GB2573518A

GB2573518A - Internet-based system and method of use thereof

Info

Publication number: GB2573518A
Application number: GB1807452.6A
Authority: GB
Inventors: Philip Gardner Stephen; Lykke Moller Gert; Erik Jensen Claus
Original assignee: Rowanalytics Ltd
Current assignee: Rowanalytics Ltd
Priority date: 2018-05-07
Filing date: 2018-05-07
Publication date: 2019-11-13
Also published as: GB201807452D0

Abstract

An Internet of Things (IoT) based system for providing individualised support to livestock in a farming environment. The system includes a data processor that receives sensor signals from a temporally logged sensor arrangement distributed within the environment, with the processor including a multi-dimensional array model against which sensor signals are compared. The sensor arrangement senses environmental conditions experienced by each animal including a food intake. The array is populated with at least one of drug characteristics, food supplement characteristics, husbandry strategies, disease characteristics of each breed of animal, animal genotype or single nucleotide polymorphism (SNP) data. The data processor further executes a software product which performs a multi-dimensional solution search in the array model, the search being based on logged sensor signals and genotype determination by DNA sequencing of each animal. The software product then computes a welfare trajectory to be used in providing an individualised husbandry for each animal, and output signals are generated by the processor.

Description

INTERNET-BASED SYSTEM AND METHOD OF USE THEREOF

TECHNICAL FIELD

The present disclosure relates generally to Internet®-based systems, namely to systems that provide individualized support to livestock animals in a farming environment. Furthermore, the present disclosure also relates to methods of (for) using aforesaid Internet®-based systems. Moreover, the present disclosure also relates to computer program product comprising a non-transitory computer-readable storage medium having computer-readable instructions stored thereon, the computer-readable instructions being executable by a computerized device comprising processing hardware to execute aforementioned methods.

BACKGROUND

Livestock farming is one of the oldest forms of occupation for people as it is a source of various animal products such as wool, meat, milk, egg and so forth that are needed by mankind on a day-to-day basis. As an example, the animal products may be used by the individuals for personal use as well as a source of income. Generally, the people tend to rear a large number of livestock, in order to produce animal products in abundance.

Often, when managing (namely, providing support to) a large number of livestock in a farming environment, the people employ specialized equipment. Examples of such specialized equipment include, sensors, processors, actuators, and so forth. As an example, a livestock farmer may deploy 100 sensors (such as temperature and humidity sensors) in the farming environment to monitor environmental conditions experienced by his/her livestock.

However, there exist limitations associated with use of the specialized equipment within the farming environment. Notably, in a cost-sensitive industry such as livestock farming, deployment of a large number of customized sensors within the farming environment is suboptimal in terms of cost. Generally, a lifetime of such customized sensors is insufficient to obtain a sufficient economy of cost, from a perspective of the livestock farmer. Furthermore, there exist difficulties associated with managing sensed data from such sensors. Notably, the existing specialized equipment is limited in its ability to communicate and process the sensed data for facilitating substantial improvement in managing livestock. As an example, processing of the sensed data is often very complex and time consuming. Eventually, a quality of management of the livestock deteriorates (or is sub-optimal), which is undesirable.

Therefore, in light of the foregoing discussion, there exists a need to address, for example to overcome, the aforementioned drawbacks associated with the existing equipment used to provide support to livestock in a farming environment.

SUMMARY

The present disclosure seeks to provide an Internet® (IoT')-based system that provides individualized support to livestock animals in a farming environment.

The present disclosure also seeks to provide a method of (for) using an Internet® (IoT')-based system that provides individualized support to livestock animals in a farming environment.

According to a first aspect, an embodiment of the present disclosure provides an Internet® (IoT')-based system that provides individualized support to livestock animals in a farming environment, wherein the Internet®-based system includes a data processing arrangement that receives in operation temporally logged sensor signals from a sensor arrangement that is spatially distributed within the farming environment, wherein the data processing arrangement includes an array model of the data processing arrangement against which the temporally logged sensor signals are compared, wherein the data processing arrangement provides output signals that control operation of the farming environment, and wherein the data processing arrangement executes a software product that in execution analyzes the temporally logged sensor signals in respect of the array model and generates the output signals, characterized in that:

(a) the software product is configured to perform multi-dimensional inferences in the array model implemented as a multi-dimensional array model, the multi-dimensional inferences being based on at least a subset of the temporally logged sensor signals and a genotype determination by DNA sequencing of each animal hosted within the farming environment;

(b) the sensor arrangement senses in operation, environmental conditions experienced by each animal, including monitoring a food intake for each animal;

(c) the multi-dimensional array model is populated with at least one of drug characteristics, food supplement characteristics, husbandry strategies depending on animal health complications, disease characteristics of each breed (genotype) of animal, animal genotype data, and Single Nucleotide Polymorphism (SNP) data; and (d) the software product is used to compute a welfare trajectory to be used in providing individualized husbandry of each animal.

The invention is of advantage in that the multi-dimensional array model, likewise the multi-dimensional inferences, are capable of providing an improved, for example more optimal, animal welfare.

By multi-dimensional is meant 3-dimensional or greater, more optionally, 4-dimensional or greater, yet more optionally 5-dimensional or greater, yet more optionally 10-dimensional or greater, yet more optionally 20-dimensional or greater, and even yet more optionally 50dimensional or greater (if a size of data set of sensor signals generated by the sensor arrangement allows). Embodiments of the invention also optionally employ a multi-dimension combinatorial grouping of an order 3 or greater, more optionally of an order 4 or greater, yet more optionally of an order 5 or greater, yet more optionally of an order 10 or greater, yet more optionally of an order 20 or greater, and even yet more optionally of an order 50 or greater (if a size of data set of sensor signals generated by the sensor arrangement allows).

The aforesaid Internet®-based system is easy to implement, costeffective, and user friendly. Notably, such an Internet®-based system can be implemented by modest, off-the-shelf hardware, in an efficient manner when configured pursuant to the present disclosure. The Internet®-based system is accurate and solves complex computational problems time-efficiently.

