Animal characteristic status

Abstract

A point-of-need, PON, instrument (200) for obtaining a metric for a physiological animal characteristic, the instrument (200) comprising: an insertable rotor cartridge (202) for sample assaying, comprising a plurality of reagent compartments (202a) for receiving an animal fluid sample, each reagent compartment (202a) comprising a reagent configured to react with an animal fluid sample; a rotor, configured to drive rotation of the rotor cartridge (202) to agitate contents of the plurality of reagent compartments (202a); an analyser (220), comprising an optical analysis setup and a processor (206), the analyser (220) being configured to perform optical analysis on the contents of each reagent compartment (202) after the reagent and animal fluid sample have reacted, to produce biomarker data; a transmitter (208) configured to communicate with a remote biomarker identification system (400), wherein the analyser (220) is configured to produce the biomarker data having a format allowing the biomarker data to be transmitted by the transmitter (208) to the biomarker identification system (400) for analysis to generate a physiological animal characteristic status.

Description

[0001]ANIMAL CHARACTERISTIC STATUS The present disclosure relates to a point-of-need instrument for obtaining a metric for a physiological animal characteristic, a physiological animal characteristic identification system and a computer-implemented method for determining a physiological animal characteristic status. Animals may undergo physiological changes in their lifetime, which may be cyclical or a permanent physiological change. Studying an animal’s physiological characteristics is of interest for scientists studying the animal. Where animals are domesticated, physiological changes may be desirable to monitor. Knowledge of physiological changes or current physiological properties of an animal may guide how their keeper (e.g. farmer) moves them between habitats, adapts feeding, changes their function (for example from a breedable or egg-laying animal to an animal being prepared for slaughter) or otherwise monitors their development, their suitability for activities such as mating, and the like. In a specific example of a physiological change in an animal as discussed herein, smoltification is a complex series of behavioural, developmental, and physiological changes where fish adapt from living in freshwater to living in salt or sea-water. Smoltification is also known as Parr-Smolt transformation and happens to salmonid fish – i.e. fish that belong to the Salmonidae family, for example Atlantic salmon. Smoltification as a physiological process enables anadrome fish species to migrate from freshwater to saltwater and survive in a changing osmotic environment and maintain the fluid balance. In the wild, fish undergo smoltification on their first migration down to the sea. In salmon fish farms, the farmed fish are transferred from the hatchery to marine growth sites in the sea. Fish farmers need to identify the best time to transfer farmed fish from freshwater to saltwater. If the fish are transferred to soon, they will not have adapted sufficiently to survive and thrive in a saltwater environment. If the fish are transferred too late, the fish will no longer be adapted for existing in freshwater. Fish can become ready to move to saltwater, but if not moved, can revert to their freshwater physiology. Therefore, there is an optimal time window in which fish need to be moved. Getting the timing wrong may stunt development of the fish and may lead to their death due to being unsuited for their environment. This is uneconomical for the farmer if fish do not grow to the required size, as well as wasteful and potentially unsanitary (where living and dead fish may be in close proximity). As animal welfare in farming is becoming of greater public interest, it is important to reduce needless loss of life before the fish are ready to be used for food and other productive uses. Transfer of animals is not just limited to aquatic farming. For example, ornamental fish and other pets, molluscs and other aquatic species, other pets, livestock and the like may also be transferred between different conditions. Typically, experts in the field are relied upon to analyse whether fish are viable to be transferred from freshwater to saltwater by performing tests on fish gill samples. Analysis is typically lethal for chosen test fish. Other physiological characteristics of fish are desirable to monitor – fish that are not anadromous (i.e. do not migrate from freshwater rivers to the ocean and back to spawn) may exhibit changes to their physiology due to stress, for example, and it may be desirable to assess their health. Generally, farmers, as well as veterinarians and feed producers, would benefit from efficient and reliable ways of benchmarking animal health, for example aquatic animal health, and obtaining biological insight – indicative of smoltification status or gonadal maturation, for example. SUMMARY STATEMENTS The disclosed technology seeks to mitigate, obviate, alleviate, or eliminate various issues known in the art. Whilst the invention is defined by the accompanying claims, various aspects of the disclosed technology including the claimed technology are set out in this summary section with examples of some preferred embodiments and indications of possible technical benefits. A first aspect of the disclosed technology relates to a point-of-need, PON, instrument for obtaining a metric for a physiological animal characteristic, the instrument comprising: an insertable rotor cartridge for sample assaying, comprising a plurality of reagent compartments for receiving an animal fluid sample, each reagent compartment comprising a reagent configured to react with an animal fluid sample; a rotor, configured to drive rotation of the rotor cartridge to agitate contents of the plurality of reagent compartments; an analyser, comprising an optical analysis setup and a processor, the analyser being configured to perform optical analysis on the contents of each reagent compartment after the reagent and animal fluid sample have reacted, to produce biomarker data; and a transmitter configured to communicate with a remote biomarker identification system, wherein the analyser is configured to produce the biomarker data having a format allowing the biomarker data to be transmitted by the transmitter to the biomarker identification system for analysis to generate a physiological animal characteristic status. The PON instrument may be for obtaining a metric for a physiological animal characteristic of a domesticated animal. The animal fluid sample may be a fluid sample from a domesticated animal. The domesticated animal may be one of: a farmed terrestrial animal including a farmed arboreal animal, a farmed aquatic animal or farmed semiaquatic animal, a pet or an animal under scientific study. For example, the domesticated animal may be a fish. The biomarker data may have a format allowing the biomarker data to be processed at a laboratory information system (LIS) of the biomarker identification system. The biomarker data may have a format meeting the health level 7, HL7, primary standards. The biomarker data may have a format meeting the HL7 Fast Healthcare Interoperability Resources, FHIR, standard. The biomarker data may have a format meeting the FHIR standard and is encoded as JavaScript Object Notation, JSON, data. Another aspect of the present disclosure provides a physiological animal characteristic identification system, comprising: the PON instrument as described above; a biomarker identification system remote from the PON instrument, comprising: a receiver, configured to receive biomarker data from the PON instrument; a memory, comprising a machine-learning model, wherein the model is a classification or regression model configured to generate a physiological animal characteristic status based on the present biomarkers in the biomarker data; and a transmitter, configured to communicate the physiological animal characteristic status to a user computing device; and a user computing device, comprising: a receiver configured to receive a signal from the biomarker identification system; and a display configured to provide a result generated by the biomarker identification system responsive to analysing the received biomarker data to identify the presence of biomarkers in the sample, the result indicating a physiological animal characteristic status. The model may be configured to identify the presence of a biomarker in the sample based on the biomarker data and attribute an animal status indicator to the biomarker data, wherein the model is configured to aggregate two or more animal status indicators to generate an overall physiological animal characteristic status. The model may comprise an XGBoost machine learning algorithm comprising an ensemble of decision trees, wherein each decision tree is configured to receive biomarker data that has been parsed from a first data format into a second data format as an input and wherein the ensemble of decision trees are configured to generate a score, wherein the physiological animal characteristic status is based on that score. The biomarker identification system further comprises a laboratory information system, LIS, wherein the LIS is configured to parse the received biomarker data from a first data format into a second data format suitable for inputting into the model and provide the parsed biomarker data as an input to the model, wherein the LIS is further configured to generate a report of the physiological animal characteristic status for communication to the user computing device. The physiological animal characteristic identification system may further comprise at least one more PON instrument according to any preceding PON instrument claim, wherein the LIS is configured to parse received biomarker data from each PON instrument into a format suitable for inputting into the model and communicate parsed biomarker data from each of the PON instruments to the model. The physiological animal characteristic status may comprise a viability status (e.g. a calculated probability) for fish to survive in salt water. The animal fluid sample may comprise any of: blood, whole blood, plasma, serum, mucus, faeces and ascites fluid. The model may be configured to determine one or more animal status indicators based on the biomarkers in the sample, and wherein the model is configured to generate the physiological animal characteristic status based on the one or more animal status indicators, wherein the one or more animal status indicators includes at least one of: a health status, a stress status, a gonade status, a metabolic status, and a smoltification status. One of the animal status indicators may be an indicator of chronic stress, the animal is a fish and the biomarkers in the sample may comprise phosphorus, sodium, calcium, growth hormone and chloride. It is noted that these biomarkers may apply for non-fish but are particularly relevant for fish. As such, one of the animal status indicators may be an indicator of chronic stress and the biomarkers in the sample may comprise phosphorus, sodium, calcium, growth hormone and chloride. Another aspect of the present disclosure provides a computer-implemented method for determining a physiological animal characteristic status, the method comprising at a biomarker identification system: receiving biomarker data obtained from an animal fluid sample from a PON instrument, wherein the biomarker is in a first data format, and parsing the biomarker data at a server remote from the PON instrument into a second format suitable for inputting into a classification or regression model; inputting the parsed biomarker data into the model at the server remote from the PON instrument, the model being configured to process the parsed biomarker data to obtain a plurality of animal status indicators; aggregating the plurality of animal status indicators into a single physiological animal characteristic status; and outputting, by the model, the physiological animal characteristic status. Another aspect of the present disclosure provides a non-transitory computer readable medium comprising computer code which, when loaded from memory and executed by one or more processor(s) or processing circuitry, causes a biomarker identification system to perform the above method. The disclosed aspects and embodiments may be combined with each other in any suitable manner which would be apparent to someone of ordinary skill in the art. BRIEF DESCRIPTION OF THE DRAWINGS Some embodiments of the disclosed technology are described below with reference to the accompanying drawings which are by way of example only and in which: Figure 1A shows a fish pen environment; Figure 1B shows a different fish pen environment; Figure 2 shows a block diagram of an example PON (point of need) instrument according to embodiments; Figure 3 shows an example rotary cartridge and PON instrument according to embodiments; Figure 4 shows an example biomarker identification system comprising a server according to embodiments; Figure 5 shows an example decision tree of a model according to embodiments; Figure 6 shows an example workflow of a biomarker identification system starting from three example PON instruments according to embodiments; Figure 7A shows an example timeframe for performing analysis according to embodiments; Figure 7B shows an example timeframe for performing analysis according to an example in which the result is delivered back to the PON instrument; Figure 8 shows a method of training a model according to embodiments; Figure 9A shows an example data flow through a physiological fish characteristic identification system according to embodiments; Figure 9B shows an example data flow through a physiological fish characteristic identification system according to examples in which the result is delivered back to the PON instrument; Figure 10 shows an example Break Down Profile from the model; Figure 11 shows an example user interface display comprising an example Break Down Profile; and Figure 12 shows a correlation between an example ML model and the PCR gold standard for analysing smoltification status. DETAILED DESCRIPTION The present figures show an example animal – fish – but as noted above, the present disclosure relates to animals in general, and more specifically domesticated animals. The fish are representative of any suitable animals. Discussion in the following example of “fish” may be generalised to “animals” or “domesticated animals” and may instead be applied to other example animals, for