基于电子鼻检测的奶牛产后子宫炎早期预警模型初步研究

doi:10.11843/j.issn.0366-6964.2026.01.048

摘要/Abstract

摘要：

旨在评估电子鼻检测奶牛粪便建立的子宫炎预警模型准确性。本研究使用电子鼻检测50头健康和22头子宫炎奶牛发病前粪便以及血液样本挥发性有机化合物（volatile organic compounds，VOCs），正交偏最小二乘判别分析（orthogonal partial least squares discriminant analysis，OPLS-DA）模型初步分析数据。按7∶3划分训练集和测试集，建立预警模型并评估电子鼻预测子宫炎的性能。结果发现，健康和子宫炎奶牛发病前的粪便以及血液明显区分并各自聚类，在本研究中粪便是建立预警模型的最佳样本来源。选择决策树（Decision Tree，DT）、随机森林（Random Forest，RF）、K-邻近算法（K-Nearest Neighbors，KNN）、梯度提升（eXtreme Gradient Boosting，XGBoost）和线性判别分析（Linear Discriminant Analysis，LDA）5种不同的机器学习算法建立模型，结合模型评估指标和受试者特征（Receiver Operating Characteristic，ROC）曲线分析预测效率。研究发现测试集准确率（Accuracy，ACC）分别达到0.88、0.96、0.88、0.91、0.98，曲线下面积（Area Under Curve，AUC）分别是0.86、0.98、0.95、0.98、1.00。因此，RF和LDA模型预测性能表现最佳，电子鼻预测子宫炎的发生具有较高的准确性。综上，电子鼻检测奶牛发病前粪便VOCs能达到预警子宫炎的目的，在奶牛疾病的预警诊断中有很大的应用前景。

关键词: 奶牛子宫炎, 电子鼻, 机器学习算法, 早期预警

Abstract:

The aim of this study was to evaluate the accuracy of the early warning model for metritis in dairy cows established by electronic nose detection. The detection of volatile organic compounds （VOCs） in feces and blood samples of 50 healthy dairy cows and 22 metritis cows before onset by electronic nose， preliminary analysis of data by orthogonal partial least squares discriminant analysis （OPLS-DA） model. The training set and test set were divided according to 7∶3， the early warning model was established and the performance of the electronic nose in predicting metritis was evaluated. The results showed that， the feces and blood of healthy and metritis dairy cows were clearly distinguished and clustered respectively. In this study feces are the best sample source for establishing an early warning model. Five different machine learning algorithms， namely decision tree （DT）， random forest （RF）， K-nearest neighbors （KNN）， eXtreme gradient boosting （XGBoost）， and linear discriminant analysis （LDA）， were used to build models. Predictive efficiency is analyzed in conjunction with model evaluation metrics and receiver operating characteristic curves （ROC）. It was found that the test set accuracy （ACC） reached 0.88、0.96、0.88、0.91、0.98. Area under curve （AUC） was 0.86、0.98、0.95、0.98 and 1.00. As a result， the RF and LDA models perform the best in terms of predictive performance. The electronic nose has a high accuracy in predicting the occurrence of uterine infections. In conclusion， the detection of fecal VOCs of dairy cows before onset by electronic nose can achieve the purpose of early warning of metritis inflammation and has a great application prospect in the early diagnosis of dairy cow diseases.

Key words: metritis, electronic nose, machine learning algorithm, early warning

中图分类号:

S857.23

王芳, 刚华晟, 张榕, 段宏伟, 胡俊杰. 基于电子鼻检测的奶牛产后子宫炎早期预警模型初步研究[J]. 畜牧兽医学报, 2026, 57(1): 553-563.

WANG Fang, GANG Huasheng, ZHANG Rong, DUAN Hongwei, HU Junjie. A Preliminary Study on the Early Warning Model of Postpartum Metritis in Dairy Cows based on Electronic Nose Detection[J]. Acta Veterinaria et Zootechnica Sinica, 2026, 57(1): 553-563.