Optionally, the SNP data includes single nucleotide polymorphisms characterizing each animal determined by using Polymerase Chain Reaction (PCR) to read genetic tissue samples derived for each animal.

Optionally, the sensor arrangement includes a plurality of sensors that are spatially distributed within the farming environment and are coupled in communication with the data processing arrangement by using a wireless dynamically reconfigurable communication network.

Optionally, the sensor arrangement includes tags that are attached to the livestock animals and that monitor and log environmental conditions experienced by each animal and movement performed by each animal.

Optionally, the wireless dynamically reconfigurable communication network is implemented as a peer-to-peer network.

Optionally, the Internet-based system collects in operation one or more pathogens present in the farming environment, genotype sequences the one or more pathogens to characterize the one or more pathogen, and employs the characterization of the one or more pathogens as an input parameter to the software product executed in the data processing arrangement to use in performing itsinference.

Optionally, the output signals are used to control at least one of a type and/or a quantity of feed provided to the animals; a time when feed is provided to the animals; additional feed supplements and/or one or more drugs to be administered to the animals; selective heating or cooling to be provided to the animals; and pathogen reducing processes to be applied to the farming environment.

Optionally, the Internet®-based system includes an avatar testing arrangement including a test animal into which genotype information from a given animal is imported, and combinations of drugs and/or feed supplements are tested to determine whether or not toxicological problems are likely to arise if a given combination of drugs and/or feed supplements are administered to the given animal.

Optionally, the avatar testing arrangement uses the test animal implemented as a fruit fly (for example, Drosophilia, but not limited thereto).

Optionally, the sensor arrangement includes a processing module that pre-processes the temporally logged sensor signals, using at least one artificial intelligence algorithm.

In a second aspect, an embodiment of the present disclosure provides a method of (for) using an Internet® (IoT)-based system that provides individualized support to livestock animals in a farming environment, wherein the Internet-based system includes a data processing arrangement that receives in operation temporally logged sensor signals from a sensor arrangement that is spatially distributed within the farming environment, wherein the data processing arrangement includes an array model of the data processing arrangement against which the temporally logged sensor signals are compared, wherein the data processing arrangement provides output signals that control operation of the farming environment, and wherein the data processing arrangement executes a software product that in execution analyzes the temporally logged sensor signals in respect of the array model and generates the output signals, characterized in that the method includes:

(a) arranging for the software product to perform multi-dimensional inferences from the array model implemented as a multi-dimensional array model, the multi-dimensional inferences being based on at least a subset of the temporally logged sensor signals and a genotype determination by DNA sequencing of each animal hosted within the farming environment;

(b) sensing in operation using the sensor arrangement, environmental conditions experienced by each animal, including monitoring a food intake for each animal;

(c) populating the multi-dimensional array model with at least one of drug characteristics, food supplement characteristics, husbandry strategies depending on animal health complications, disease characteristics of each breed (genotype) of animal, animal genotype data, and Single Nucleotide Polymorphism (SNP) data; and (d) using the software product to compute a welfare trajectory to be used in providing individualized husbandry of each animal.

By multi-dimensional is meant 3-dimensional or greater, more optionally, 4-dimensional or greater, yet more optionally, 5-dimensional or greater, yet more optionally 10-dimensional or greater, yet more optionally 20-dimensional or greater, and even yet more optionally 50dimensional or greater (if a size of data set of sensor signals generated by the sensor arrangement allows). Embodiments of the invention also optionally employ a multi-dimension combinatorial grouping of an order 3 or greater, more optionally of an order 4 or greater, yet more optionally of an order 5 or greater, yet more optionally of an order 10 or greater, yet more optionally of an order 20 or greater, and even yet more optionally of an order 50 or greater (if a size of data set of sensor signals generated by the sensor arrangement allows).

Optionally, the method includes arranging for the SNP data to include single nucleotide polymorphisms characterizing each animal, determined by using Polymerase Chain Reaction (PCR) to read genetic tissue samples derived for each animal.

Optionally, the method includes arranging for the sensor arrangement to include a plurality of sensors that are spatially distributed within the farming environment and are coupled in communication with the data processing arrangement by using a wireless dynamically reconfigurable communication network.

Optionally, the method includes arranging for the sensor arrangement to include tags that are attached to the livestock animals and that monitor and log environmental conditions experienced by each animal and movement performed by each animal.

Optionally, the method includes arranging for the wireless dynamically reconfigurable communication network to be implemented as a peer-topeer network.

Optionally, the method includes arranging for the Internet-based system to collect in operation one or more pathogens present in the farming environment, perform genotype sequencing of the one or more pathogens to characterize the one or more pathogens, and employ the characterization of the one or more pathogens as an input parameter to the software product executed in the data processing arrangement to use in performing its search.

Optionally, the method includes using the output signals to control at least one of a type and/or a quantity of feed provided to the animals; a time when feed is provided to the animals; additional feed supplements and/or one or more drugs to be administered to the animals; selective heating or cooling to be provided to the animals; and pathogen reducing processes to be applied to the farming environment.

Optionally, the method includes arranging for the Internet-based system to include an avatar testing arrangement including a test animal into which genotype information from a given animal is imported, and combinations of drugs and/or feed supplements are tested to determine whether or not toxicological problems are likely to arise if a given combination of drugs and/or feed supplements are administered to the given animal.

Optionally, the method includes arranging for the avatar testing arrangement to use the test animal implemented as a fruit fly.

Optionally, the method includes arranging for the sensor arrangement to include a processing module that pre-processes the temporally logged sensor signals, using at least one artificial intelligence algorithm. An artificial intelligence algorithm is capable of changing its parameters dynamically in operation by way of adaptive learning when confronted with data; such an algorithm, for example, implemented or mimic a hierarchical arrangement of pseudo-analog variable state machines, whose analog states are changed depending upon learning from being confronted with data.

