays, MS-based metabolomics promotes the development in many aspects, such as the key steps of Alzheimer’s disease [14] and various aspects of aging-associated metabolic features [25]. Machine learning is a branch of artificial intelligence and belongs to the interdisciplinary field of statistics and computer science [5]. Machine learning is capable of processing and making connections with large amounts of data. The operator needs to provide a large amount of known data and continuously optimize it based on the known framework models so that the established model can continuously learn. The optimized model is used to predict the potential features or dig out biomarkers based on the provided data [8]. In particular, machine learning’s feature of mining connections from large amounts of data is very suitable for the large data sets of untargeted metabolomics, so machine learning is widely used in metabolomics analysis. Bo Shen et al. combined targeted metabolomics and machine learning techniques to input serum metabolites of COVID-19 patients for classification verification. The results showed that the serum metabolites of COVID-19 patients produced significant changes and were distinguished by biomarkers. COVID-19 patients have significant effects compared to healthy people [23]; Hiroshi Kuwabara et al. collected a large number of cases and used a decision tree algorithm to determine the correlation between biomarkers in saliva and rectal cancer [12]. Liang Liang et al. also used machine learning methods to determine the relationship between the species metabolite content in the urine of pregnant women and the delivery period of pregnant women [13]. In the present study, we used several bioinformatics techniques to verify the relationship between mouse fecal metabolites and intermittent fasting, and several machine learning models were established to predict characteristic biological metabolites, and the most ideal model was selected to predict the promising biomarker. The schematic diagram for this work is shown in Fig. 1. The obtained results could provide new aspects to interpret metabolomic datasets. Fig. 1Schematic diagram of experimental operations and data analysis Schematic diagram of experimental operations and data analysis Materials and methods Sample preparation Sixteen 4-week-old female C57BL/6J mice were bought from Hangzhou Ziyuan Experimental Animal Science and Technology Co. Ltd. The mice were acclimatized for 7-day at Anhui Agricultural University and were co-housed together under the conditions of Specific Pathogen Free (SPF), controlled temperature (25 ± 2°C), and half-light/half-dark cycle. The animal studies were approved by the committee of Animal welfare and ethics, Anhui Agricultural University with the ID of AHAUXMSQ2023052. In this experiment, all mice were fed under the same conditions except for the dietary pattern. A total of sixteen mice were randomly divided into EODF group and Control group, with eight mice in each group (n = 8). The EODF group was every 16 h fasting group which is composed of 8 h feeding period and 16 h fasting period every day for 90 consecutive days. The diets and water in the control group were provided ad libitum. The feces samples were collected and 400 µL of 20% methanol was added for homogenization. The mixture was centrifuged at 300 rpm for 15 min, then supernatant was collected and stored − 80°C before untargeted metabolomics assay. The metabolites were identified using SCIEX ExionLC (Framingham, MA), an Ultra-high-performance liquid chromatography tanem SCIEX X500R Quadrupole-time of flight mass spectrometry (UHPLC-Q-TOF) equipped with an electrospray ionization (ESI) source. The chromatographic conditions are specifically added in Supporting information S1-S2, Quality control information is described in detail in the supporting information Figure. S1. Identification and enrichment analysis of differentially metabolized compounds The acquired data were visualized using the R package “ggplot 2 (3.4.0)” to show significant differential metabolites between different diets in the form of volcano plots. The data were clustered and visualized using the Microbiology website (http://www.bioinformatics.com.cn) to show the differences in the form of heat maps. Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis [10] was performed using Metaboanalyst (https://www.metaboanalyst.ca/) [18], and demonstrated alterations in fecal metabolic pathways between EODF group and control group in the form of bubble plots. Machine learning modeling and model evaluation To avoid the overfitting problem due to the excessive number of features and to enhance the generalization ability of the model, several methods were adopted to screen the features. First, the correlation coefficient between features was calculated by the correlation coefficient method, and features with higher correlation (higher than 0.8) were retained, which reduced the number of features from 595 to 105.Subsequently, the top 20% of features were selected based on the chi-square test, which reduced the number of features from 105 to 21.In addition, the selected features were normalized and scaled according to the formula Z = (X-u)/s, where X is the original feature data and u is the mean. Normalized scaling was performed, where X is the original feature data, u is the mean, and s is the standard deviation to ensure that the variance of the data in each dimension is 1 and the mean is 0 (Information on the 21 metabolites after standardized scaling is given in detail in Supporting Information Table S3). Limited by the number of samples, three-fold cross validation is used ine the establishment of all machine learning models to enhance model generalization and reduce overfitting. The 16 samples were divided into three groups that did not overlap each other (fold), with two of the groups as the training set and the other as test set, three consecutive iterations are performed to average the performance metrics (e.g., accuracy, mean square error, etc.) evaluated in all three folds to obtain the overall performance evaluation of the model. The results of binary machine learning models are plotted separately as Receive Operating Characteristic Curve (ROC curves), and the points of the calculated binary machine learning models of the three-fold are averaged to obtain the mean ROC curve and the mean Area Under the Curve (AUC) value. Corresponding coefficient of variation, recall, accuracy and f1 score were used to evaluate the robustness and reliability of the developed models. In present study, five binary classification machine learning models with simpler structures were selected for model construction, including Decision Tree, K-Nearest Neighbors, Random Forest, Support Vector Machine, Naive Bays. The models were implemented using the scikit-learn Python package in the Anaconda platform (https://anaconda.org/anaconda/conda) through Pypi version 1.5.2. Shapely additive exPlanations (SHAP) analysis The Shapely Additive exPlanations (SHAP) is an important tool to interpret the machine learning model. In the explanatory model built by SHAP, all the features are assigned a certain contribution value, which is the SHAP value of the feature to the model, and the SHAP value of each feature can be used to determine the degree of influence of the feature on the model building as well as the positive and negative influences (negative/positive). In the field of life sciences, SHAP analysis has been widely used for interpretation of machine learning models in various fields. In this experiment, we use the “shap (0.46.0)” package in Python (3.12) to analyze the Random Forest model. Results and discussion Untargeted metabolomics detects changes in fecal metabolites in mice Untargeted metabolomics is often used to identify all metabolites in a sample under two dietary patterns. In this experiment, a total of 7400 differential metabolites (VIP > 1 and p-value < 0.05) were detected in mouse feces using UPLC-HRM, among which 595 differential metabolites with significant changes were selected for subsequent data-mining of biomarkers. A good case in point, non- targeted metabolomic analysis of plasma metabolites was performed in imiquimod-induced mice, and the results showed that 37 amino acids and 40 carnitines were significantly differential metabolites, which were the potential biomarkers of psoriasis [3]. In the current study, the metabolites with significant differences were mostly lipids and lipid-like molecules and organic acids and derivatives (described in Table S4 in the Supporting Information). This indicates that lipids and lipid-like molecules and organic acids and derivatives are closely related to the dietary diet in mice. Lipids and lipid-like molecules and organic acids are involved in obesity-related metabolic pathways [16]. Alterations in fecal metabolic markers and metabolic pathways induced by intermittent fasting in mice In this experiment, sixteen mice were randomly divided into control group and EODF group, with 8 mice in each group. The top 30 metabolites with significant difference were selected for hierarchical clustering based on the untargeted metabolomics data. As shown in Fig. 2A, significant differences in fecal metabolites between intermittent fasting and normal feeding dietary styles were observed. Subsequently, KEGG metabolic pathway enrichment analysis was performed based on the differential metabolites in mouse feces. The detailed information for top 20 metabolic pathways was presented in