example farmed terrestrial, aquatic or semiaquatic animals, pets, or animals that are being studied for their physiological characteristics. Example farmed terrestrial animals are cows, sheep, alpaca and the like. “Farmed” does not exclusively refer to farming for animal by-products / meat and may include animals located on farms or otherwise captive for entertainment or other purposes. The meaning of “farmed” here refers to animals that are not wild, are kept for profit / as part of a business (i.e. not a pet) and whose characteristics are to be determined as described herein. Pets are another example of domesticated animals. For the purposes of the present application, “domesticated” also includes an animal under scientific observation / study, for example a bird whose migration is being monitored. Animals, including wild animals, may have their physiological properties determined as discussed herein; however, it is foreseen that the importance of the present disclosure lies with monitoring animals that are owned or being studied (and so are likely to be tagged or otherwise identifiable amongst wild animals). Figure 1A shows an example animal, fish 102, in a fish pen environment 100. The fish 102 may be anadromous, meaning that they migrate from freshwater to sea water to spawn their young. In a non-wild environment, the fish are not free to migrate under instinct. The keepers of the fish, for example fish farmers, need to replicate a migration for the fish 102 by transferring the fish 102 into saltwater for the fish 102 to develop properly. Figure 1A shows fish 102 that may be in a stage of their life before they are ready to move to saltwater (which may be sea water or a manufactured equivalent of sea water) and are residing in freshwater. Figure 1B shows a second fish pen environment 104. The fish 106 depicted in Figure 1B are larger in size than the fish 102 in the first fish pen environment 100. Figure 1B shows fish 106 that may be at a later stage of development versus the fish 102 in the first fish pen environment 100. The second fish pen environment 104 may be larger than the first fish pen environment 100 to accommodate the larger fish. The second fish pen environment 104 may contain sea water or other saltwater, such that the environment is suitable for fish 106 that can survive and thrive in sea water or saltwater. The second fish pen environment 104 may be located in the sea, for example. Fish 102 may be transferred from the first fish pen environment 100 to the second fish pen environment 104 at the right time in their development, where they are anadromous or otherwise evolve to live in saltwater. This may involve releasing fish 102 into a sea-based pen, for example. In this way, the life cycle of anadromous fish 102, 106 in the wild can be replicated in captivity. Being able to identify the best time to transfer fish 102 is useful to keepers of the fish 102, to upkeep fish health in a suitable environment for their stage of life. Whether or not to transfer a fish 102 to saltwater may be identified by an indication of saltwater viability, which may be a measure of the fish’s 102 likelihood or suitability for survival in saltwater. Fish 102 viable for life in saltwater may exhibit saline osmoregulation, i.e. the ability to extract oxygen without dying from electrolyte imbalance. An option for determining viability is for a veterinarian or other expert to assess fish 102, in person. A sample may be obtained from a fish 102 at the fish’s location and assessed. The sample is transported to a laboratory. Common methods for assessing viability include performing a polymerase chain reaction (PCR) test on a gill sample. The present invention removes the need for a veterinarian or other expert to be present at the fish’s location (or other example animal) and removes the need for any assessment in a real-world laboratory environment. The present invention can be used directly by the fish farmer or other non- expert in fish physiology or other non-expert. The present invention can be used without causing death of the fish (or other example animal). A point-of-need (PON) instrument 200 may be used to collect a fluid sample; in this example, a fluid sample from a fish 102 in the first fish pen environment 100. Figures 2 and 3 show an example PON instrument 200. Figure 2 shows a block diagram of features of the PON instrument 200 and Figure 3 shows an example view of the PON instrument 200 alongside a rotor cartridge 202. Point of need (PON) instruments 200 are often used in fields such as food safety, environmental monitoring, and veterinary medicine, where the ability to quickly detect and identify pathogens or contaminants is crucial. Point of need testing is testing in the field, for example at or near a fish farm, for example at the fish pen environment 100. A point of need sample may be obtained using a PON instrument 200, configured to provide rapid and accurate test results at the location where the test is needed, rather than requiring samples to be sent to a laboratory for analysis. The sample may be a blood sample, a scale sample, a skin scrape, a gill biopsy or clip, a fin biopsy or clip or the like. Taking a scale sample may include performing a mucus smear. The sample may be a mucus sample or include mucus. The sample may be a faecal sample or include faeces. The sample may be a sample of ascites fluid or another bodily fluid that may appear or build up due to a health condition. Throughout this application and the drawings, “blood sample” may be interchanged with another type of sample from a fish 102 or other animal, except where the description relates particularly to blood. A blood sample may be a whole blood sample or a plasma sample, for example, or may be a serum sample. The sample may be taken using a needle or other sampling mechanism in a lethal or non-lethal manner, by swab sampling of saliva or mucus, by extracting urine or fecal matter, by homogenization of organs or larval stages. Biomarkers are identified in the animal fluid sample. In the present example, biomarkers are identified in the fish fluid sample. A biomarker is a biological molecule that can be identified in a sample as an indicator of a process having occurred or of a condition, for example a disease. The PON instrument 200 may be configured to allow certain blood characteristics from a blood sample to be isolated and measured in vitro. The PON instrument 200 comprises an analyser 220 configured to analyse an animal, e.g. fish, fluid sample and the analyser 220 may comprise a blood assay device, for example. A blood sample assay device is an analytic device used to analyse blood samples for the presence of specific substances, such as proteins, enzymes, electrolytes or antibodies. Blood sample assay devices typically use a small amount of blood and perform the analysis using various techniques such as photometric assays, haematology, immunoassays, enzymatic assays, or molecular diagnostics. In the present PON instrument 200, the analyser 220 is configured to perform optical assessments of the animal fluid sample. The PON instrument 200 may be configured to receive a rotor cartridge 202. The rotor cartridge 202 may comprise multiple reagent compartments 202a. Each compartment 202a may comprise a chemical reagent and may include a dilutant. Each compartment 202a may be configured to receive animal fluid and may be pre-calibrated (before receiving any animal fluid) to a specific chemistry test. The reagent material may be lyophilised reagent beads, for example, which are suitably stabilised for point-of-need testing. The rotor cartridge 202 may be single-use – used for one round of analysis. The rotor cartridge 202 may be pre-calibrated for each test to ensure accuracy and consistency. The rotor cartridge 202 may comprise a reserve – for example in the centre – to receive the animal fluid sample, which is distributed to the reagent compartments 202a when the rotor cartridge 202 is spun. The animal fluid sample then reacts with the reagents in each compartment 202a. In the example where the animal is fish, only 90-120 µl of fish fluid sample is required to be added to the rotor cartridge 202. More can be added, but there is no need to take a lethal amount of sample from the animal. The reaction causes a change in the light absorption properties of the fluid in the reagent compartments 202a, which may be due to a shift in colour of the solution, or due to a change in opacity or translucency of the fluid or changes to optical characteristics of the fluid. From the change in light absorption properties / optical characteristics, the concentration of a particular biomarker in the sample can be identified, as described below. Example biomarkers are: a potassium cation (K+) biomarker; an amylase (AMY) biomarker; an albumin (ALB) biomarker; a cholesterol (CHOL) biomarker; an alkaline phosphatase (ALT) biomarker; a sodium cation (NA+) biomarker; a chloride anion (Cl-) biomarker; a magnesium (MG) biomarker; a carbon dioxide (CO2) biomarker; a calcium (Ca) biomarker; and a phosphorus (P) biomarker. The biomarkers that are tested for may be determined by the type of physiological animal characteristics being investigated. For example, blood analytes phosphorus, sodium, calcium, growth hormone and chloride are suitable candidates for chronic stress assessment. The PON instrument 200 may comprise a housing 200a, configured to house an analyser 220 of the PON instrument 200 and to receive the rotor cartridge 202 for analysis. Once the rotor cartridge 202 is loaded with the sample, it may be inserted into an opening 222 in the PON instrument housing 200a. The opening 222 may be a slot in the housing 200a, which may or may not include a cover, or a drawer, or the housing 200a may be openable to receive the rotor cartridge 202 or may even separate into parts to re-build around the rotor cartridge 202 once inserted. The housing 200a may house any of the PON instrument 200 elements described herein. Inserting the rotor cartridge 202 into the opening 222 may automatically align the compartments 202a with elements of the analyser 220 that will analyse the sample and may automatically align the rotor cartridge 202 with a rotor to spin the rotor cartridge 202. The analyser 220 may comprise the rotor configured to drive spinning of the rotor cartridge and may comprise a motor to drive the rotor, and a power supply (for example a battery) configured to supply power to the motor. The PON instrument 200 may comprise a user interface 204, which may comprise physical buttons, a touchscreen, a camera or a microphone, for example, to receive a user input and cause the PON instrument 200 to perform a function. The PON instrument 200 may comprise a display 212, configured to display information to a user, which may be a touchscreen and so also serve as a user interface 204. The user interface 204 may enable the user to power up / down the PON instrument 200 and / or instruct the PON instrument to perform its function. The PON instrument 200 may enable a user to cause the rotor cartridge 202 to be spun and / or the analyser 220 may comprise a sensor configured to sense that a rotor cartridge 202 has been inserted into the analyser 220 and cause the rotor cartridge 202 to be spun automatically after insertion. The analyser 220 may be configured to spin the rotor at a sufficiently high speed to distribute the sample evenly across the reagent compartments 202a, ensuring a uniform reaction and accurate measurements. The analyser 220 of the PON instrument 200 may comprise an optical analysis setup. The optical analysis setup may comprise a light source, a wavelength selector, and a light detector. The optical analysis setup may be configured to perform absorption spectroscopy, transmission turbidimetry or absorption spectrophotometry. As the rotor cartridge 202 spins, the light source may be configured to illuminate each reagent compartment 202a at one or more specific wavelengths, and the light detector may be configured to measure the amount of light that is not absorbed by the reacted sample – i.e. any light that is transmitted, reflected or otherwise is not absorbed. Different biomarkers that may be in the sample will absorb the light differently (i.e. will absorb a specific wavelength), giving each biomarker a unique absorption “fingerprint”. Transmission spectra will therefore show whether a biomarker is present or not. The light detector may be a photodiode detector or a photodiode detector array, for example. The light detector may be a photoelectric cell. Wavelength selector examples include prisms, diffraction gratings and filters. For transmission turbidimetry, the wavelength selector may be a filter between the light source and the rotor cartridge to limit the wavelength of light reaching the sample. The wavelength selector may be configured to control the specific wavelength for each test. As different biomarker molecules absorb light at different wavelengths, using the correct wavelength is crucial for accurate measurements. The PON instrument 200 may be configured to detect light (at the light detector) at more than one different wavelength, to allow for the PON instrument 200 to identify different biomarkers in the sample. The PON instrument 200 may be configured to detect light at 340, 405, 450, 505, 546, 600, 630 and / or 850nm wavelengths for example. The light source may be in the visible and may be in the infrared, to provide these desired wavelengths to be detected. More than one light source may be included in the PON instrument 200 to provide the desired breadth of wavelengths. The PON instrument 200 may comprise a processor or processing circuity 206 and a memory 210. Once the light detector has detected light that was not absorbed in the reagent compartments 202a, the PON instrument 200 may be configured to compare absorption of wavelengths with earlier data (which may be test data) at the processor 206. The earlier data may be retrieved from remote storage (e.g. cloud storage) or may be retrieved from the memory 210 of the PON instrument 200. The memory 210 may be configured to store test data, which may comprise data from sample(s) obtained earlier by the PON instrument 200 and / or may comprise data from sample(s) obtained earlier by a different sampling instrument. The earlier sample data may comprise test data, taken from a control group of fish 102 or another animal for example. This earlier sample data may be used to calibrate the PON instrument 200, for example, and / or may be used as reference data against which the sample data obtained by the PON instrument 200 may be compared. The PON instrument 200 may be configured