In a third aspect, embodiments of the present disclosure provide a software product recorded on machine-readable non-transitory (nontransient) data storage media, wherein the software product is executable upon computing hardware for implementing the aforementioned method; in other words, the present disclosure provides a computer program product comprising a non-transitory computerreadable storage medium having computer-readable instructions stored thereon, the computer-readable instructions being executable by a computerized device comprising processing hardware to execute the aforesaid method.

Additional aspects, advantages, features and objects of the present disclosure would be made apparent from the drawings and the detailed description of the illustrative embodiments construed in conjunction with the appended claims that follow.

It will be appreciated that features of the present disclosure are susceptible to being combined in various combinations without departing from the scope of the present disclosure as defined by the appended claims.

DESCRIPTION OF THE DRAWINGS

The summary above, as well as the following detailed description of illustrative embodiments, is better understood when read in conjunction with the appended drawings. For the purpose of illustrating the present disclosure, exemplary constructions of the disclosure are shown in the drawings. However, the present disclosure is not limited to specific methods and instrumentalities disclosed herein. Moreover, those in the art will understand that the drawings are not to scale. Wherever possible, like elements have been indicated by identical numbers.

Embodiments of the present disclosure will now be described, by way of example only, with reference to the following diagrams wherein:

FIG. 1 is a schematic illustration of an Internet®-based system that provides individualized support to livestock animals in a farming environment, in accordance with an embodiment of the present disclosure;

FIG. 2 is a schematic illustration of a farming environment having a sensor arrangement spatially distributed therein, in accordance with an embodiment of the present disclosure;

FIG. 3 is a process diagram illustrating steps that are implemented by the Internet-based system for enabling a software product to perform inferences from a multi-dimensional array model, in accordance with an embodiment of the present disclosure;

FIG. 4 is a process diagram illustrating steps that are implemented by an avatar testing arrangement, in accordance with an embodiment of the present disclosure; and

FIG. 5 illustrates steps of a method of (for) using an Internet-based system that provides individualized support to livestock animals in a farming environment, in accordance with an embodiment of the present disclosure.

In the accompanying drawings, an underlined number is employed to represent an item over which the underlined number is positioned or an item to which the underlined number is adjacent. A non-underlined number relates to an item identified by a line linking the non-underlined number to the item. When a number is non-underlined and accompanied by an associated arrow, the non-underlined number is used to identify a general item at which the arrow is pointing.

DESCRIPTION OF EMBODIMENTS

In overview, embodiments of the present disclosure are concerned with an Internet® (IoT)-based system that provides individualized support to livestock animals in a farming environment, and a method of operating an Internet® (IoT)-based system.

Referring to FIG. 1, illustrated is a schematic illustration of an Internet®based system 100 that provides individualized support to livestock animals in a farming environment, in accordance with an embodiment of the present disclosure. The Internet®-based system 100 includes a data processing arrangement 102 and a sensor arrangement 104. Notably, the data processing arrangement 102 includes a multi-dimensional 106 of the data processing arrangement 102; by multi-dimensional is meant 3-dimensional or greater, more optionally, 4-dimensional or greater, yet more optionally, 5-dimensional or greater, yet more optionally 10-dimensional or greater, yet more optionally 20-dimensional or greater, and even yet more optionally 50-dimensional or greater (if a size of data set of sensor signals generated by the sensor arrangement allows). Embodiments of the invention also optionally employ a multidimension combinatorial grouping of an order 3 or greater, more optionally of an order 4 or greater, yet more optionally of an order 5 or greater, yet more optionally of an order 10 or greater, yet more optionally of an order 20 or greater, and even yet more optionally of an order 50 or greater (if a size of data set of sensor signals generated by the sensor arrangement allows).

The data processing arrangement 102 receives in operation, temporally logged sensor signals from the sensor arrangement 104 that is spatially distributed within the farming environment (shown in FIG. 2). The temporally logged sensor signals are compared against the multidimensional array model 106 of the data processing arrangement 102. Furthermore, the data processing arrangement 102 provides output signals that control operation of the farming environment. The data processing arrangement 102 executes a software product that in execution analyzes the temporally logged sensor signals in respect of the multi-dimensional array model 106 and generates the output signals.

In the Internet®-based system 100, (a) the software product is configured to perform inferences from the multi-dimensional array model 106, the inferences being based on at least a subset of the temporally logged sensor signals and a genotype determination by DNA sequencing of each animal hosted within the farming environment;

(b) the sensor arrangement 104 senses in operation, environmental conditions experienced by each animal, including monitoring a food intake for each animal;

(c) the mutli-dimensional array model 106 is populated with at least one of drug characteristics, food supplement characteristics, husbandry strategies depending on animal health complications, disease characteristics of each breed (genotype) of animal, animal genotype data, and SNP data; and (d) the software product is used to compute a welfare trajectory to be used in providing individualized husbandry of each animal.

Notably, (a) will be explained in more detail hereinafter, in conjunction with FIG. 3; (b) will be explained in more detail hereinafter, in conjunction with FIG. 2; (c) will be explained in more detail hereinafter, in conjunction with description for array model', and (d) will be explained in more detail hereinafter, following the description for FIG. 3.

Throughout the present disclosure, the term Internet®-based system relates to an arrangement of Internet®-compatible devices (namely, devices that are communicably coupled to other devices, via the Internet®) that can be employed to provide individualized support to the livestock animals in the farming environment. Notably, such Internet®compatible devices relate to physical devices (for example, such as cameras, beacons, appliances, sensors, processing machines, and so forth) having at least a transceiver module and a wireless communication interface that allows for the physical devices to connect to the Internet®, and exchange information via the Internet®. As a result, when the Internet®-compatible devices are deployed in the farming environment, they form the Internet®-based system 100 that provides individualized support to the livestock animals in the farming environment. Therefore, the data processing arrangement 102 and the sensor arrangement 104 of the Internet®-based system 100 are arrangements of the aforesaid Internet-compatible devices. Furthermore, the Internet®-compatible devices may also be referred to as Internet®-compatible hardware.