图/表 6

图1

图2

表1

表2

图3

图4

参考文献 58

[1]	PINEDO P，SANTOS J E P，CHEBEL R C，et al.Early-lactation diseases and fertility in 2 seasons of calving across US dairy herds［J］.J Dairy Sci，2020，103（11）：10560-10576.
[2]	DAETZ R，CUNHA F，BITTAR J H，et al.Clinical response after chitosan microparticle administration and preliminary assessment of efficacy in preventing metritis in lactating dairy cows［J］.J Dairy Sci，2016，99（11）：8946-8955.
[3]	SILVA T V，D E OLIVEIRA E B，PÉREZ-BÁEZ J，et al.Economic comparison between ceftiofur-treated and nontreated dairy cows with metritis［J］.J Dairy Sci，2021，104（8）：8918-8930.
[4]	DUBUC J，DUFFIELD T F，LESLIE K E，et al.Risk factors for postpartum uterine diseases in dairy cows［J］.J Dairy Sci，2010，93（12）：5764-5771.
[5]	SHELDON I M，CRONIN J G，BROMFIELD J J.Tolerance and innate immunity shape the development of postpartum uterine disease and the impact of endometritis in dairy cattle ［J］.Annu Rev Anim Biosci，2019，7：361-384.
[6]	CASARO S，PRIM J G，GONZALEZ T D，et al.Unraveling the immune and metabolic changes associated with metritis in dairy cows［J］.J Dairy Sci，2023，106（12）：9244-9259.
[7]	JUODŽENTIS V，ŠIUGŽDAITĖ J，PALUBINSKAS G，et al.Influence of puerperal metritis on the first ovulation after calving in dairy cows［J］.Pol J Vet Sci，2020，23（4）：619-626.
[8]	PÉREZ-BÁEZ J，SILVA T V，RISCO CA，et al.The economic cost of metritis in dairy herds［J］.J Dairy Sci，2021，104（3）：3158-3168.
[9]	PAIANO R B，BIRGEL D B，BONILLA J，et al.Alterations in biochemical profiles and reproduction performance in postpartum dairy cows with metritis［J］.Reprod Domest Anim，2020，55（11）：1599-1606.
[10]	GIULIODORI M J，MAGNASCO R P，BECU-VILLALOBOS D，et al.Metritis in dairy cows：risk factors and reproductive performance［J］.J Dairy Sci，2013，96（6）：3621-3631.
[11]	GARZON A，HABING G，LIMA F，et al.Defining clinical diagnosis and treatment of puerperal metritis in dairy cows：A scoping review［J］.J Dairy Sci，2022，105（4）：3440-3452.
[12]	FIGUEIREDO C C，MERENDA V R，DE OLIVEIRA E B，et al.Failure of clinical cure in dairy cows treated for metritis is associated with reduced productive and reproductive performance［J］.J Dairy Sci，2021，104（6）：7056-7070.
[13]	MENTA P R，FERNANDES L，PRIM J，et al.A randomized controlled trial evaluating the efficacy of systemic ceftiofur administration for metritis therapy in dairy cows and the effect of metritis cure on economically important outcomes［J］.J Dairy Sci，2024，107（9）：7092-7105.
[14]	MENTA P R，MACHADO V S，PIÑEIRO J M，et al.Heat stress during the transition period is associated with impaired production，reproduction，and survival in dairy cows［J］.J Dairy Sci，2022，105（5）：4474-4489.
[15]	GOSHEN T，SHPIGEL N Y.Evaluation of intrauterine antibiotic treatment of clinical metritis and retained fetal membranes in dairy cows［J］.Theriogenology，2006，66（9）：2210-2218.
[16]	BORCHARDT S，BURNETT T A，DRILLICH M，et al.Association of uterine health in the first lactation with transition cow health and reproductive performance in the second lactation of Holstein dairy cows［J］.J Dairy Sci，2024，107（11）：9706-9718.
[17]	JANOVICK N A，DRACKLEY J K.Prepartum dietary management of energy intake affects postpartum intake and lactation performance by primiparous and multiparous Holstein cows［J］.J Dairy Sci，2010，93（7）：3086-3102.
[18]	BARRACLOUGH R A C，SHAW D J，BOYCE R，et al.The behavior of dairy cattle in late gestation：Effects of parity and dystocia［J］.J Dairy Sci，2020，103（1）：714-722.
[19]	STANGAFERRO M L，WIJMA R，CAIXETA L S，et al.Use of rumination and activity monitoring for the identification of dairy cows with health disorders：Part III.Metritis［J］.J Dairy Sci，2016，99（9）：7422-7433.