to retrieve test data or other earlier sample data from cloud storage or other remote storage. The PON instrument 200 may comprise a transmitter and receiver or transceiver 208 such that the PON instrument 200 can transmit and receive data – for example to / from cloud storage. The PON instrument 200 may be configured to generate biomarker data at the processor or processing circuitry 206 based on the detected light. The biomarker data generated at the PON instrument 200 may comprise an amount of a particular biomarker present in the sample. The biomarker data generated at the PON instrument 200 may comprise indicators of quantities of more than one type of biomarker present in the sample. The processor or processing circuity 206 may generate biomarker data for sending for remote analysis. In absorption spectroscopy, the transmission spectra at the detector show where absorption has taken place at the sample. The biomarker data may be generated from the transmission spectra at the PON instrument 200 processor 206. Different absorption of wavelengths indicates the presence / absence / abundance of different biomarkers. The amount of each biomarker in the sample may therefore be calculated from the absorption of wavelengths. The processor 206 or processing circuitry may be configured to identify substances present in the sample. For example, the processor 206 or processing circuitry may be configured to identify proteins, enzymes, electrolytes or antibodies in the input sample. The memory 210 may comprise code 214, comprising operating system application(s) 216 for the PON instrument 200. In the example where the animal is fish, the memory 210 may further comprise a fish analysis client application 218. The operating system 216 may be configured to run basic functions of the PON instrument 200. A role of the operating system 216 is to execute the fish analysis client application 218. The fish analysis client application 218 may cause the analyser 220 to perform analysis of the sample in the rotor cartridge 202 once the sample has been received. The fish analysis client application 218 may cause the processor 206 to process the results of analysis by the analyser 220. The fish analysis client application 218 may be configured to send a request to a remote biomarker identification system 400 to perform a fish analysis service based on the data generated by the analyser 220. Another suitable animal analysis service by an animal analysis client application may be performed having the features described above. Multiple PON instruments 200 may be used to obtain multiple samples in the field – for example a team of users may be in the field and wanting to analyse fish 102 or another animal in bulk. Multiple rotor cartridges may be used and inserted into the same PON instrument 200 in succession, to work through multiple samples quickly and efficiently. The processor 206 may sort the analysis results from the analyser 220 to distinguish between samples / between rotor cartridges 202 (and so samples). The transmitter or transceiver 208 may be configured to transmit a digital signal and the receiver or transceiver 208 may be configured to receive a digital signal. The processor or processing circuitry 206 may be configured to process a received digital signal. The transmitter or transceiver 208 may be configured to transmit a signal comprising information about the input blood / mucus sample, gleaned from the processor or processing circuitry 206. The transmitter or transceiver 208 may be configured to transmit a request to the biomarker identification system 400, which may comprise a remote platform for analysing biomarker data. The transmitter or transceiver 208 may be configured to transmit a request for a physiological animal characteristic status / report, for example for a result / report comprising a fish smoltification status to the biomarker identification system 400. The PON instrument 200 may include an input / output or I / O 222 configured to allow data input and data output. Figure 2 shows an example input, the user interface 204 and an example output, the display 212. The I / O 222 may comprise any suitable mechanism to input and output data to and from the PON instrument 200, for example a keyboard (which may be touchscreen or physical buttons) or microphone, or a printer. The display 212 may be configured to display information to the user, visually. The displayed information may be instructions for use of the PON instrument 200 for example, or other useful indications such as battery life, date, time and the like. In one example, the PON instrument 200 is configured to receive results from a biomarker identification system having sent readings via the transmitter 208 and the display 612 may be configured to display the physiological animal characteristic status or report to the user. The PON instrument display 212 may be configured to show other results generated at the instrument. An example PON instrument 200 that may be used to obtain a sample and identify biomarkers in the sample is the Seamaty SMT-120VP. The Seamaty SMT-120VP can measure multiple analytes simultaneously by analyzing the detected light at different wavelengths. This device typically prints out results of optical analysis of the sample. The biomarker information in this form may be accessible to a veterinarian or other relevant professional to understand, but not (typically) to a farmer or pet-owner, for example. In the present disclosure, the biomarker data is transmitted for remote analysis, such that farmer- friendly indications of animal physiological characteristics can be reported to a user computing device 703a with additional information and simple presentation for the farmer, or other animal keeper or other non-expert. A remote server 438 may be configured to perform biomarker identification – the figures depict an example biomarker identification system 400. The server 438 may be a cloud server (i.e. a virtual, not physical, server running in a cloud computing environment). The server 438 is labelled in Figure 4 as an “ML model server” because the server 438 is configured to run a classification or regression model, which is used to characterise the animal, for example the fish 102, from which the sample was taken, using biomarkers in the sample. The model is described in further detail below, in relation to Figure 5. The PON instrument 200 may be configured to communicate the biomarker data to a biomarker identification system 400 that is remote from the instrument 200. Figure 4 shows an example server 438 of the biomarker identification system 400 and its example components. By remote, it is meant that the biomarker identification system 400 is not part of the same physical item as the PON instrument 200 (e.g. not in the housing 200a). The biomarker identification system 400 is not a laboratory in the traditional sense – the PON instrument 200 does not output a blood sample (or other sample) to be shipped away to a physical laboratory. The PON instrument 200 may be configured to communicate biomarker data as a digital signal to the biomarker identification system 400. The biomarker identification system 400 may comprise a receiver 408 configured to receive the biomarker data. Received data may be forwarded to an LIS 418 of the biomarker identification system 400. The LIS 418 may be configured to parse the data from the format it arrived at the biomarker identification system 400 in, into a suitable format to be input to the classification / regression model. The LIS 418 may be configured to run scripts to coerce input biomarker data into functional data frames to be input into the model. The functional data frames may be stored in .json or JSON format in a cloud database. The functional data frames may be stored in a BLOB (binary large object) format in a cloud database. It is beneficial to send the data away from the PON instrument 200 for analysis and evaluation. Although the PON instrument 200 may comprise on-board processing circuitry that may be configured to identify biomarkers in a sample, transferring the biomarker data outside of the PON instrument 200 to the remote server 438 leads to several benefits, including but not limited to: 1) Combining several biomarkers for augmented and specialized diagnostics 2) Combining results from several PON instruments in the diagnostic algorithm 3) Easy implementation of auxiliary diagnostic services, new and / or updated algorithms and features, implementation of new biological conditions. Beneficially, sending the biomarker from the PON instrument 200 in a first data format and then parsing the data into a second – different – data format for assessment may enable easy implementation of auxiliary diagnostic services, new and / or updated algorithms and features, implementation of new biological conditions. The PON instrument 200 is configured to provide web connectivity and functional parsing of blood sample results directly to a database of the remote server 438 and the present biomarker identification services. Device firmware of the PON instrument 200 may be configured to support this feature, i.e. to redirect data to the LIS (laboratory information system) end-point that is compatible with both HL7 (health level 7) and FHIR (HL7 Fast Healthcare Interoperability Resources) data messages (payloads) as described below. Metadata can be included in the biomarker data. For example, the biomarker data obtained from a particular animal fluid sample (e.g. fish fluid sample) may be combined with metadata such that the LIS 418 can identify a time that the sample was obtained, a location where the sample was obtained, an environmental condition when the sample was obtained (e.g. ambient temperature or salinity) and the like from the metadata. The processor 206 may be configured to include metadata to the biomarker data indicating which rotor cartridge 202 the data came from, for example. Including metadata may aid the LIS 418 in organising biomarker data arriving at the biomarker identification system 400. The receiver 408 may be configured to receive a request from the PON instrument 200 to perform an assessment of biomarker data using the classification / regression model. A request may be sent to the model to perform analysis of the biomarker data, to provide analysis of the fish 102 from which the fish fluid sample was obtained, in the example where the animal is a fish, and the relevant animal otherwise. A physiological animal characteristic status may then be received following classification of the biomarker data by the model. In an example, the physiological fish characteristic status is a smoltification status. As above, a fish fluid sample is obtained from fish 102 and biomarkers are identified in the sample. The biomarker(s) searched for in the sample are biomarkers that can point towards smoltification characteristics. The parsed biomarker data is suitable to input into the model to produce a smoltification status. A smolitification status request is issued to the model, and an output of the model is a smolitification status. Where the server 438 is a cloud-based server, its components may be cloud based. The biomarker identification system 400 may include infrastructure to support a cloud-based server 438, such as a processor 406. The receiver or transceiver 408 may be configured to receive a request from the PON instrument 200 to perform an animal analysis service. The processor or processing circuitry 406 may be configured to execute a computer program product comprising computer code 420 to cause the server 438 to provide the animal analysis service, which may include retrieving biomarker data received from the PON instrument 200, parsing the data at the LIS 418, inputting the parsed data into the model and running the model to generate a physiological animal characteristic status. Figure 4 shows operating system application(s) 414, which may be configured to run basic functions of the server 438. A role of the operating system 414 is to execute animal analysis service application(s) 416. The animal analysis service application(s) 416 may be configured to cause the model to run and analyse the data from the animal fluid sample. The animal analysis service application(s) 416 may be configured to be executed by the processor or processing circuitry 406 to perform a animal analysis service as described herein. The server 438 may comprise an I / O 420 for inputting and / or outputting data to and from the server 438. For example, the server 438 may comprise a user interface 404, which may be configured to allow a user to control operations of the server 438. For example, a user may be enabled to trigger animal analysis using the user interface 404. The server 438 may comprise a display 412. The display 412 and user interface 404 may be provided by the same touchscreen, for example. The model may be accessed by the animal analysis service application(s) 416 via model serving, rather than being embedded in the application 416. An API (application programming interface) may be implemented to access the model. The API may be remote – i.e. not located at the PON instrument 200, which is where a request for the biomarker identification system to perform animal analysis originates and not located at the server 438. The API may be a web API. The request for animal analysis may be formatted as a HTTP request message; however, a suitable request format is a JSON (JavaScript Object Notation) data frame. The JSON data frame may comprise parsed biomarker data that has been parsed by the LIS 418. The JSON data exchange format is considered suitable for this use because of its lightweight human-readable text, for simple processing, in a manner in which it is simple to relay biomarker data. Figure 5 shows an example decision tree used to characterise the fish 102 of the example of Figure 1. The model may comprise a machine-learning (ML) classification algorithm. The ML classification algorithm may use decision trees to generate fish status indicators based on input biomarker data indicative of biomarker content in the fish fluid sample. A fish status indicator is an indicator of whether the fish (from which the sample was obtained) meets a particular physiological criteria and multiple fish status indicators may be considered together (for example, reviewed as a set or aggregated) to arrive at a physiological fish characteristic status. Of course, where the animal is not a fish 102, a decision tree according to Figure 5 may be used to characterise the animal as described above. In order to determine an ML_score (score output from the model) on which the physiological fish