It will be appreciated that the aforesaid Internet®-compatible devices of the Internet®-based system 100 can be implemented by way of readily available off the shelf' hardware, such as Internet®-compatible tags, Internet®-compatible beacons, Internet®-compatible sensors, Internet®-compatible cameras, Internet®-compatible processing devices, and the like. Such Internet®-compatible devices are often mass-produced, and can therefore be employed within the farming environment to implement a cost-effective Internet®-based system 100. Notably, since agriculture industry is highly cost-sensitive, such an Internet®-based system 100 is a highly attractive provision for providing individualized support to livestock animals in the farming environment.

Alternatively, the Internet®-compatible devices of the Internet-based system 100 could also be implemented by way of custom hardware.

It will be appreciated that the Internet-based system 100 may also be referred to as an Internet-of-Things based system.

Throughout the present disclosure, the term data processing arrangement relates to an arrangement of Internet-compatible devices having data processing capabilities. Notably, the data processing arrangement 102 constitutes a powerful computing engine that facilitates in performing data processing for provision of individualized support to livestock animals in the farming environment.

Furthermore, the multi-dimensional array model 106 of the data processing arrangement 102 generally relates to hardware, software, firmware, or a combination of these for storing information in an organized (namely, structured) manner, thereby, allowing for easy storage, access (namely, retrieval), updating and analysis of such information. The term array mode! also encompasses array model servers that provide the aforesaid array model services to the data processing arrangement 102. It will be appreciated that the array model 106 serves as a data repository of the data processing arrangement 102.

As mentioned previously in (c), the array model 106 is populated with at least one of: drug characteristics, food supplement characteristics, husbandry strategies depending on animal health complications, disease characteristics of each breed (genotype) of animal, animal genotype data, and Single Nucleotide Polymorphism (SNP) data. Furthermore, optionally, the array model 106 is populated with at least one of: genotype information of animals, recommended diet of animals, recommended medication for animals, recommended sleeping pattern of animals, medication list, and food supplement list. Notably, the array model 106 is populated with comprehensive information pertaining to health of animals, and such comprehensive information is utilized by the Internet-based system 100 to provide precision support for maintaining good health of the livestock animals.

Optionally, the SNP data includes single nucleotide polymorphisms characterizing each animal determined by using Polymerase Chain Reaction (PCR) to read genetic tissue samples derived for each animal. The term single nucleotide polymorphisms relates to genetic variations among a given species of organisms. Notably, such single nucleotide polymorphisms are identified upon analysis of DNA sequences of members of the given species, wherein each single nucleotide polymorphism represents a variation in a single nucleotide within DNA sequences of different members of the given species. It will be appreciated that the genetic tissue samples derived for each animal allow for obtaining DNA sequences for each animal, wherefrom, the single nucleotide polymorphisms characterizing each animal are determined by using Polymerase Chain Reaction (PCR). Furthermore, Polymerase Chain Reaction (PCR) is a molecular biology technique that allows for amplification of a segment of a given DNA sequence across several orders of magnitude whilst making multiple copies of the segment of the given DNA sequence.

As an example, the array model 106 may be populated with husbandry strategies depending on animal health complications and the SNP data of livestock animals. In such an example, the husbandry strategies may include (i) administering a given diet to an animal suffering from intestinal instabilities (namely, tummy ache), (ii) providing selective heating to animals having a metabolic rate lying within a given range, (iii) providing preventive medication to animals having a history of medical illnesses, (iv) fumigating animal stalls and indoor accommodation with ozone gas in an event of fungal or mould growth to kill microbes without leaving any environmentally damaging residues, and moving the livestock animals to an open field environment whilst implementing such fumigation operation, and the like.

Optionally, the array model 106 of the data processing arrangement 102 includes a log of expected sensor signals (namely, forecasted sensor signals) from the farming environment against which the temporally logged sensor signals are compared. In such a case, the aforesaid comparison allows for identifying inconsistencies between the expected sensor signals and the temporally logged sensor signals from the sensor arrangement 104, thereby, allowing for the Internet®-based system 100 to appropriately generate the output signals for providing individualized support to the livestock animals.

Optionally, the array model 106 is periodically updated and populated with latest information pertaining to provision of individualized support to the livestock animals. Such latest information may be obtained by the array model 106 from at least one external server (not shown), learnings derived from historical welfare trajectories that have been previously computed upon execution of the software product by the data processing arrangement 102, and the like.

Throughout the present disclosure, the term sensor arrangement relates to an arrangement of Internet-compatible devices having sensing capabilities. The sensor arrangement 104 senses in operation, environmental conditions experienced by each animal, including monitoring a food intake for each animal. Examples of the environmental conditions include, but are not limited to, temperature within the farming environment, humidity within the farming environment, sunlight exposure within the farming environment, air quality within the farming environment, and chemicals within the farming environment.

Furthermore, optionally, the sensor arrangement 104 senses in operation, attributes and/or activities of the livestock animals within the farming environment. Examples of the attributes and/or activities of the livestock animals include, but are not limited to, appearance of the livestock animals, movement of the livestock animals, sleeping pattern of the livestock animals, posture of the livestock animals, breathing pattern of the livestock animals, and food intake of the livestock animals. Referring to FIG. 2, illustrated is a schematic illustration of a farming environment 200 having the sensor arrangement 104 spatially distributed therein, in accordance with an embodiment of the present disclosure. In operation, the sensor arrangement 104 temporally logs sensor signals, and transmits the temporally logged sensor signals to the data processing arrangement 102. Specifically, such a temporal log of sensor signal describes sensor signals received from within the farming environment 200, as a function of time. Beneficially, such temporal logs of sensor signals allow for the Internet®-based system 100 to temporally monitor the farming environment 200 in a manner that the Internet®based system 100 is able to efficiently provide individualized support to livestock animals in the farming environment 200.