[20]	BARRAGAN A A，LAKRITZ J，CARMAN M K，et al.Short communication：Assessment of biomarkers of inflammation in the vaginal discharge of postpartum dairy cows diagnosed with clinical metritis［J］.J Dairy Sci，2019，102（8）：7469-7475.
[21]	POMEROY B，SIPKA A，HUSSEN J，et al.Counts of bovine monocyte subsets prior to calving are predictive for postpartum occurrence of mastitis and metritis［J］.Vet Res，2017，48（1）：13.
[22]	MCNEEL A K，REITER B C，WEIGEL D，et al.Validation of genomic predictions for wellness traits in US Holstein cows［J］.J Dairy Sci，2017，100（11）：9115-9124.
[23]	DERVISHI E，ZHANG G，HAILEMARIAM D，et al.Urine metabolic fingerprinting can be used to predict the risk of metritis and highlight the pathobiology of the disease in dairy cows［J］.Metabolomics，2018，14（6）：83.
[24]	ALI A S，JACINTO J G P，MΫNCHEMYER W，et al.Estrus detection in a dairy herd using an electronic nose by direct sampling on the perineal region［J］.Vet Sci，2022，9（12）：688.
[25]	CHO Y S，JUNG S C，OH S.Diagnosis of bovine tuberculosis using a metal oxide-based electronic nose［J］.Lett Appl Microbiol，2015，60（6）：513-516.
[26]	MAURER D L，ELLIS C K，THACKER T C，et al.Screening of microbial volatile organic compounds for detection of disease in cattle：development of lab-scale method［J］.Sci Rep，2019，9（1）：12103.
[27]	LOUTFI A，CORADESCHI S，MANI G K，et al.Electronic noses for food quality：A review［J］.J Food Eng，2015，144：103-111.
[28]	TAECHACHOKEVIVAT N，KOU B，ZHANG T，et al.Evaluating the performance of herd-specific Long Short-Term Memory models to identify automated health alerts associated with a ketosis diagnosis in early lactation cows［J］.J Dairy Sci，2024，107（12）：11489-11501.
[29]	FERRAZ P A，POIT D A S，FERREIRA PINTO L M，et al.Accuracy of early pregnancy diagnosis and determining pregnancy loss using different biomarkers and machine learning applications in dairy cattle［J］.Theriogenology，2024，224：82-93.
[30]	SADEGHI H，BRAUN H S，PANTI B，et al.Validation of a deep learning-based image analysis system to diagnose subclinical endometritis in dairy cows［J］.PLoS One，2022，17（1）：e0263409.
[31]	RISVANLI A，TANYERI B，YILDIRIM G，et al.Metrisor：A novel diagnostic method for metritis detection in cattle based on machine learning and sensors［J］.Theriogenology，2024，223：115-121.
[32]	ZHANG E，WANG F，YIN C，et al.Application of an electronic nose for the diagnosis of ketosis in dairy cows［J］.Food Biosci，2024，60：104355.
[33]	DRAGONIERI S，QUARANTA V N，PORTACCI A，et al.Effect of food intake on exhaled volatile organic compounds profile analyzed by an electronic nose［J］.Molecules，2023，28（15）：5755.
[34]	LIMA F S，VIEIRA-NETO A，SNODGRASS J A，et al.Economic comparison of systemic antimicrobial therapies for metritis in dairy cows［J］.J Dairy Sci，2019，102（8）：7345-7358.
[35]	GRANTZ J M，MUKHOPADHYAY A，JANNASCH A H，et al.Plasma oxylipin profile of postpartum dairy cows categorized into different systemic inflammatory grades in the first week after parturition［J］.JDS Commun，2023，5（2）：155-160.
[36]	MENTA P R，NEVES R C，MACHADO V S.Association of time to metritis diagnosis with circulating concentration of metabolites，minerals，and haptoglobin in Jersey cows［J］.J Dairy Sci，2023，106（7）：5029-5042.
[37]	BRUINJÉ T C，MORRISON E I，RIBEIRO E S，et al.Progesterone profiles in postpartum dairy cows with inflammatory disorders［J］.J Dairy Sci，2024，107（9）：7153-7164.
[38]	BRUINS M，BOS A，PETIT P L，et al.Device-independent，real-time identification of bacterial pathogens with a metal oxide-based olfactory sensor［J］.Eur J Clin Microbiol Infect Dis，2009，28（7）：775-780.
[39]	XIAO Q，XIA Y.Insights into dendritic cell maturation during infection with application of advanced imaging techniques［J］.Front Cell Infect Microbiol，2023，13：1140765.
[40]	SAEZ A，HERRERO-FERNANDEZ B，GOMEZ-BRIS R，et al.Pathophysiology of inflammatory bowel disease：Innate immune system［J］.Int J Mol Sci，2023，24（2）：1526.