supplementary information Table S5 and illustrated in Fig. 2B in the form of bubble plots. Riboflavin Metabolis, also known as vitamin B2, is a flavin-like compound that provides energy to the body by participating in the metabolism of sugars, fats and proteins. Several articles have shown that Riboflavin Metabolism contributes to ATP synthesis by participating in major metabolic pathways such as the tricarboxylic acid cycle and the electron transport chain. Histidine metabolism is a compound closely related to energy metabolism, which is mainly mediated by the condensation of ATP and PRPP (Ribose Phosphate Pyrophosphate) to form N-5-Ribose Phosphate. Histidine metabolism is a compound closely related to the energy metabolism of the body, which is mainly formed through the condensation of ATP and PRPP (ribulose phosphate pyrophosphate) to form N-5-phosphate ribulose 1-pyrophosphate (ribulose phosphate-ATP), which can be further synthesized. In addition to this, Histidine metabolism has a close relationship with the secretion of gastric acid, which is facilitated through the binding of the H2 receptor on the cells of gastric lining, thus helping digestion. In this study, a total of 595 metabolites with significant differences were screened from 7400 metabolites based on the parameters of p-value < 0.05 and VIP > 1. The up-regulated metabolites and down-regulated metabolites were then categorized based on log2 FoldChange and presented as volcano plots in Fig. 2C. Among the 595 significant metabolites, 324 down-regulated and 271 up-regulated metabolites were identified. Fig. 2Differences in fecal metabolites caused by intermittent fasting and normal feeding. (A) The heatmap shows top 30 representative metabolites calculated based on the correlation coefficients in the 16 groups of samples, hierarchically clustering each mouse according to the two dietary patterns. (B) KEGG enrichment pathway analysis of the top 20 metabolites. (C) Intermittent fasting-induced changes in differential metabolites, variable importance in projection (VIP) > 1, p-value < 0.05 Differences in fecal metabolites caused by intermittent fasting and normal feeding. (A) The heatmap shows top 30 representative metabolites calculated based on the correlation coefficients in the 16 groups of samples, hierarchically clustering each mouse according to the two dietary patterns. (B) KEGG enrichment pathway analysis of the top 20 metabolites. (C) Intermittent fasting-induced changes in differential metabolites, variable importance in projection (VIP) > 1, p-value < 0.05 Predicting metabolites in intermittent fasting based on machine learning In recent years, machine learning has been widely used in the fields of biomedicine and food chemistry, and has attracted a great deal of attention. In the field of biomedicine, machine learning is often used for disease prediction as well as antibiotic screening and prediction. Deep learning model has been applied to identify a novel type of antibiotic effective in killing Methicillin-Resistant Staphylococcus Aureus (MRSA) from over 12 million compounds [26]. Moreover, machine learning has been widely accepted in the many other fields, such as food chemistry for composition identification or origin traceability [22]. Somaieh Soltani et al. predicted the bitterness of 229 peptides by establishing three machine learning models [24]. The application of machine learning methods in various fields has remarkably enhanced the efficiency of data analysis and offered more novel ideas for scientific research to solve the practical dilemma. In this experiment, 595 metabolites were screened based on p-value < 0.05 and VIP > 1. However, since the number of 595metabolites was much larger than the number of samples, which would lead to overfitting of the data and make the predictions of the model meaningless, we used several feature screening methods to reduce the number of features from 595 to 21, and performed standardized scaling to avoid certain features that were too large or too small from affecting the stability of the model. In this experiment, due to limited sample data, in order to prevent the model from overfitting, the three cross-validation method was used to establish the model, where the sixteen samples were divided into three groups that do not overlap each other, and two of them were used as the training set and the other as the test set. Three consecutive iterations were performed to build the three-group model, and the performance metrics (e.g., accuracy, mean squared error, etc.) evaluated in all three folds were averaged, and the ROC curves and average ROC curves of each group were plotted to obtain the overall performance evaluation of the model. Due to the limited sample data, the use of overly complex