Optionally, the sensor arrangement 104 includes a plurality of sensors 202-210 that are spatially distributed within the farming environment 200 and are coupled in communication with the data processing arrangement 102 by using a wireless dynamically reconfigurable communication network. Notably, the plurality of sensors 202-210 are arranged at different spatial locations (or spatially distributed) within an entire region of the farming environment 200. The plurality of sensors 202-210 could include sensors, such as, but not limited to, image sensors, movement sensors, orientation sensors, temperature sensors, humidity sensors, soil moisture sensors, sunlight exposure sensors, food intake monitoring sensors (for example, such as weight sensors in animal feeding trays), gas sensors, air pollution sensors, air flow sensors, and chemical sensors. Furthermore, the aforesaid wireless dynamically reconfigurable communication network allows for the plurality of sensors 202-210 to exchange information (for example, such as the temporally logged sensor signals) with the data processing arrangement 102 in an efficient manner whilst being arranged within the farming environment 200. Specifically, the dynamic reconfigurability of such a network allows for the Internet®-based system 100 to conveniently employ and adapt available proprietary Internet-compatible sensors, as per requirement.

Optionally, the wireless dynamically reconfigurable communication network is implemented as a peer-to-peer network. The peer-to-peer (P2P) network implementation of the wireless dynamically reconfigurable communication network allows for deploying the plurality of sensors 202210 within the farming environment 200 at very modest infrastructure costs. Furthermore, optionally, such plurality of sensors 202-210 can be recycled from one generation of animals to a next generation of animals (for example, by reusing the plurality of sensors 202-210). In an example, the wireless dynamically reconfigurable communication network may be implemented as a peer-to-peer network that is dynamically reconfigured by using a periodic calibration routine based on signal strength to find principal Eigenvector routes of communication between the plurality of sensors 202-210 and the data processing arrangement 102. Computation of Eigenvectors for communication network is described, for example, in a granted European patent EP1700421B1 (Canright, Telenor AS).

Furthermore, optionally, each of the plurality of sensors 202-210 includes a transceiver module (not shown) and a wireless communication interface (not shown) that enables the sensor to be directly or indirectly coupled in communication with the data processing arrangement 102. In one embodiment, each of the plurality of sensors 202-210 are Internet®-compatible (namely, communicably coupled to other devices such as the data processing arrangement 102, other sensors, and the like, via the Internet®). In such a case, each of the plurality of sensors 202-210 are directly coupled in communication with the data processing arrangement 102. In another embodiment, at least one of the plurality of sensors 202-210 is Internet-compatible and the remaining sensors of the plurality of sensors 202-210 are coupled in communication with the at least one Internet®-compatible sensor. In such a case, the at least one Internet®-compatible sensor is directly coupled in communication with the data processing arrangement 102 whereas the remaining sensors of the plurality of sensors 202-210 are indirectly coupled in communication with the data processing arrangement 102, via the at least one Internet®-compatible sensor.

Optionally, operational power for the plurality of sensors 202-210 of the sensor arrangement 104 is provided by way of at least one of: a nonrechargeable charging arrangement, a rechargeable charging arrangement. In an example, the operational power for the plurality of sensors 202-210 of the sensor arrangement 104 may be provided by way of solar panel rechargeable batteries. In another example, the operational power for the plurality of sensors 202-210 of the sensor arrangement 104 may be provided by way of a wireless resonant inductive charging arrangement implemented within the farming environment 200. In yet another example, the operational power for the plurality of sensors 202-210 of the sensor arrangement 104 may be provided by way of non-rechargeable batteries.

Optionally, the sensor arrangement 104 includes tags 212 and 214 that are attached to the livestock animals and that monitor and log environmental conditions experienced by each animal and movement performed by each animal. Optionally, in this regard, the tags 212-214 include sensors (for example, such as movement sensors, orientation, sensors sunlight exposure sensors, and the like) to monitor and log the environmental conditions experienced by each animal and the movement performed by each animal. In other words, the tags 212-214 that are attached to the livestock animals are sensor-equipped. For sake of simplicity, the tags 212-214 can also be understood to be a constituent of a plurality of sensors of the sensor arrangement 104. In such a case, the tags 212-214 can be understood to be sensors 212 and 214 of a plurality of sensors 202-214 of the sensor arrangement 104.

Optionally, each of the tags 212-214 further include a memory module whereat the environmental conditions experienced by each animal and the movement performed by each animal can be logged.

Optionally, operational power for the tags 212-214 of the sensor arrangement 104 is provided by way of at least one of: a nonrechargeable charging arrangement, a rechargeable charging arrangement. In an example, the operational power for the tags 212214 of the sensor arrangement 104 may be provided by way of the wireless resonant inductive charging arrangement implemented within the farming environment 200.

Optionally, the sensor arrangement 104 is operated in intermittent mode, to allow for optimal utilization of the operational power by the sensor arrangement 104. More optionally, the tags 212-214 that are attached to the livestock animals are operated in the intermittent mode. As an example, the tags 212-214 may be operated in intermittent mode (namely, in a discontinuous or irregular manner), based upon a temporal density of activity of the livestock animals upon which the tags 212-214 are attached.

In the exemplary implementation of FIG. 2, in the sensor arrangement 104, the plurality of sensors 202-210 may include image sensors 202 and 204 arranged on pillars within the farming environment 200, a gas sensor 206 for sensing presence of gases such as carbon dioxide, hydrogen sulphide and methane within the farming environment 200, a temperature sensor 208 and a humidity sensor 210. Furthermore, in the sensor arrangement 104, the tags 212-214 that are attached to the livestock animals may include motion sensors and sunlight exposure sensors implemented thereon. Optionally, the pillars upon which the image sensors 202 and 204 are arranged, may include the wireless resonant inductive charging arrangement for providing operational power for the tags 212-214 of the sensor arrangement 104.

Optionally, the temperature sensors (such as the temperature sensor 208 of FIG. 2) are implemented as thermistor or integrated-circuit solidstate temperature sensors. In an example, the temperature sensors may be housed in animal feeding troughs, water troughs, barriers between animal pens, doors and gates of the animal pens, roofs/ceilings within the farming environment 200, the tags 212-214 that are attached to the livestock animals, and the like.