characterisation status (in this example, although more generally an “animal” characterisation status) can be based, the model may be set to “regression” mode, to use regression trees as appropriate. The model may otherwise be set in a classification mode to predict binary conditions, as opposed to the values shown in the leaves in Figure 5. The model may be configured to tune itself automatically based on the parsed biomarker data that is input into the model. The trees may be boosted such that the ensemble of decision trees is generated where each new instance is trained to emphasize previously mis- modelled instances. Depending on what animal characteristic is being assessed (e.g. smolt status, health status, stress status, gonade status, metabolic status), two or more animal status indicators may be available to indicate the result of the animal characteristic assessment. Where two animal status indicators are available, one may be positive and one may be negative: an example animal status indicator is “poor health” and another is “good health”. Another example animal status indicator is “fertile” and another is “infertile”. Based on the biomarkers appearing in the animal fluid sample, various animal statuses can be identified. The model may be configured to generate a score, to be compared with a threshold, and based on that score may attribute a animal status indicator. To get a picture of the animal’s physiological characteristics, multiple animal status indicators may be compiled to reflect different physiological properties of the animal. The animal status indicators may not be limited to being binary (e.g. positive or negative) – for example, the model may be configured to generate a predictive animal status indicator. In an example where the animal is a fish, an example predictive fish status indicator is “likely smoltification within [timeframe]”. Where the fish status indicator is predictive, the model may be configured to identify a score – as above – and proceed to identify that the score is within a pre-determined range of values. The range of values may be provided when training the model and / or may be entered during real use of the model, for example by a user using a user interface. Based on the range of values, the model may determine a window of time within which the score is likely to rise or fall in order to enter a different range of values, and thus predict when the fish status will change. For example, the model may determine a score of 96 out of a possible 100 (by way of example only). The model may be trained to identify that 96 out of 100 means that the fish from which the sample was taken is currently saltwater viable. This may be the fish status indicator – “saltwater viable”, which may be provided to the user. However, the model may (for example, in addition) determine that because the score is between 80 and 100 (by way of example only), that fish will desmoltify in 2 weeks, for example. Thus, a predictive fish status indicator of “will not be saltwater viable after 2 weeks” may be provided by the model. The animal status indicator(s) may be provided to the user as well as or instead of an overall physiological animal characterisation status. The physiological animal characterisation status may only comprise one animal status indicator, such that they are one and the same, where the user only wants to analyse one particular animal status. The model may employ one or more decision trees to arrive at the physiological animal characterisation status. Figure 5 shows an example ML algorithm decision tree, labelled Tree 1. Biomarkers that are present in the sample (and so in the biomarker data) are assessed to grow the decision tree. A first biomarker is assessed at step 502. The example first biomarker is potassium. The biomarker data is analysed and the split of the node is based on information Gain as shown. The values are examples only. Following step 502, the decision tree arrives at a leaf at step 504; i.e. a node that will not be further split. A branch follows step 502 and another biomarker is assessed at step 506 (a decision node) – this time, in the example, magnesium is the biomarker to be assessed. Another leaf 508 and another branch 510 follow. At decision node 510 another biomarker is assessed – this time, in this example, total carbon dioxide. Following step 510, two leaves are arrived at – steps 512 and 514. This is only an example of a single decision tree that could contribute to the model. The units annotated in the example decision tree of Figure 5 are normalized (i.e. all variables scaled according to respective (max-min) / standard deviation) rather than using the absolute biomarker values obtained from fish fluid samples. The scaling of the biomarker data may take place at the LIS 418, for example. The example tree shown in Figure 5 may be one of an ensemble that the model comprises. The model is configured to generate a physiological animal characteristic status based on the results at the leaves, by generating an ML_score and interpreting the meaning of the score (a numerical score) in terms of a animal health status. In some embodiments the ML algorithm comprises an embodiment of the XGBOOST ML algorithm, which is a supervised ML algorithm that implements optimized distributed gradient boosting ML algorithms using the gradient boosting framework. XGBoost stands for extreme gradient boosting, which is a scalable distributed gradient-boosted decision tree (GBDT) ML library providing parallel trees boosting using supervised machine learning ML algorithm to implement multiple decision trees, ensemble learning, and gradient boosting. The XGBoost algorithm uses second-order gradients of the loss function in addition to the first order gradients, based on a Taylor expansion of the loss function for the test data used as training data to configure the XGBoost ML algorithm in some embodiments of the disclosed technology. In some embodiments of the XGBoost algorithm, a Taylor expansion of other loss functions may be used instead in some embodiments, for example, a logistic loss for binary classification, such as a binary classification of smoltification status indicating a fish is salt-water viable or non-viable may be used. An alternative to the XGBoost algorithm or similar gradient boosting ML algorithms may be any implementation of the Random forest algorithm, variance clustering or an ML algorithm that uses deep learning. A gradient boosting algorithm takes a set of labelled (also known as target) training instances as input and builds a model that aims to predict the label (target) of each training example correctly based on other non-label information known for a sample (the features of the instance). Once trained, the model should be able label future data accurately with unknown labels. One or more decision trees can be used to minimise a suitable loss function. To prevent overfitting, the XGBoost algorithm transforms the loss function into an objective function containing regularisation terms which adds penalty terms to the loss function for adding new decision tree leaves to the model which add a penalty proportional to the size of the leaf weights (see Figure 5 for an example). Without the regularisation terms, gradient boosted models can quickly become large and over fit to noise present in the training data, which would result in a poor performance of the algorithm when presented with new test data (such as the biomarker data). XGBoost is an example of an ML algorithm which includes an exhaustive list of six or seven hyperparameters and these are automatically selected. Each hyperparameter may be tuned, having values from zero to infinity. The hyperparameters provide ML model specifications – i.e. tree depth, number of trees, learning rate, loss reduction, sample size etc. These hyperparameters affect how the final ML model interprets new data during training and when making predictions. XGBoost performs parallel processing such that the model can be trained more quickly than models having algorithms that do not perform parallel processing. Retraining of the model may be triggered when new test data is received, automatically. A graphics processing unit (GPU) may be used for calculations to solve parallel problems and in this context a GPU algorithm may be used for gradient boosting. The training data set may be divided into equal sized groups or quantiles. This improves the accuracy of the trained algorithm and simplifies the tree construction algorithm and enables implementation to be performed more efficiently. The model may generate a score (or ML_score) based on the values at the leaves of the decision tree(s). The score may be a number on which the physiological animal characteristic status is based. For example, the score may be a number between 1 and 100. The model may be configured to compare the score with a threshold value and, based on the comparison, generate the physiological animal characteristic status. The model may be configured to output a score “x”. This may be compared with a threshold “y” which determines whether the animal meets a physiological criterion or not – any of the example assessments shown and described in relation to Figure 6 below may have a threshold “y” determining which of two assessment outcomes are true of the sample. In one example, smolt status is being assessed and based on the score it can be determined whether the fish from which the sample was taken is saltwater viable or not. The ML_score may be “93” and the threshold may be “50”, where a value above 50 indicates that the fish is viable to be transferred to saltwater, for example – the model may be configured to determine that 93 exceeds and 50 and generate a physiological fish characteristic status reflecting that the fish is viable to be transferred. The threshold (“y”) may be a probability cut-off between 0-100% - e.g.50 as above. The threshold for a particular physiological status that is being assessed may be a predetermined and fixed value. The threshold may be a continuously drifting threshold that changes as the model is trained and re-trained. The threshold may be adjusted based on new biomarker data, for example. Through use of the model, the threshold may be adjusted to make the ML_score more accurate, for example. The threshold may be adjusted based on one or more scripts configured to determine optimal threshold values for the physiological statuses being assessed, using ROC curves for example. The scripts may be automated scripts and the changes to the threshold may occur at any given time with the biomarker data available. The physiological animal characteristic status may be an overall result based on multiple animal status indicators – i.e. a single result may be generated by compiling multiple statuses to paint an overall picture of the animal’s wellbeing. The physiological animal characteristic status may be based on a compilation of the values at one or more (for example, each) leaves of the ML algorithm decision tree. The model may be configured to output a comment based on the ML_score, which is not the ultimate physiological animal characteristic status, but is instead a comment on a single animal characteristic that can contribute to an overall physiological animal characteristic status. The comment may indicate to a non-professional what the ML_score means, for example as above a score of “93” where the threshold is “50” may result in a comment like “positive result” or “saltwater viable” or the like, but the overall physiological fish characteristic status may be output as “healthy fish” based on more than one ML_score. The generation of the physiological animal characteristic status may occur outside of the model. The model may be configured to output the score to another part of the biomarker identification system 400 – the processor 406 and / or the LIS 418 may be configured to generate the physiological animal characteristic status based on the score, for example. For the model to assess the biomarkers in the sample, the biomarker data is sent from the PON instrument 200 to the server 438. Biomarker data may be input into the processor 206, which may record the data and may organise the data at the PON instrument 200. The data may be structured suitably to input into the model. The data may be structured suitably to send to a laboratory information system (LIS) for parsing, for input into the model. Example data formats include a format meeting the health level 7 (HL7) primary standards and a format meeting the HL7 Fast Healthcare Interoperability Resources (FHIR) standard. An example suitable data format is JSON (JavaScript Object Notation) for data transfer. The HL7 primary standards are a set of international data standards for transferring clinical data in particular. This data format is flexible to allow different systems to communicate. At the time of this application, the HL7 primary standards include: Version 2.x Messaging Standard – an interoperability specification for health and medical transactions; Version 3 Messaging Standard – an interoperability specification for health and medical transactions; Clinical Document Architecture (CDA) – an exchange model for clinical documents, based on HL7 Version 3; Continuity of Care Document (CCD) – a US specification for the exchange of medical summaries, based on CDA; Structured Product Labelling (SPL) – the published information that accompanies a medicine, based on HL7 Version 3; Clinical Context Object Workgroup (CCOW) – an interoperability specification for the visual integration of user applications. Source: https: / / en.wikipedia.org / wiki / Health_Level_7 Version 2 comprises non-XML encoding syntax based on segments / lines and one-character delimiters. This data format is compatible with laboratory information systems (LIS). Version 3 comprises XML encoding syntax. The following is an example of an actual HL7 message transferred from a PON instrument 200 to an LIS API, as an example of HL7 format data: MSH|^~\&|SMT|SMT- 120VP|||20230513220842||ORU^R01|9|P|2.3.1|A|||0||ASCII|||V1.00.01.09| PID|9|4|||^||A|||Delphinidae||||||||1||||||||||||| OBR|9||2|121004686|||20230513212647||||||1^^||BloodSerum|||71^922542^01985922542^Bl ood gas plus parameters|0^0^0^1||||||||||||||||||||||||||||| OBX|1|NM|BUN / CREA|BUN / CREA|150.158|||N|||F||||||| OBX|2|NM|Ca|Ca|2.69|mmol / L||N|||F||||||| OBX|3|NM|GLU|GLU|6.09|mmol / L||N|||F||||||| OBX|4|NM|LAC|LAC|12.12|mmol / L||N|||F||||||| OBX|5|NM|Cl|Cl|140.9|mmol / L||N|||F||||||| OBX|6|NM|BUN|BUN|3.47|mmol / L||N|||F||||||| OBX|7|NM|tCO2|tCO2|7.7|mmol / L||N|||F||||||| OBX|8|NM|Na|Na|152.3|mmol / L||N|||F||||||| OBX|9|NM|Mg|Mg|1.35|mmol / L||N|||F||||||| OBX|10|NM|Crea|Crea|23.1|umol / L||N|||F||||||| OBX|11|NM|PHOS|PHOS||mmol / L||N|> 6.00||F||||||| OBX|12|NM|K|K|6.09|mmol / L||N|||F||||||| OBX|13|NM|pH|pH|6.86|||N|||F||||||| The example includes example biomarkers (e.g. Ca) and indicates that the sample is a sample of blood serum. Concentrations of the present biomarkers are indicated in mmol / L. The FHIR format is a type of HL7 format that provides a secure way to transfer data between computer systems regardless of how the data is stored in those systems. An example data format for biomarker data being transmitted from the PON instrument 200 is FHIR encoded as JSON. Data in this format may be delivered to the LIS 418. A benefit of using FHIR encoded as JSON is that messages may be sent via HTTP clients and be utilized in conventional application programming interfaces (APIs). Figure 6 shows system architecture of the biomarker identification system 400. One or more PON instruments 200i, 200ii, 200iii (three are shown in the example of Figure 6) are used in the field to obtain samples from animal 102. In the example where the animal is fish, the number of fish 102 that may be represented by a small number of samples may be up to 100,000 or even 200,000 in a fish farm environment. A population level in the 100,000s means that multiple PON instruments 200 may be used at the fish location to collect a sufficient number of samples to represent the population level. In the present disclosure, the biomarker data converges at the biomarker identification system 400. Other data may also be provided to the biomarker identification system 400, if a different assaying device has been used on a sample. For example, instead of observing absorption and collecting biomarker data, a PON instrument may collect data to indicate biomarkers through PCR testing or the like. Once the data has been collected, the biomarker identification system 400 is device agnostic – in other words, the instrument used does not matter to the system 400. The LIS 418 may be configured to receive data in different formats and parse the data consistently; so long as the data is parsed consistently for input to the model, the biomarker identification system 400 is measuring device agnostic. In this way, the PON instrument 200 may be developed / progressed and the biomarker identification system 400 will still be a relevant tool. PON instruments 200i, 200ii, 200iii may therefore represent different types of assaying device, not just the PON instrument 200 described herein. Once biomarker contents have been identified in a sample from an instrument 200i, 200ii, 200iii, biomarker information may be sent to a biomarker metrics data store 402. The biomarker metrics data store 402 may be located in network-connected storage or distributed cloud storage, for example. The biomarkers metrics data store 402 may be configured to provide input data to a server 438 of the biomarker identification system, for analysis. The server 438 may be configured to perform one or more assessments of input biomarker information. Figure 6 shows a collection of example assessments and example results, which are passed to output analysis 436 for generating a classification for outputting. The classification may be a physiological fish characteristic status that is based on multiple assessments. Each assessment may generate a fish status indicator. The physiological fish characteristic status may be a result of one of the assessments (i.e. may be a fish status indicator, which goes on to be output from the analysis) or may compile results from multiple assessments. From left to right in Figure 6, there is shown an assessment performing smolt (or smoltification) status analysis 604, health status analysis 606, stress status analysis 610, reproductive and sexual health status analysis (or gonade analysis) 612 and metabolic status analysis 614. Another assessment that may be performed is a general physiological analysis, where a general physiological status indicates biological function as circulation, respiration, fluid balance. Each assessment may be performed by a decision tree of the ensemble of decision trees forming the model. An animal health status is an example animal status indicator and may represent whether an animal is free from disease, infections, virus, bacteria and parasites. The animal health status may be based on immune response to pathogens and inflammatory response in mucosal and intestinal barriers, which may be indicated by the biomarkers in the sample. Which biomarkers are selected to use in identifying which animal health statuses may evolve over time, as modelling develops or as testing techniques progress. Presently, biomarkers are ranked as to how useful they are in indicating a particular animal heath status and a number of the better ranking biomarkers are assessed. For example, for identifying stress, biomarkers phosphorus, Sodium, Calcium, Growth hormone, Chloride, Cortisol, Ureic acid, Magnesium, Glucose, Globulin, Total protein, Albumin may be assessed as these are presently considered by the inventors to be useful biomarkers for identifying stress in animals. In an example where the animal is fish, a fish stress status may represent whether fish 102 have the ability to adapt and survive handling as part of the farming procedures, including crowding, pumping, sorting, grading and transportation within or between farming sites. The stress status of the fish 102 may be based on the presence of cortisol, adrenaline or noradrenaline in conjunction with any panel of related biomarkers that produce predictive value with regard to animal stress responses. A gonade status may represent sexual maturation. A metabolic status may represent growth performance, ability to convert feed to growth, and quality of the aquatic organisms in terms of economic value – the status may include colour, texture and biochemical composition. In an example where the animal is fish, fish gonade status may be identifiable using one or more of the same biomarkers as for stress, for example, or different biomarkers. It may be possible to identify fish gonade status from certain biomarkers, impossible from others, and then of the possible biomarkers there may be a ranking of which are most useful (or provide the best indication) of gonade status. This may be true of any fish health status being explored. The model may be configured to perform all of the possible assessments that it is trained to perform, automatically, once parsed biomarker data has been input into the model. The model may perform all of the types of assessment that it is capable of performing on input biomarker data and may return an output indicating that a meaningful result cannot be obtained from the input biomarker data where a meaningful result cannot be obtained. For example, the model may determine that a biomarker required for determining stress status is absent and therefore that stress status cannot be concluded, so a stress status analysis is not performed. Or, stress status analysis is performed in the absence of a required biomarker and the result is returned as inconclusive. Alternatively, the request for a biomarker analysis service that is issued by the PON instrument 200 may include a request for one or more particular assessments to be performed. For example, the user may not want to receive analysis results relating to animal stress and may only be concerned about gonade status. The request may comprise a request for gonade status analysis only and the model may be configured to generate only a gonade status score, leading to a physiological animal characteristic status based only on the gonade status score. The model may only step through one decision tree to arrive at a gonade status score. The results of each assessment may be binary and provide one of two possible animal status indicators. For example, the results of the smolt status analysis 604 in Figure 6 are shown to be saltwater viable 616 or saltwater non-viable 618. In the case of assessing health status, health status analysis 606 is performed and the result may be poor health 620 or good health 622 – each of which is an animal status indicator. In the case of assessing stress status, stress status analysis 610 is performed and the result may be high stress 624 or low stress 626. In the case of assessing sexual and reproductive health, gonade status analysis 612 is performed and the result may be infertile or non-fertile 628 or fertile 630. Some fish lay eggs and some fish give birth to live fry – it may be that gonade status analysis returns a result that a fish 102 is currently pregnant, for example, depending on what sort of fish 102 are being analysed. In the case of assessing metabolic status, metabolic status analysis 614 is performed and the result may be thriving 632 or not thriving 634. Biomarkers assessed by the model to determine metabolic status may include any combination of total cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, Aspartate aminotransferase (AST), alanine aminotransferase (ALT), blood glucose levels, vitamin levels - in conjunction with any panel of related biomarkers that produce predictive value with regard to metabolic dysfunction in animals. The model is configured to analyse results from assessments of the input biomarker information at output analysis 636. Output analysis 636 may comprise compiling outcomes from multiple decision trees. The outcome of output analysis comprises the physiological animal characteristic status. The physiological animal characteristic status may comprise one or more animal status indicators. For example, the physiological animal characteristic status may include each animal status indicator determined by the animal analysis server 438. The physiological animal characteristic status may include an aggregation of two or more animal status indicators. The physiological animal characteristic status may be an overall conclusion based on output analysis 636, for example “viable to be moved from freshwater to saltwater”. The physiological animal characteristic status may comprise multiple animal status indicators, for example “healthy and fertile but not suitable to be transferred to saltwater”. This is because multiple assessments may be performed on samples and multiple biomarkers may be identified and classified to form a conclusion at output analysis 636, as shown in Figure 6. Information to be output for a user to view may be in the form of an animal analysis report, comprising the physiological animal characteristic status. The biomarker identification system 400 may comprise a processor or processing circuitry 406 configured to process the outcome of output analysis 436 into the report (which may take a user-friendly form, such as words, numbers or a chart). Additionally or alternatively to providing the physiological animal characteristic status as a conclusion of viability of whether fish 102 are capable of surviving in saltwater for example, the physiological animal characteristic status may comprise an indicator of saline osmoregulation value, for example, from which the user can infer whether fish 102 are viable for survival in saltwater. The nature of the physiological animal characteristic status may depend on the preferences, skills or knowledge of the intended end user; for example, a fish farmer may want the result in the form of “yes” or “no” fish 102 can be moved to saltwater, whereas a scientist studying fish 102 or a veterinarian, for example, may want to see an indicator such as a saline osmoregulation value or other more detailed indicators. The biomarker identification system 400 may comprise a mobile application, for example, or a webpage viewable in a web browser for presentation of the animal analysis report. The animal analysis report is transmitted to the user computing device, which is operable to provide a customer portal user interface or dashboard through which a user (e.g. farmer) can access the physiological animal characteristic status based on data from the PON instrument from which the biomarker data was sent. The user computing device 703a may be any suitable device having web access and the user interface may be web-based to provide the physiological animal characteristic status. The user computing device 703a may be portable, such as a smart phone or tablet; this would be particularly useful for a user in possession of the PON instrument 200i, 200ii, 200iii in the field, looking to get feedback from the server 438 as quickly and efficiently as possible, where they access the customer portal user interface or dashboard on their user computing device 703a whilst at the point of need. They could take action to transfer a fish 102, for example, while still at the point of need and based on the output analysis 636. The animal analysis report may be delivered to more than one computing device provided the customer portal user interface or dashboard is accessible using the computing device, such that a user who is not currently with the PON instrument 200 can see the results or so that a user who wishes to study the results away from the animal’s location can do so. The outcome of output analysis 636 or a report based thereon may be deliverable via the customer portal user interface or dashboard to another computing device as a file, for example as an email attachment. A benefit of the present disclosure is that the physiological animal characteristic status can be received at the point of need via the user computing device 703a to allow quick action based on the results, but it may also be beneficial to deliver results to a laboratory or other location away from the point of need, for example for data collation or study. The physiological animal characteristic status may be delivered in the form of a report / as part of a report, which may be a summary of the outcome of output analysis 636, which may include text (e.g. “animal is not viable”), tables of numbers, charts, diagrams and the like. The report may be electronically actionable, for example the user may be able to choose between different possible representations of the outcome of output analysis 636. The report may be actionable – i.e. there may be interactive elements to the manner in which the outcome of output analysis 636 is presented to the user. The physiological animal characteristic status may be supplemented for presentation to a user; for example, one or more external links, enabling the end user to follow a link to obtain additional information beyond the report, may be provided to the user alongside the report or as part of the report. Information generated as part of output analysis 636 may be