machine learning models may lead to over-fitting of the model. In this experiment, the Random Forest (Mean AUC = 0.99, Mean Accuracy = 0.94) (Fig. 3A) and the Support Vector Machine (Mean AUC = 0.99, Mean Accuracy = 0.94) (Fig. 3B) models have the highest area under the curve, i.e., AUC. AUC is an important parameter for judging the prediction rate of machine learning, the larger the AUC, the higher the prediction rate of the model. In machine learning models, Precision, Recall, Accuracy, and f1-source are four important metrics for evaluating models. In general, when all four metrics are high and remain in equilibrium, this model will be considered to be ideal. In this experiment, all the five machine learning models we selected have better performance, while the Random Forest and the Support Vector Machine have higher AUC. Thus, the Random Forest and the Support Vector Machine are the best two out of the five machine learning models (Table 1). Since the Random Forest is obtained by integrating multiple Decision Trees, we believe that the Random Forest has better generalization under the same parameters. In summary, we chose the Random Forest as the most ideal machine learning model for the next step of the analysis. However, among the five machine models selected for this experiment, all of them had an AUC value of greater than 0.5, which indicated that the prediction rate of mildly fasted mice versus normally fed mice is higher than random prediction. In this experiment, all models were modeled using default parameters. Table 1Performance parameters of five machine learning modelsModelPrecisionRecallAccuracyF1-scoreAUC Decision Tree 1.000.920.940.950.96 Random Forest 1.000.920.940.950.99 K-Nearest Neighbors 1.000.920.940.950.96 Naive Bays 1.000.920.940.950.96 Support Vector Machine 1.000.920.940.950.99Note: Precision = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{TP}{TP\:+\:FP}\:$$\end{document};Recall =\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{TP}{TP\:+\:FN}$$\end{document}; Accuracy =\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{TP\:+\:TN}{TP\:+\:TN\:+\:FP\:+\:FN}$$\end{document}; F1-score = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{2TP}{2TP\:+\:FP\:+\:FN}$$\end{document}; Table 1Performance parameters of five machine learning modelsModelPrecisionRecallAccuracyF1-scoreAUC Decision Tree 1.000.920.940.950.96 Random Forest 1.000.920.940.950.99 K-Nearest Neighbors 1.000.920.940.950.96 Naive Bays 1.000.920.940.950.96 Support Vector Machine 1.000.920.940.950.99Note: Precision = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{TP}{TP\:+\:FP}\:$$\end{document};Recall =\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{TP}{TP\:+\:FN}$$\end{document}; Accuracy =\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{TP\:+\:TN}{TP\:+\:TN\:+\:FP\:+\:FN}$$\end{document}; F1-score = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{2TP}{2TP\:+\:FP\:+\:FN}$$\end{document}; Performance parameters of five machine learning models Note: Precision = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{TP}{TP\:+\:FP}\:$$\end{document};Recall =\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{TP}{TP\:+\:FN}$$\end{document}; Accuracy =\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{TP\:+\:TN}{TP\:+\:TN\:+\:FP\:+\:FN}$$\end{document}; F1-score = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{2TP}{2TP\:+\:FP\:+\:FN}$$\end{document}; Where, TP is the true positive which were predicted to be positive and actually were positive; FP is the False positive which were predicted to be positive but actually negative; True Negative (TN) was samples that were predicted to be negative and actually were negative; False Negative (FN) was samples that were predicted to be negative but actually positive; Precision, reflects the proportion of correct classification among the number of positive samples classified by the model; Recall, proportion of positive samples among correctly classified samples; Accuracy, proportion of correctly classified samples; F1-score, the average of Precision and Recall; AUC, area under ROC, the larger the AUC, the better the model. Fig. 3Five machine learning models were established to predict metabolic biomarkers between intermittent fasting and normal feeding in mice based on 595 significant differential metabolites. The machine learning modes are Decision Tree (A), Random Forest (B), K-Nearest Neighbors (C), Naive Bays (D), Support Vector Machine (E), respectively. (F) is the confusion matrix of Random Forest (Color depth represents the number of samples). The model uses three-fold cross validation and plots the receiver operating characteristic (ROC) curves, the blue line represents the mean of each ROC curve, representing the overall performance