Optionally, the humidity sensors (such as the humidity sensor 210 of FIG. 2) are implemented as thin-film polyamide sensors.

Optionally, the air flow sensors are implemented as heated wire pair transducers or thermistor pair transducers.

Optionally, the sensor arrangement 104 includes a processing module that pre-processes the temporally logged sensor signals, using at least one artificial intelligence algorithm. Notably, such pre-processing utilizes hierarchical configurations of analog variable state machines implemented using micropower RISC processors. Such artificial intelligence algorithms are well known in the art. It will be appreciated that such pre-processing of the temporally logged sensor signals allows for filtering out unwanted data (sensor signals) whilst retaining seemingly relevant data (sensor signals) for subsequent processing by the data processing arrangement 102. In other words, the aforesaid pre processing operation allows for reducing an amount of data that is to be communicated by the sensor arrangement 104 to the data processing arrangement 102, into simpler representative sensor signals. As a result, unnecessary data transmission within the Internet-based system 100 is reduced. Consequently, the software product executed by the data processing arrangement 102 can implement processing steps for analysis of the simpler representative sensor signals (which are a subset of the temporally logged sensor signals) in respect of the array model 106 to generate the output signals with minimum time wastage (or maximum time utilization).

As an example, such pre-processing may be employed for the image sensors 202 and 204 within the farming environment 200, to substantially reduce an amount of video data captured by the image sensors 202 and 204.

The software product executed by the data processing arrangement 102 is configured to perform inferences from the multi-dimensional array model 106, the inferences being based on at least a subset of the temporally logged sensor signals and a genotype determination by DNA sequencing of each animal hosted within the farming environment 200. Notably, information stored in the array model 106 acts as an addressable solution space that substantially represents all valid solutions that satisfy all constraints of the Internet-based system 100. In other words, the array model 106 includes valid Cartesian sub-spaces of states or combinations satisfy a conjunction of all Internet®-based system 100 constraints for all interconnected variables such as the subset of the temporally logged sensor signals and the genotype determination by DNA sequencing of each animal hosted within the farming environment 200.

Referring to FIG. 3, illustrated is a process diagram 300 illustrating steps that are implemented by the Internet®-based system 100 for enabling the software product to perform the inferences from the multi dimensional array model 106, in accordance with an embodiment of the present disclosure. Optionally, the Internet®-based system 100 collects in operation one or more pathogens present in the farming environment 200, genotype sequences the one or more pathogens to characterize the one or more pathogen, and employs the characterization of the one or more pathogens as an input parameter to the software product executed in the data processing arrangement 102 to use in performing its search. Notably, in the process diagram 300, at a step 302, the Internet®-based system 100 collects the one or more pathogens present in the farming environment 200, at a step 304, the Internet®-based system 100 genotype sequences the one or more pathogens to characterize the one or more pathogen, and at a step 306, the Internet-based system 100 employs the characterization of the one or more pathogens as an input parameter to the software product executed in the data processing arrangement 102 to use in performing its inferences. Optionally, at the step 302, the sensor arrangement 104 in operation, collects the one or more pathogens present in the farming environment 200. Additionally or alternatively, optionally, at the step 302, the one or more pathogens present in the farming environment 200 are collected manually by a person. Optionally, at the step 304, the software product executed by the data processing arrangement 102 genotype sequences the one or more pathogens to characterize the one or more pathogen. In such a case, the aforesaid genotype sequencing operation allows for the Internet-based system 100 to identify characteristics of the one or more pathogens such as, breeding conditions of the one or more pathogens, lifespan of the one or more pathogens, and the like. Optionally, at the step 306, the software product employs the input parameter to substantially reduce computational complexity of the inferences by way of performing the inferences only on a portion of the addressable solution space that substantially relates to the input parameter.

Optionally, at the step 306, the Internet®-based system 100 employs a plurality of input parameters to be provided to the software product for use in performing its search. It will be appreciated that employing different input parameters allow for the Internet®-based system 100 to determine how various input parameters affect an output as represented in the addressable solution space.

Optionally, the input parameter includes features of one or more SNPs obtained from performing DNA base readout from biological samples of the one or more pathogens. Optionally, in this regard, the DNA base readout from biological samples of the one or more pathogens is performed as part of a GWAS (Genome-Wide Association Study). Furthermore, large SNP-genotyping arrays for tens of thousands of pathogens (cases and controls) and hundreds of thousands of SNPs, such as 600,000 pathogens and 2.5 million SNPs, are available in multiplexed formats for huge throughput. In such a case, features of the one or more SNPs obtained as part of the GWAS are stored in the array model 106. It will be appreciated that the GWAS datasets can be generated using well-known techniques, including but not limited to, SNP genotyping using SNP microarrays, exome sequencing, genome sequencing and so forth. The input parameter obtained from performing the DNA base readout of the biological samples of the one or more pathogens comprise information about SNP genotypes associated with the one or more pathogens.

The software product executed by the data processing arrangement 102 is used to compute a welfare trajectory to be used in providing individualized husbandry of each animal. Specifically, the software product executed by the data processing arrangement 102 computes the welfare trajectory after performing the inferences from the multidimensional array model 106, as described hereinabove. Notably, a welfare trajectory of a given animal includes at least one constraint and/or at least one environmental condition pertaining to the farming environment 200, wherein the at least one constraint and/or the at least one environmental condition is favourable for the given animal. The welfare trajectory is created based upon the information stored in the array model 106 and results of the inferences in the multi-dimensional array model 106. Optionally, the welfare trajectory is also based upon the input parameter (namely, the characterization of the one or more pathogens collected from the farming environment 200). It will be appreciated that the information stored in the array model 106, the results of the inferences from the multi-dimensional array model 106, and optionally, the input parameter, constitute a dataset upon which the software product implements processing operations, to compute the welfare trajectory to be used in providing individualized husbandry of each animal.