transmitted to a data store and may be accessible to a user via a link associated with or included in their report. The report may be supplemented with additional information alongside the outcome of output analysis 636, such as educational information about how to address issues with the animal identified by the biomarker identification system 400. The report may include one or more external links to additional content such as webpages or videos that may include relevant information for interpreting the output analysis 436 or report, for example a video detailing how to reduce stress in animals. Additional content such as images or video may be included in the report to supplement the outcome of the output analysis 636. In one example, the display 212 of the PON instrument 200 itself may be configured to display the physiological animal characteristic status to a user. The PON instrument 200 may comprise a computer application configured to receive and display the physiological animal characteristic status application. Otherwise, a computer application on the user computing device may be so configured, or the dashboard / user interface may be provided in a browser as described above. The computer application may comprise a dashboard for user navigation to allow them to view the physiological animal characteristic status, multiple physiological animal characteristic statuses, historical physiological animal characteristic statuses or to select a desired format for presenting the physiological animal characteristic status (e.g. a table). The output analysis 636 (without being summarised into a physiological animal characteristic status) or a report based thereon may be viewable via the dashboard. The PON instrument 200 or the user computing device 703a may be configured to store earlier output analysis 636 or reports based thereon and the user may be enabled to access earlier output analysis 636 or reports based thereon via the computer application, for example. This may enable a user to follow progress of animal from an earlier time to the present time. The time taken between inserting the rotor cartridge into the PON instrument 200 and receiving the physiological animal characteristic status or another output analysis report may be less than 24 hours. Figure 7A shows a representation of the turnaround between taking the animal fluid sample and receiving the physiological animal characteristic status or other output (e.g. data in a different form from the physiological animal characteristic status). The turnaround may be significantly shorter than other solutions available at the time of this application. For example, the time taken between inserting the rotor cartridge 202 into the PON instrument 200 (step 701) and receiving the physiological animal characteristic status or another output analysis report (step 703) may be less than 12 hours, may be less than 6 hours, may even be less than 1 hour. It is estimated that the time taken between inserting the rotor cartridge 202 into the PON instrument 200 (step 701) and receiving the physiological animal characteristic status or another output analysis report (step 703) may be as low as 15 minutes, 12 minutes or even less. The constraints on the turnaround time are the time taken for chemical reactions between the sample and reagents and the optical analysis performed by the analyser 220. The transmission of the biomarker data, the processing time, the time for analysis at the server 438 (step 702) and the time to provide the output (e.g. the physiological animal characteristic status) is minimal. Currently, the analyser 220 can perform optical analysis on a sample and generate biomarker data in around 12 minutes. This may be reduced in future and the biomarker identification system only minimally adds to the time taken between inserting the sample and receiving results from the model. Step 701 is performed at the location of the animal and it is envisioned that the user computing device 703a will accompany the user at the point of need such that they can receive the results at the point of need (here, the location of the animal). Figure 7B shows an example turnaround where the result is delivered back to the PON instrument 200, where the rest of the process is the same as shown and described in relation to Figure 7A. Figure 8 depicts a method 800 for training the model. Figure 8 shows a training method 800 including taking a sample 804. The sample may be taken using a PON instrument 200 as described above. The sample(s) for use in training the model may be obtained in a different manner from the sample(s) that are taken when using the model for real assessments. There is no need to use the PON instrument 200 to generate the training data – for example, to generate sufficient test data, animal may be sampled en masse and using the PON instrument 200 may not obtain enough of a bulk of data. The method 800 includes inputting 802 empirical animal status information. Empirical animal status information may relate to a combination of characteristics of the animal, for example empirical smoltification status information (information indicating whether a fish is undergoing or has undergone smoltification), empirical health status information (information indicating whether a fish 102 is healthy or not healthy), empirical stress status information (information indicating whether a fish 102 is stressed or not stressed) or empirical reproductive status information (information indicating whether a fish 102 is fertile or not), for example. The empirical animal status information (any of the status examples) may have been obtained from tests performed on a test group of animals. Public data may be used as input data; however, the present model is trained using original test results. The empirical animal status information may have been obtained from test data using typical animal analysis methods such as PCR testing. The model may be trained in order to produce a score and / or physiological animal characterisation status that is sufficiently representative of the sample compared with the empirical animal status information. The empirical animal status information may comprise indications of biomarkers present in test samples and what physiological animal characteristic statuses animal having those biomarkers present in their samples possess. For example, the empirical animal status information may describe what biomarkers are present, in what relative amounts, in a animal that is chronically stressed. The method 800 also includes taking a sample 804 and obtaining biomarker data 806. This may be done using the PON instrument 200 or another suitable analysis tool (as at this stage, the model is being trained and efficiency at the point of need is not necessarily the goal when obtaining the biomarker data). Training data is generated 808 based on the obtained sample biomarkers 806 and the input empirical animal status information. From the empirical information, the meaning of the biomarker data can be interpreted and appropriately designated a animal status indicator. At step 808, the model may be trained to recognise the presence of an example biomarker over or under an example threshold amount as being indicative of an example animal status indicator. Once the training data has been generated, the training data is input into the model at step 810. The model is trained – see step 812. The model may be continuously trained in use, as new biomarker data is provided to the model. The training protocol trains thousands of iterations and selects the best model, which is ranked for example using ROC-AUC. Figure 9A shows a data flow between an example PON instrument 200 and an example ML server 438 for obtaining a physiological animal characteristic status. As shown, the data flow 900 comprises obtaining a sample 901 as described above. The PON instrument 200 receives the sample as an input, is configured to analyse the sample 904 to identify biomarkers in in the sample 906. As shown in Figure 9A, the PON instrument 200 may be configured to indicate some analysis results 916 (labelled as on-board result, meaning not yet sent to the ML server 438 but instead calculated at the PON instrument 200) simply based on analysis performed at the PON instrument 200, without first requesting an animal analysis service. The results may be displayed at the display 212, for example. These results could be detected wavelength results from the light detector, may be data processed by the processor 206 showing spectra after the analyser 220 has analysed the sample, and the like. Any data that has not yet been input to the model could be provided to the user of the PON instrument 200 via the display 212. The biomarker data from the PON instrument 200 may be transmitted to the ML server 438 at step 906. The biomarker data may be provided as an input for the model at step 908. As described above, the biomarker data sent from the PON instrument 200 may be processed first, before being input into the model, to take a suitable format for being input into the model. For example, the biomarker data may be parsed at the LIS 418. At step 910, the model may be run on the input data to determine a physiological animal characteristic status. At step 914, the physiological animal characteristic status (or output data in any suitable format, such as a collection of animal status indicators) is transmitted to the user computing device 703a for the user to receive. In one example described above and as shown in Figure 9B, the results are transmitted to the PON instrument 200 itself – this may be an alternative to the user computing device 703a if the PON instrument has the necessary computing capabilities, or in addition (which may supplement the results sent to the user computing device). In any event, it is anticipated that the user computing device 703a will be used to access the results at the PON location and so the benefit of being able to take action based on the results while at the PON location can be achieved. Steps taking place at the PON location are shown enclosed in the dashed box in Figure 9A. In the example of Figure 9B, the benefit of getting results at the point of need is inherent from the results being received to the PON instrument 200 which was used to transmit the biomarker data to the server 438. So far, the model for classifying animals has been described in terms of producing a physiological animal characteristic status. A physiological animal characteristic status may also be referred to as an ML_prediction, as an output of the model. Figure 10 shows an example break down profile of a sample. A user interface may be provided that includes one or more input fields for entering biomarker values as feature variables for the model. The user may be enabled to enter biomarker values – for example by typing – or the input fields may be automatically populated based on data received to the server 438. The biomarkers that can be entered in this particular example are potassium, sodium, chloride, total CO2, calcium, magnesium and phosphate. These values may be altered as the model is trained, for example. The biomarkers may be altered to train the model for more than (in this example) seven biomarkers. The biomarkers may be chosen based on what physiological characteristic(s) is / are being explored by the PON instrument 200, as desired by the farmer. In Figure 10, an intercept is indicated by a dotted line and the label “xgboost”. This intercept represents the model default prediction if all input variables are set to zero. In the example shown, an XGBoost algorithm has been used to make the prediction. The Break Down profile shows the exact contribution of each biomarker to the ML_prediction. The Break Down Profile shows the prediction made by the model as “92.6”. From that value, or ML_score, it can be established that the ML_prediction is “Saltwater viable”. The threshold values for saltwater viability may be input into the model during training or the model may be updated as time goes on as scientists identify appropriate thresholds, for example. The model may be trained based on threshold values found to indicate a animal’s health status in test data, for example. Currently, for saltwater viability, the scores are: Score >80: Saltwater viable and thriving Score 40-80: Saltwater viable, but struggling Score <40: Not saltwater viable Therefore, a prediction of 92.6 as in this example would indicate a saltwater viable and thriving fish from which the sample was taken. From the information presented in Figure 10, calculations can be made to determine how long this particular fish will be saltwater viable (i.e. how long is the window before the fish desmoltifies?). Such a prediction is possible when the ML model has been trained with sufficient amounts of empirical data describing the fish biomarker profile environmental milieu (fish age, water temperature, water salinity, water pH etc.) so that it is able to predict the developmental status and future developmental trajectory of the fish, which in combination with the ML_prediction can determine (i) the health status of the fish (e.g. smoltification or stress status) and (ii) the duration of that status being valid (e.g. how long before desmoltification). In the example of Figure 10, the window is around 3 weeks. Figure 11 shows an example user interface that provides input fields (as described above) for entering biomarker values as feature variables for the model. However, in this case the ML_prediction is negative – “Not saltwater viable”. This physiological fish characteristic status is based on the ML_score being 28, which is calculated using the fish biomarker values entered in the input fields. In this case, the range of saltwater non-viable predictions would be between 0-40 as set out above. Based on the information presented in Figure 11, it can be established that the fish from which the sample was taken can be expected to be saltwater viable in 4 weeks. This estimation may be arrived at by the model being sufficiently trained with empirical data describing the fish biomarker profile environmental milieu (fish age, water temperature, water salinity, water pH etc.) as above to predict – based on the ML_score – a range of time for the status to change sufficiently to push the ML_prediction from negative to positive. Of course, the farmer or other user is not only concerned with one individual fish 102. From the fish’s sample, population extrapolation can be performed to indicate that around 35% (in this example) of the fish are saltwater viable at the time when this particular fish is not. Extrapolating a population benchmark is feasible when