of the model Five machine learning models were established to predict metabolic biomarkers between intermittent fasting and normal feeding in mice based on 595 significant differential metabolites. The machine learning modes are Decision Tree (A), Random Forest (B), K-Nearest Neighbors (C), Naive Bays (D), Support Vector Machine (E), respectively. (F) is the confusion matrix of Random Forest (Color depth represents the number of samples). The model uses three-fold cross validation and plots the receiver operating characteristic (ROC) curves, the blue line represents the mean of each ROC curve, representing the overall performance of the model SHAP analysis to identify key metabolites as biomarkers in distinguishing patterns of the two dietary patterns SHAP analysis is a commonly used machine learning interpretable model that calculates the contribution value of each feature to that modeling. The lack of transparency due to the black-box nature of the machine learning model makes the results presented by the machine learning model unconvincing, and the use of SHAP analysis allows for an interpretable analysis of the machine learning model and assigns a certain level of contribution to each feature value [15]. In this experiment, depending on the size of the contribution value, the biomarkers can be identified for both dietary models. The Random Forest model was selected as the most desirable model and further analyzed the interpretability using SHAP analysis. Twenty metabolites with the highest mean contribution was selected to plot a bar chart, and the magnitude of the SHAP value is the contribution of the feature to the model (Fig. 4A), i.e., the extent to which the metabolite is influential in distinguishing between the two dietary approaches. The top 20 differential metabolites between 2 dietary patterns were described in supplementary information Tables S6. It can be seen that the metabolites Ganoderenic Acid C, (±)-3’,4’-Methylenedioxy-5,7-Dimethylepicatechin, Ganoderic Acid Beta contribute to the differentiation between the two dietary patterns in mice. In addition, combining the SHAP values as well as the feature values, potential explanations for the dependence of each trait on the model can be calculated (Fig. 4B). In this figure, red color indicates higher Feature value while blue color indicates lower Feature value, and the higher the Feature value, the better the metabolite’s classification prediction for distinguishing between the two dietary styles. It can be seen that Ganoderenic Acid C, (±)-3’,4’-Methylenedioxy-5,7-Dimethylepicatechin, Ganoderic Acid Beta has a negetive correlation in terms of contribution to SHAP and they were more likely to be present in the Control group. Tuberonic Acid, Genistein 7-O-Glucoside-4’-O-Apioside had a positive correlation for the contribution of SHAP to that and they were more likely to be present in EODF group mice. Fig. 4The contribution of each metabolite was calculated through SHAP analysis. (A) The metabolites in feces were ranked according to their SHAP value, and the top 20 were extracted as biomarkers that distinguished mice between intermittent fasting and normal feeding. (B) The fecal metabolite trends of mice under two dietary modes through SHAP value and feature value The contribution of each metabolite was calculated through SHAP analysis. (A) The metabolites in feces were ranked according to their SHAP value, and the top 20 were extracted as biomarkers that distinguished mice between intermittent fasting and normal feeding. (B) The fecal metabolite trends of mice under two dietary modes through SHAP value and feature value Collectively, 595 metabolites with significant difference were discovered between two dietary patterns, which play critical roles in butanoate metabolism, alanine, aspartate and glutamate metabolism, phenylalanine, tyrosine and tryptophan biosynthesis, riboflavin metabolism, D-Glutamine and D-glutamate metabolism, and nitrogen metabolism. After further screening and standardized scaling of 595 metabolites, 21 metabolites from the screening process were used to characterize the five machine learning models were established, and eventually Random Forest was optimal to conduct further interpretation using SHAP analysis for the prediction of metabolites biomarker. Consequently, the metabolites Ganoderenic Acid C showed more contribution to differentiate between normal diet and intermitten fasting dietary in mice. Overall, machine learning can effectively mine data from a large number of metabolites and identify potential biomarkers. Our work provides new insights for metabolic biomarker analysis and lay theoretical foundation for the selection of healthier dietary lifestyle. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1