Metritis is an inflammatory response of the uterus to traumatic, microbial and chemical insults. Bacteria are generally responsible for metritis, which most commonly manifests as endometritis [1,2]. Endometritis has a significant negative impact on reproductive performance. Its prevalence is high at 40–60 days postpartum with a rate of approximately 26 %, but reported rates vary between 5 % and 50 % [3].

The most common bacteria isolated from animals with endometritis are Escherichia coli, Trueperella pyogenes, Prevotella melaninogenica and Fusobacterium pyogenes [3,4]. Diagnosing endometritis on the basis of clinical signs is challenging as these are often absent. The definitive diagnosis of endometritis is made by histological, histopathological and microbiological examination. Transrectal ultrasound is also useful in the diagnosis [5,6].

One of the most important steps in treatment is the identification of the microorganisms causing the infectious disease. The identification process can extend to determining the serotype, genotype and species of the bacterium [7]. Determining the taxonomy of a bacterium is essentially based on determining its DNA and rRNA oligonucleotide sequences. However, these techniques are too difficult, time-consuming and expensive to be performed routinely in every laboratory. Therefore, identification is often based on the phenotypic profile of the bacteria. This involves subjecting a bacterial sample to various biochemical and physiological laboratory tests to determine its responses [[8], [9], [10]].

Most microorganisms produce significant primary metabolites (amino acids, proteins, carbohydrates, vitamins, acetone, ethanol, organic acids, etc.) and secondary metabolites (antibiotics, toxins, alkaloids, etc.) [11]. One of the most important metabolites for bacteria is glucose. Once glucose enters damaged bacterial cells, it is either stored as glucose in specific depots or it is further degraded, breaking down into end products depending on the type of degradation (aerobic or anaerobic). The mode of degradation and the resulting end products vary greatly between microorganisms. Attempts have been made to identify microorganisms on the basis of these products [11].

In recent years, there have been developments in the use of gases in human breath as non-invasive biomarkers for the clinical diagnosis of diseases such as bronchopneumonia, asthma, diabetes, various cancers and Helicobacter pylori infection [[12], [13], [14], [15]]. Microorganisms can produce certain species-specific volatile organic compound species during reproduction [16]. Gases released by bacteria cultured in a sealed bioreactor during their growth have been collected and analyzed by gas chromatography, showing that they emit some trace gases. The host immune response can also produce gases specific to certain pathogens [[17], [18], [19], [20], [21]]. Thus, it has been reported that it may be possible to use volatile organic compound (VOC) species in the diagnosis of some infectious agents using this method. However, there has been little investigation of the cellular sources or mechanisms responsible for the observed gases other than nitric oxide.

Recent advances have made it possible to detect and classify gases in the environment. Chemical sensors play an important role in this. These sensors consist of a chemical sensing surface and a unit that converts chemical interactions into electrical signals, enabling the development of various applications [22]. For example, a portable electronic 'nose' is capable of detecting various gases that cause bad breath in humans. This device can detect volatile sulphur compounds (hydrogen sulphide, dimethyl sulphide, methyl mercaptan and sulphur dioxide) produced by oral bacteria using gas chromatography techniques. Such devices can contain a range of chemical sensors and can detect volatile sulphur compounds produced by bacteria such as Treponema denticola, Porphyromonas gingivalis, Tanneralla forsythia, Fusobacterium nucleatum, Prevotella intermedia, Prevotella nigrescens and Actinobacilli [22,23].

The development of such gas sensors has been an active area of research in recent years, and many different materials have been developed. Some of these materials include graphene [24], carbon nanotubes [25], conducting polymers [26] and metal oxides [27]. Gas sensors based on nanocarbon materials (such as graphene and carbon nanotubes) have high sensitivity and selectivity, fast response and recovery, and low processing time. However, their industrial use has not yet been fully realised [24].

Conductive polymer gas sensors have good sensitivity but are affected by moisture, resulting in low repeatability [26]. Metal Oxide Semiconductor (MOS) gas sensors may not have the same sensitivity performance as others, but they are the most suitable for industrial use due to their low cost and mature technologies and are typically used in industrial applications [28]. Microelectromechanical systems (MEMS) sensors have become increasingly popular in recent years, mainly due to their small size, high sensitivity and low power consumption. One of their major advantages is their faster response time compared to MOS and electrochemical sensors [29]. However, MEMS sensors have a narrower range of sensor types compared to MOS and electrochemical sensors.

Metritis is one of the most important problems in the cattle industry and affects reproductive performance to varying degrees, especially the annual calving rate, which is one of the most important economic aspects of cattle farming. One of the problems is the inability to accurately, rapidly and reliably detect the factors that cause metritis in cows. Current methods, such as microbiology or polymerase chain reaction (PCR), are time-consuming and require specialised laboratories and expertise, resulting in labour, time and economic losses.

This study presents a device called Metrisor, which has been developed using gas sensors to quickly diagnose metritis with high accuracy and effectiveness. The device works by detecting the gases emitted by microorganisms as they grow. It is capable of analysing a wide range of gas measurements using current machine learning technologies. It can quickly determine whether an animal has metritis and identify the bacteria that cause it in the animal's environment. In addition, it can take 25 different gas readings simultaneously using a total of 19 different sensors, and is cost-effective and portable. The results are simultaneously evaluated by 10 different machine learning algorithms. The device is powered by electricity.