[41]	周佳丽，丁宝隆，马子明，等.奶牛子宫内膜炎与胃肠微生物相关性及益生菌作用的研究进展［J］.畜牧兽医学报，2024，55（8）：3321-3330.
	ZHOU J L，DING B L，MA Z M，et al.Research progress on the correlation between endometritis and gastrointestinal microorganisms and the role of probiotics in dairy cows［J］.Acta Veterinaria et Zootechnica Sinica，2024，55（8）：3321-3330.（in Chinese）
[42]	JEON S J，CUNHA F，VIEIRA-NETO A，et al.Blood as a route of transmission of uterine pathogens from the gut to the uterus in cows［J］.Microbiome，2017，5（1）：109.
[43]	JONES K，CUNHA F，JEON S J，et al.Tracing the source and route of uterine colonization by exploring the genetic relationship of Escherichia coli isolated from the reproductive and gastrointestinal tract of dairy cows［J］.Vet Microbiol，2022，266：109355.
[44]	GRIMARD B，HUMBLOT P，et al.Influence of postpartum energy restriction on energy status，plasma LH and oestradiol secretion and follicular development in suckled beef cows［J］.J Reprod Fertil，1995，104（1）：173-179.
[45]	CLARAMUNT M，MEIKLE A，SOCA P.Metabolic hormones，grazing behaviour，offspring physical distance and productive response of beef cow grazing at two herbage allowances［J］.Animal，2020，14（7）：1520-1528.
[46]	INGVARTSEN K L，ANDERSEN J B.Integration of metabolism and intake regulation：A review focusing on periparturient animals［J］.J Dairy Sci，2000，83（7）：1573-1597.
[47]	MANN S，URH C，SAUERWEIN H，et al.Short communication：The association of adiponectin and leptin concentrations with prepartum dietary energy supply，parity，body condition，and postpartum hyperketonemia in transition dairy cows［J］.J Dairy Sci，2018，101（1）：806-811.
[48]	BRIANO C，MEIKLE A，VELAZCO J I，et al.Metabolic and hormonal profiles and productive performance in primiparous and multiparous cows grazing different forage allowance in late gestation［J］.Theriogenology，2024，227：68-76.
[49]	GONZÁLEZ L A，TOLKAMP B J，COFFEY M P，et al.Changes in feeding behavior as possible indicators for the automatic monitoring of health disorders in dairy cows［J］.J Dairy Sci，2008，91（3）：1017-1028.
[50]	NEAVE H W，LOMB J，WEARY D M，et al.Behavioral changes before metritis diagnosis in dairy cows［J］.J Dairy Sci，2018，101（5）：4388-4399.
[51]	PAUDYAL S，MAUNSELL F P，RICHESON J T，et al.Rumination time and monitoring of health disorders during early lactation［J］.Animal，2018，12（7）：1484-1492.
[52]	BURFEIND O，BRUINS M，BOS A，et al.Diagnosis of acute puerperal metritis by electronic nose device analysis of vaginal discharge in dairy cows［J］.Theriogenology，2014，82（1）：64-70.
[53]	VIDAL G，SHARPNACK J，PINEDO P，et al.Impact of sensor data pre-processing strategies and selection of machine learning algorithm on the prediction of metritis events in dairy cattle［J］.Prev Vet Med，2023，215：105903.
[54]	LEGGIERI M C，MAZZONI M，FODIL S，et al.An electronic nose supported by an artificial neural network for the rapid detection of aflatoxin B₁ and fumonisins in maize［J］.Food Control，2021，123：107722.
[55]	KAMPHUIS C，MOLLENHORST H，HEESTERBEEK J A，et al.Detection of clinical mastitis with sensor data from automatic milking systems is improved by using decision-tree induction［J］.J Dairy Sci，2010，93（8）：3616-3627.
[56]	MOAWED S A，MAHROUS E，ELASWAD A，et al.Milk yield prediction in Friesian cows using linear and flexible discriminant analysis under assumptions violations［J］.BMC Vet Res，2024，20（1）：392.
[57]	WANG J，BELL M，LIU X，et al.Machine-learning techniques can enhance dairy cow estrus detection using location and acceleration data［J］.Animals （Basel），2020，10（7）：1160.
[58]	FAUVEL K，MASSON V，FROMONT E，et al.Towards sustainable dairy management-a machine learning enhanced method for estrus detection［C］.Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.2019：3051-3059.

