1. What do we want the dog to detect?

Firstly, it is essential to define precisely what we want the dog to detect, as training it to recognise a family of compounds, such as opioids, is not the same as training it to identify a specific substance, such as fentanyl. This distinction involves deciding whether the objective is to detect a set of chemical signals or a specific compound.

However, even in the latter case, the odour profile is usually composed of multiple signals. It is therefore essential to clearly determine the detection objective in order to select and design a truly representative sample with which to train the dog.

2. Sample collection

Once the detection objective has been defined, the sample collection system must be established. At this stage, two key questions arise: how will the collection be carried out, and how many samples will be required?

There is no single answer, and it requires careful prior consideration. The collection method will depend largely on the detection objective and the way in which the odour propagates. When dealing with substances such as narcotics or explosives, it is standard practice to work with real samples or synthesised compounds, always within the legal framework and in compliance with all administrative requirements.

Conversely, in fields such as health, animal welfare or the study of cadavers, in addition to addressing the relevant ethical considerations, it is essential to analyse the pathways of odour emission, which involves deciding whether samples should be taken from faeces, sweat, saliva, urine or other biological matrices.

Once the type of sample has been defined, the sample size must be determined. In some cases, this is limited by legal or access issues, so minimums are established to ensure representative results. In others, the aim is to maximise the number of available samples. In any situation, it is essential that the samples are representative of the study population, including the variability associated with factors such as age, sex, race, specific conditions or temporal evolution.

It is also essential to include negative samples and controls, as these enable us to distinguish and validate which volatile organic compounds (VOCs) are truly characteristic of the positive samples.

3. Sample analysis

Once sampling has been completed, the samples are analysed. Given that the aim is to study the odour profile, the reference technique is HS-SPME-GC-MS, which allows the identification of volatile organic compounds present in the headspace, that is, the closest technological equivalent to an electronic nose.

The first step involves optimising the analytical method. To do this, various parameters specific to the technique must be evaluated and adjusted, such as the type of SPME fibre, temperatures, extraction and injection times, and chromatographic conditions. This phase is key to ensuring adequate sensitivity, reproducibility and resolution of the compounds.

The results are presented as chromatograms, which show the different signals associated with the compounds present in the sample. From these, it is possible to characterise the complete odour profile and identify the compounds individually, often with the aid of databases and specialist literature.

In cases where samples are particularly complex or contain compounds at very low concentrations, it may be necessary to use more advanced techniques that offer greater sensitivity, such as HS-SPME-GC/MS-TOF. This is particularly relevant given that dogs can detect compounds at extremely low concentrations, even down to parts-per-quadrillion levels.

Once the method has been optimised, all samples are analysed and the data obtained are recorded, including blanks and negative controls. The chromatographic signals provide an overview of the odour profile, whilst the area of each peak provides information on the relative abundance of the various compounds present.

4. Search for representative VOCs

Once all samples have been analysed and the list of compounds along with their relative abundances has been obtained, it is necessary to return to the initial objective to identify which volatile organic compounds (VOCs) are truly representative and universal and will serve as the basis for creating the aid.

In the case of specific substances, such as narcotics or explosives, where detection focuses on a particular compound or a small number of them, the aim is to identify the VOC (or VOCs) that best represents that substance. To this end, it is essential to have a thorough understanding of the compound’s synthesis and degradation processes, as these determine the possible by-products or metabolites present. This knowledge allows us to distinguish between compounds specific to the target substance and those originating from the matrix, such as excipients, plasticisers or other contaminants.

Furthermore, knowledge of physicochemical properties, particularly vapour pressure, is key to prioritising those VOCs most likely to be detected by the dog.

On the other hand, when the objective is the detection of a complex profile, such as in diseases, animal welfare or the study of cadavers, the focus shifts towards identifying common patterns within a family of VOCs. In these cases, it is necessary to perform a joint analysis of the positive samples to identify shared compounds, ruling out those present in blanks and negative samples.

This analysis is carried out using multivariate data analysis techniques, including machine learning tools, which enable the identification of complex patterns and relationships within the data.

5. Synthesis of the aid

Once the representative VOCs have been identified, the next step is to design and develop an aid that reproduces the odour profile observed in the actual samples. This process depends largely on the results obtained in the previous stages and may involve the procurement or synthesis of the selected compounds.

The main objective is to reproduce, on a substrate or device, an odour profile that is as faithful as possible to the original, maintaining relative abundance ratios like those detected experimentally. In the case of complex profiles, greater representation of VOCs considered universal is usually prioritised.

This stage marks the completion of the technological phase of the process, culminating in the production of a functional aid prototype at laboratory level.

6. Testing with dogs

However, it is necessary to move from the laboratory to the field, which involves conducting tests with dogs. However promising a prototype may be at the experimental stage, the final assessment rests with the dog.

It is therefore essential to subject the lure to practical trials in which its actual effectiveness is evaluated. The results obtained in terms of specificity and selectivity will be the key indicators of the device’s performance.

To ensure the reliability of the results, it is necessary to conduct trials with different dogs, from different units and handlers, to minimise bias and obtain a more robust evaluation. Furthermore, the objectivity of the process is based both on quantitative results and on the observation of the animals’ behaviour.

Should the results fail to meet the established thresholds, it will be necessary to return to the laboratory phase to make improvements to the prototype. This iterative process will be repeated until the defined validation criteria are met.

7. Performance testing

Finally, to assess the device’s actual performance, it is necessary to evaluate its stability and service life. Stability must be analysed under both storage and usage conditions, determining how long it retains an effective odour profile.

Furthermore, it is essential to establish its operational lifespan, that is, how many uses or training sessions it allows before its performance declines significantly.

These assessments are initially carried out under controlled laboratory conditions and are subsequently verified in real-world conditions through tests with dogs, thereby validating the device’s performance both experimentally and in the field.