a sample size of 10-40 fish have been analysed, i.e. using basic statistics to infer the status of the greater / general population. The distribution of ML scores from several sampled animals is evaluated by a one-tailed t- test in relation to the point estimate representing the cutoff value (e.g.80). This population benchmark hypothesis test will indicate if the sampled animals mean ML score, notwithstanding some degree of variance, is significantly above the cutoff. Alternatively, a 95% confidence interval of the sampled animal ML scores may be calculated and the population benchmark would be negative if the cutoff value exceeds the lower confidence interval limit. The farmer or other user can then make an informed decision about whether there is a sufficiently high percentage of the population ready to transfer to risk losing the remaining fish (which are not ready to move). With respect to other physiological characteristics, it may be that the model outputs a physiological animal characteristic status indicating that the sampled animal is “healthy” or “fertile” or “stressed” or any other of the described statuses. From this, the result may be extrapolated to a population level and the farmer / other user may be presented with an indication of the percentage of the animal population that are “healthy”, etc. From this, they may make an informed decision about the well-being of their animals and if any actions need to be taken to improve the quality of care given to the animal. Figure 12 shows the ML_prediction results from another, different ML model fitted to data produced from actual PCR testing, in which gene expression status of a gill gene required for saline osmoregulation is tested. The x-axis is the machine learning prediction of smoltification status (an example animal status indicator). The y-axis is the measured PCR result. The points show different smoltification statuses, which are given binary labels 0 and 1 meaning “not viable” and “viable” accordingly. This is another ML model (a regression model trained to predict the PCR ‘gold standard’ metric of smoltification) intended to demonstrate and prove that the types of biomarkers being used by the model of this disclosure can be used to infer the objective smoltification status. The 45 degree line shows an optimal correlation between the predicted status and PCR result. The other line shows results attained by the present inventors, representing a strong correlation between the ML physiological animal characterisation status (which in this example is a smoltification status) from the alternative, earlier, model and the current gold standard of PCR testing. The alternative ML model featured in Figure 12 uses the same biomarkers as the model, but it is trained and validated to predict the objective PCR status of salmon. As shown, using the biomarkers and an ML model comes close to an actual PCR analysis result and thus the feasibility of the present model is evidenced (0-1 probability classification approach). As set out herein, versus PCR testing, the present approach reduces animal mortality as well as time to retrieve the output results and allows for different biomarkers to be assessed to look into different physiological animal characteristics. There is no need to delivery physical samples to a laboratory and the speed of returning a result to the user computing device 703a is limited only by how long the chemical reactions take to produce the biomarker data. The following clauses are examples of the present disclosure: 1. A point-of-need, PON, instrument for obtaining a metric for a physiological fish characteristic, the instrument comprising: an insertable rotor cartridge for sample assaying, comprising a plurality of reagent compartments for receiving a fish fluid sample, each reagent compartment comprising a reagent configured to react with a fish fluid sample; a rotor, configured to drive rotation of the rotor cartridge to agitate contents of the plurality of reagent compartments; an analyser, comprising an optical analysis setup and a processor, the analyser being configured to perform optical analysis on the contents of each reagent compartment after the reagent and fish fluid sample have reacted, to produce biomarker data; a transmitter configured to communicate with a remote biomarker identification system, wherein the analyser is configured to produce the biomarker data having a format allowing the biomarker data to be transmitted by the transmitter to the biomarker identification system for analysis; a receiver configured to receive a signal from the biomarker identification system; and a display configured to provide a result generated by the biomarker identification system responsive to analysing the received biomarker data to identify the presence of biomarkers in the sample, the result indicating a physiological fish characteristic status. 2. The PON instrument of clause 1, wherein the biomarker data has a format allowing the biomarker data to be processed at a laboratory information system (LIS) of the biomarker identification system. 3. The PON instrument of clause 1 or clause 2, wherein the biomarker data has a format meeting the health level 7, HL7, primary standards. 4. The PON instrument of clause 3, wherein the biomarker data has a format meeting the HL7 Fast Healthcare Interoperability Resources, FHIR, standard. 5. The PON instrument of clause 4, wherein the biomarker data has a format meeting the FHIR standard and is encoded as JavaScript Object Notation, JSON, data. 6. A physiological fish characteristic identification system, comprising: the PON instrument of any preceding clause; and a biomarker identification system remote from the PON instrument, comprising: a receiver, configured to receive biomarker data from the PON instrument; a memory, comprising a machine-learning model, wherein the model is a classification model configured to generate a physiological fish characteristic status based on the present biomarkers in the biomarker data; and a transmitter, configured to communicate the physiological fish characteristic status to the PON instrument. 7. The physiological fish characteristic identification system of clause 6, wherein the classification model is configured to identify the presence of a biomarker in the sample based on the biomarker data and attribute a fish status indicator to the biomarker data, wherein the classification model is configured to aggregate two or more fish status indicators to generate an overall physiological fish characteristic status. 8. The physiological fish characteristic identification system of clause 6 or clause 7, wherein the classification model comprises an XGBoost machine learning algorithm comprising an ensemble of decision trees, wherein each decision tree is configured to receive biomarker data that has been parsed from a first data format into a second data format as an input and wherein the ensemble of decision trees are configured to generate a score, wherein the physiological fish characteristic status is based on that score. 9. The physiological fish characteristic identification system of any one of clauses 6, 7 or 8, wherein the biomarker identification system further comprises a laboratory information system, LIS, wherein the LIS is configured to parse the received biomarker data from a first data format into a second data format suitable for inputting into the classification model and provide the parsed biomarker data as an input to the classification model, wherein the LIS is further configured to generate a report of the physiological fish characteristic status for communication to the PON instrument. 10. The physiological fish characteristic identification system of clause 9, further comprising at least one more PON instrument according to any preceding PON instrument claim, wherein the LIS is configured to parse received biomarker data from each PON instrument into a format suitable for inputting into the classification model and communicate parsed biomarker data from each of the PON instruments to the classification model. 11. The PON instrument of any one of clauses 1 to 5 or the physiological fish characteristic identification system of any one of clauses 6 to 10, wherein the physiological fish characteristic status comprises a viability status for fish to survive in salt water. 12. The PON instrument of any preceding PON instrument clause or the physiological fish characteristic identification system of any preceding physiological fish characteristic identification system clause, wherein the fish fluid sample comprises any of: blood, whole blood, plasma, serum, mucus, faeces and ascites fluid. 13. The physiological fish characteristic identification system of any preceding physiological fish characteristic identification system clause, wherein the classification model is configured to determine one or more fish status indicators based on the biomarkers in the sample, and wherein the classification model is configured to generate the physiological fish characteristic status based on the one or more fish status indicators, wherein the one or more fish status indicators includes at least one of: a health status, a stress status, a gonade status, a metabolic status, and a smoltification status. 14. The PON instrument of any preceding PON instrument clause or the physiological fish characteristic identification system of any preceding physiological fish characteristic identification system clause, wherein one of the fish status indicators is an indicator of chronic stress and wherein the biomarkers in the sample comprise phosphorus, sodium, calcium, growth hormone and chloride. 15. A computer-implemented method for determining a physiological fish characteristic status, the method comprising at a biomarker identification system: receiving biomarker data obtained from a fish fluid sample from a PON instrument, wherein the biomarker is in a first data format, and parsing the biomarker data at a server remote from the PON instrument into a second format suitable for inputting into a classification model; inputting the parsed biomarker data into the classification model at the server remote from the PON instrument, the classification model being configured to process the parsed biomarker data to obtain a plurality of fish status indicators; aggregating the plurality of fish status indicators into a single physiological fish characteristic status; and outputting, by the classification model, the physiological fish characteristic status. 16. A non-transitory computer readable medium comprising computer code which, when loaded from memory and executed by one or more processor(s) or processing circuitry, causes a biomarker identification system to perform a method according to clause 15. Where the disclosed technology is described with reference to drawings in the form of block diagrams and / or flowcharts, it is understood that several entities in the drawings, e.g., blocks of the block diagrams, and also combinations of entities in the drawings, can be implemented by computer program instructions, which instructions can be stored in a computer-readable memory, and also loaded onto a computer or other programmable data processing apparatus. Such computer program instructions can be provided to a processor of a general purpose computer, a special purpose computer and / or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer and / or other programmable data processing apparatus, create means for implementing the functions / acts specified in the block diagrams and / or flowchart block or blocks. In some implementations and according to some aspects of the disclosure, the functions or steps noted in the blocks can occur out of the order noted in the operational illustrations. For example, two blocks shown in succession can in fact be executed substantially concurrently or the blocks can sometimes be executed in the reverse order, depending upon the functionality / acts involved. Also, the functions or steps noted in the blocks can according to some aspects of the disclosure be executed continuously in a loop. The description of the example embodiments provided herein have been presented for the purposes of illustration. The description is not intended to be exhaustive or to limit example embodiments to the precise form disclosed, and modifications and variations are possible in light of the above teachings or may be acquired from practice of various alternatives to the provided embodiments. The examples discussed herein were chosen and described in order to explain the principles and the nature of various example embodiments and its practical application to enable one skilled in the art to utilize the example embodiments in various manners and with various modifications as are suited to the particular use contemplated. The features of the embodiments described herein may be combined in all possible combinations of methods, apparatus, modules, systems, and computer program products. It should be appreciated that the example embodiments presented herein may be practiced in any combination with each other. It should be noted that the word “comprising” does not necessarily exclude the presence of other elements, features, functions, or steps than those listed and the words “a” or “an” preceding an element do not exclude the presence of a plurality of such elements, features, functions, or steps. It should further be noted that any reference signs do not limit the scope of the claims, that the example embodiments may be implemented at least in part by means of both hardware and software, and that several “means”, “units” or “devices” may be represented by the same item of hardware. The various example embodiments described herein are described in the general context of methods, and may refer to elements, functions, steps or processes, one or more or all of which may be implemented in one aspect by a computer program product, embodied in a computer-readable medium, including computer-executable instructions, such as program code, executed by computers in networked environments. In the drawings and specification, there have been disclosed exemplary aspects of the disclosure. However, many variations and modifications can be made to these aspects which fall within the scope of the accompanying claims. Thus, the disclosure should be regarded as illustrative rather than restrictive in terms of supporting the claim scope which is not to be limited to the particular examples of the aspects and embodiments described above. The invention which is exemplified herein by the various aspects and embodiments described above has a scope which is defined by the following claims.