排名 Rank	特征名称 Sensor	评分 Score	排名 Rank	特征名称 Sensor	评分 Score
1	S6	0.030 995	17	S25	0.000 110
2	S3	0.001 080	18	S1	0.000 097
3	S28	0.000 940	19	S14	0.000 082
4	S27	0.000 692	20	S29	0.000 075
5	S9	0.000 424	21	S8	0.000 073
6	S2	0.000 417	22	S16	0.000 071
7	S19	0.000 367	23	S11	0.000 062
8	S18	0.000 356	24	S17	0.000 059
9	S26	0.000 341	25	S7	0.000 055
10	S15	0.000 276	26	S5	0.000 039
11	S23	0.000 227	27	S21	0.000 033
12	S32	0.000 214	28	S4	0.000 024
13	S24	0.000 203	29	S31	0.000 014
14	S10	0.000 165	30	S20	0.000 009
15	S30	0.000 148	31	S12	0.000 002
16	S13	0.000 115	32	S22	0.000 000

数据集Dataset	评价指标 Evaluation metrics	DT	RF	KNN	XGBoost	LDA
Train	ACC	0.84	0.94	0.77	0.91	0.98
	TPR	0.86	0.94	0.89	0.89	0.99
	TNR	0.86	0.94	0.89	0.89	0.99
	PPV	0.85	0.98	0.85	0.90	0.97
Test	ACC	0.88	0.96	0.88	0.91	0.98
	TPR	0.86	0.94	0.89	0.89	0.99
	TNR	0.86	0.94	0.89	0.89	0.99
	PPV	0.85	0.98	0.85	0.90	0.97
	Precision	0.92	0.95	0.97	0.92	1.00
	Recall	0.89	1.00	0.84	0.95	0.97
	F1 Score	0.91	0.97	0.90	0.94	0.99