Optionally, to compute welfare trajectories for the livestock animals, the software product is operated to:

(I) perform pre-filtering of the dataset, to reduce a number of inputs that are considered for computing the welfare trajectories;

(II) perform mining within the pre-filtered dataset to find distinct ncombinations of SNP genotypes and/or other types of features found in the livestock animals;

(ill) perform (ii) repeatedly for mining a plurality of random permutations of properties using a same set of mining parameters;

(iv) find networks of distinct n-combinations sharing one or more properties;

(v) find networks using the same set of parameters from among the distinct n-combinations and from among the plurality of random permutations, compare null hypothesis and determine one or more pvalues with FDR. correction to eliminate random observations;

(vi) perform annotation of the networks with a semantically normalised knowledge graph containing information about the shared one or more properties; and (vii) perform re-clustering of the networks, after correlating the networks found at (v) with the semantically normalized knowledge graph of (vi) containing information about the shared one or more properties.

Optionally, at (I), the data processing arrangement 102 is operable to perform the pre-filtering of the datasets via at least one of: removal of SNPs where a minor allele frequency associated therewith is below a threshold at which it can satisfy a >MinCases criterion for a population of livestock animals; removing SNPs which are approximately co-located and within linkage disequilibrium regions using linkage disequilibrium based clumping; removing SNPs where a major:minor allele distribution is close to 50:50; and including or selecting SNPs that are relevant to a hypothesis or other analytical strategy. In one example, the pre-filtering of the specific datasets includes removing SNPs where a minor allele frequency associated therewith is below a threshold at which it can satisfy a >MinCases criterion for a population. The MinCases criterion for the population is a numerical value that denotes a minimum number of cases within the population that satisfy a requirement, such as, a minimum number of cases that have a specific SNP. Such a MinCases criterion can be automatically specified, such as, by the data processing arrangement 102. Alternatively, the MinCases criterion can be manually specified by a user of the Internet-based system 100. In one example, the prefiltering of the specific datasets includes removing SNPs which are approximately co-located and within linkage disequilibrium regions, using linkage disequilibrium based clumping. In another example, the prefiltering of the specific datasets includes removing SNPs where a major:minor allele distribution is close to 50:50, such as, where the major:minor allele distribution is 52:48.

Optionally, at (ii), the data processing arrangement 102 performs in operation mining within the pre-filtered dataset by finding combinations of SNP genotypes which occur in a plurality of cases in the plurality of animals in the farming environment 200 (>MinCases) or in zero or just a few controls (<MaxControls), analyzing in ascending levels of order, in an n number of layers, one layer at a time and storing the n-combinations as N-states in an output data structure. The MaxControls criterion for the population is a numerical value that denotes a maximum number of controls within the population that satisfy a requirement, such as, a maximum number of cases that have a specific SNP. Such MaxControls criterion can be either specified automatically, such as by the data processing arrangement 102, or manually by a user of the Internetbased system 100. The data processing arrangement 102 is operable to perform mining to determine all such SNPs that are associated with the >MinCases and the <MaxControls criterions respectively, such as SNPs that occur with a maximum number of cases and a minimum number of controls. Optionally, during mining of the input parameter and data by the data processing arrangement 102, a new data-type such as epigenetics data (a series of new elements) can be incorporated as features into the input parameter in addition to the SNP genotypes. Furthermore, the data processing arrangement 102 is operable to perform the mining in ascending levels of order, in an n number of layers, one layer at a time and storing the n-combinations in an output data structure. For example, the data processing arrangement 102 analyses SNPs at a layer 1, wherein SNPs associated with all cases and controls within the population are analyzed. In an example embodiment, the data processing arrangement 102 terminates the determination of combinations of n-SNPs (or n-combinations) in successive layers. In another example embodiment, the data processing arrangement 102 is operable to determine n-combinations in 20 or more layers.

Optionally, upon successful determination of the n-combinations of SNP genotypes and/or other types of features found in the livestock animals in one layer that satisfy the >MinCases and the <MaxControls criterions, the MinCases is incremented by the data processing arrangement 102 for analysis of a successive layer. Such incrementing of the MinCases criterion upon successful determination of n-combinations of SNPs in previous layers, enables the data processing arrangement 102 to determine largest possible subgroups of cases that are associated with the n-combination of SNPs within the population. In an example, the data processing arrangement 102 comprises at least one multicore GPU. The data processing arrangement 102 is operable to employ the GPU and/or the FPGA for the determination of the n-combination of SNPs in the n number of layers. More optionally, the Graphics Processing Unit (GPU) and/or the Field Programmable Gate Array comprise a memory associated therewith. In one example, the memory is implemented as a random access memory (RAM).

Optionally, the data processing arrangement 102 is operable to store the n-combinations of SNPs determined in each layer, as well as individual identifiers for the cases. For example, the data processing arrangement 102 is operable to assign a binary vector (indicated by BV hereinafter) to each case, subsequent to determination of the n-combinations of SNPs within a layer. The binary vector can take a value of '0' that indicates a specific case not being associated with the n-combination of SNPs for the layer, and a value of '1' that indicates the specific case being associated with a specific n-combination of SNPs. The BV is updated by the data processing arrangement 102 for each specific n-combination in each layer and is employed for determination of n-combinations of SNPs for subsequent layers (whereas the individual identifiers for each of the livestock animals in the farming environment 200 are input to the GPU of the data processing arrangement 102 prior to initiation of operation thereof). Moreover, the data processing arrangement 102 is operable to store the n-combinations of SNPs determined in each layer and the BV values associated with the cases, such as, within the memory (such as a random access memory or RAM) associated with the GPU. For example, the data processing arrangement 102 is operable to store the ncombinations of SNPs and the BV values associated with the cases in the output, such as an output represented by N-states.

Optionally, at (ill), the data processing arrangement 102 is operable to perform execution of permutations or repeating mining for the plurality of random permutations of the properties (for example, such as genotype sequences and pathogens) using the same set of mining parameters. The data processing arrangement 102 is operable to repeat the mining (as explained in detail hereinabove) a predefined number of times for each of the livestock animals in the farming environment 200, with the plurality of random permutations thereof. It will be appreciated that the execution of the permutations by the data processing arrangement 102 provides statistical significance to the n-states determined by the data processing arrangement 102 and enables to increase a confidence associated therewith.