Claims

CLAIMS 1. A point-of-need, PON, instrument for obtaining a metric for a physiological animal characteristic, the instrument comprising: an insertable rotor cartridge for sample assaying, comprising a plurality of reagent compartments for receiving an animal fluid sample, each reagent compartment comprising a reagent configured to react with an animal fluid sample; a rotor, configured to drive rotation of the rotor cartridge to agitate contents of the plurality of reagent compartments; an analyser, comprising an optical analysis setup and a processor, the analyser being configured to perform optical analysis on the contents of each reagent compartment after the reagent and animal fluid sample have reacted, to produce biomarker data; and a transmitter configured to communicate with a remote biomarker identification system, wherein the analyser is configured to produce the biomarker data having a format allowing the biomarker data to be transmitted by the transmitter to the biomarker identification system for analysis to generate a physiological animal characteristic status.

2. The PON instrument of claim 1, wherein the instrument is for obtaining a metric for a physiological animal characteristic of a domesticated animal and wherein the animal fluid sample is a fluid sample from a domesticated animal.

3. The PON instrument of claim 2, wherein the domesticated animal is one of: a farmed terrestrial animal including a farmed arboreal animal, a farmed aquatic animal or farmed semiaquatic animal, a pet or an animal under scientific study.

4. The PON instrument of claim 3, wherein the domesticated animal is a fish.

5. The PON instrument of any preceding claim, wherein the biomarker data has a format allowing the biomarker data to be processed at a laboratory information system (LIS) of the biomarker identification system.

6. The PON instrument of any preceding claim, wherein the biomarker data has a format meeting the health level 7, HL7, primary standards.

7. The PON instrument of claim 6, wherein the biomarker data has a format meeting the HL7 Fast Healthcare Interoperability Resources, FHIR, standard.

8. The PON instrument of claim 7, wherein the biomarker data has a format meeting the FHIR standard and is encoded as JavaScript Object Notation, JSON, data.

9. A physiological animal characteristic identification system, comprising: the PON instrument of any preceding claim; a biomarker identification system remote from the PON instrument, comprising: a receiver, configured to receive biomarker data from the PON instrument; a memory, comprising a machine-learning model, wherein the model is a classification or regression model configured to generate a physiological animal characteristic status based on the present biomarkers in the biomarker data; and a transmitter, configured to communicate the physiological animal characteristic status to a user computing device; and a user computing device, comprising: a receiver configured to receive a signal from the biomarker identification system; and a display configured to provide a result generated by the biomarker identification system responsive to analysing the received biomarker data to identify the presence of biomarkers in the sample, the result indicating a physiological animal characteristic status.

10. The physiological animal characteristic identification system of claim 9, wherein the model is configured to identify the presence of a biomarker in the sample based on the biomarker data and attribute an animal status indicator to the biomarker data, wherein the model is configured to aggregate two or more animal status indicators to generate an overall physiological animal characteristic status.

11. The physiological animal characteristic identification system of claim 9 or claim 10, wherein the model comprises an XGBoost machine learning algorithm comprising an ensemble of decision trees, wherein each decision tree is configured to receive biomarker data that has been parsed from a first dataformat into a second data format as an input and wherein the ensemble of decision trees are configured to generate a score, wherein the physiological animal characteristic status is based on that score.

12. The physiological animal characteristic identification system of any one of claims 9, 10 or 11, wherein the biomarker identification system further comprises a laboratory information system, LIS, wherein the LIS is configured to parse the received biomarker data from a first data format into a second data format suitable for inputting into the model and provide the parsed biomarker data as an input to the classification model, wherein the LIS is further configured to generate a report of the physiological animal characteristic status for communication to the user computing device.

13. The physiological animal characteristic identification system of claim 12, further comprising at least one more PON instruments according to any preceding PON instrument claim, wherein the LIS is configured to parse received biomarker data from each PON instrument into a format suitable for inputting into the model and communicate parsed biomarker data from each of the PON instruments to the model.

14. The PON instrument of any one of claims 1 to 8 or the physiological animal characteristic identification system of any one of claims 9 to 13, wherein the animal is a fish and wherein the physiological animal characteristic status comprises a viability status for fish to survive in salt water.

15. The PON instrument of any preceding PON instrument claim or the physiological animal characteristic identification system of any preceding physiological animal characteristic identification system claim, wherein the animal fluid sample comprises any of: blood, whole blood, plasma, serum, mucus, faeces and ascites fluid.

16. The physiological animal characteristic identification system of any preceding physiological animal characteristic identification system claim, wherein the model is configured to determine one or more animal status indicators based on the biomarkers in the sample, and wherein the model is configured to generate the physiological animal characteristic status based on the one or more animal statusindicators, wherein the one or more animal status indicators includes at least one of: a health status, a stress status, a gonade status, a metabolic status, and a smoltification status.

17. The PON instrument of any preceding PON instrument claim or the physiological animal characteristic identification system of any preceding physiological animal characteristic identification system claim, wherein one of the animal status indicators is an indicator of chronic stress, wherein the animal is a fish and wherein the biomarkers in the sample comprise phosphorus, sodium, calcium, growth hormone and chloride.

18. A computer-implemented method for determining a physiological animal characteristic status, the method comprising at a biomarker identification system: receiving biomarker data obtained from an animal fluid sample from a PON instrument, wherein the biomarker is in a first data format, and parsing the biomarker data at a server remote from the PON instrument into a second format suitable for inputting into a classification or regression model; inputting the parsed biomarker data into the model at the server remote from the PON instrument, the model being configured to process the parsed biomarker data to obtain a plurality of animal status indicators; aggregating the plurality of animal status indicators into a single physiological animal characteristic status; and outputting, by the model, the physiological animal characteristic status.

19. A non-transitory computer readable medium comprising computer code which, when loaded from memory and executed by one or more processor(s) or processing circuitry, causes a biomarker identification system to perform a method according to claim 18.

Images