Optionally, at (iv), the data processing arrangement 102 finds in operation the networks of distinct n-combinations sharing one or more properties. In other words, at (iv) the data processing arrangement 102 performs in operation execution of a network analysis procedure. For example, the networks of distinct n-combinations can be different Nstates that have at least one common SNP. In a first example, N-states determined by the data processing arrangement 102 in a third layer comprise 6-states A to F, such as, A [31 52 2470], B [31 2181 7751], C [31 52 8421], D [31 2470 2641], E [2181 2641 5112] and F [5112 7751

8421]. In such an example, the data processing arrangement 102 finds in operation (namely, is operable to find) networks from the N-states corresponding to each SNP common to one or more N-states, such as, 31 [A, B, C, D], 52 [A, C], 2470 [A, D], 2641 [D, E], 2181 [B, E], 8421 [C, F], 7751 [B, F] and 5112 [E, F].

Optionally, at (v), the data processing arrangement 102 performs in operation execution of a network validation procedure to eliminate random observations. In one example, the data processing arrangement 102 compares in operation a number of pseudo-cases within the network associated with permutations, against a number of cases within the network before performing the permutations. In such an example, if the number of pseudo-cases within the network is more than the number of cases within the network before performing the permutations for more than 50 networks out of 1,000 networks (or the p-value is more than 0.05), the null hypothesis is validated.

Optionally, the data processing arrangement 102 determines in operation a penetrance of the networks, such that the penetrance is associated with an amount of population that corresponds to the network. In an example, the penetrance is expressed as a percentage value.

Optionally, the data processing arrangement 102 is operable to merge identical networks, such as, networks having all identical properties. Optionally, the data processing arrangement 102 determines in operation a p-value for each network, against a network having higher NC and density than the network. The p-value indicates a probability that a SNP is associated with a particular phenotype, wherein the phenotype is any one of: a physical trait, a disease and so forth. Furthermore, p-value represents the significance of a genetic difference between two populations (case and control) at a particular locus on a gene.

Optionally, the data processing arrangement 102 performs in operation FDR. correction during multiple testing of the networks to compare the null hypothesis. In one example, the data processing arrangement 102 is operable to employ a technique such as Benjamini-Hochberg procedure or Benjamini-Hochberg-Yekutieli procedure to correct for the multiple testing on the networks. For example, the data processing arrangement 102 employs in operation a FDR of 1% for comparing null hypothesis. It will be appreciated that comparing the null hypothesis for the networks enables the data processing arrangement 102 to eliminate substantially random n-combinations that may have been determined by the data processing arrangement 102.

Optionally, at (vi), the data processing arrangement 102 performs in operation execution of a network annotation procedure. The semantically normalized knowledge graph contains information about the shared one or more properties including but not limited to, SNPs, genes, pathways, targets, drugs, diseases, pharmacogenetics and drug interaction and so forth. In one embodiment, the data processing arrangement 102 in operation selects one or more properties from the SNPs, genes, pathways, targets, drugs, diseases, pharmacogenetics and drug interaction in the semantically normalized knowledge graph. Subsequently, the one or more properties selected by the data processing arrangement 102, is correlated with the network of SNPs determined by the data processing arrangement 102, to determine information about the SNPs, such as, if the SNPs are in a coding, non-coding or eQTL (expression quantitative trait loci) region; the genes or pathways that are affected by the SNPs; diseases or phenotypes associated with the SNPs; a location within the genome where the SNPs are occurring; known phenotypic associations of the SNPs within the networks; if the diseases or phenotypes associated with the SNPs are druggable, and so forth.

Optionally, at (vii), the data processing arrangement 102 performs in operation the re-clustering of the networks by merging networks comprising at least one common SNP therein. In the first example, the networks comprising the SNPs 31, 52 and 2470 share the N-states A, C, D therebetween. Thus, the networks corresponding to the 31, 52 and 2470 can be merged into a cluster, such that the merged network has a connection between the N-states. In such an example, the data processing arrangement 102 merges in operation the cases corresponding to the N-states into the cluster. Furthermore, hypothesis driven criteria based on biological insights, role of specific metabolic pathways, phenotypic factors, clinical factors, and the like, may be applied and tested in the re-clustering stage by re-segmenting the case and control populations based on specific conditions.

According to an embodiment, the re-clustering is used to correlate validated networks with extended phenotypic and clinical data to find biological explanations for observed associations. The data processing arrangement 102 correlates in operation phenotypic and clinical data to find the biological explanations for the observed associations of the SNPs within the cluster. In one example, the phenotypic and clinical data is stored in the array model 106, for example, as semantically normalized knowledge graphs. In another example, the phenotypic and clinical data can be associated with merged networks corresponding to various other populations. In yet another example, data processing arrangement 102 is operable to correlate (namely, correlates in operation) hypothesisdriven criteria comprising biological insights, role of metabolic pathways, lifestyle data and so forth, find the biological explanations for the observed associations of the SNPs within the cluster.

Optionally, the dataset retrieved from the array model 106 by the data processing arrangement 102 comprises epigenetic data. However, the epigenetic data may correspond to continuous variables. In such an example, the data processing arrangement 102 converts the epigenetic data from the continuous variables into finite domains.

Optionally, the data processing arrangement 102 in operation, finds at least one other feature that is selected from omics, or non-genetic factors. As mentioned hereinabove, the data processing arrangement 102 correlates phenotypic and clinical data to find the biological explanations for the observed associations of the SNPs within the cluster. Furthermore, such phenotypic and clinical data associated with the cases can be used to determine the at least one other feature from omics, or non-genetic factors. In one example, the data processing arrangement 102 in operation finds cases and controls that share at least one nongenetic factor, such as a phenotypic, clinical and/or lifestyle factor. Furthermore, the data processing arrangement 102 performs high-order combinatorial association of the non-genetic factors and genetic factors, such as, presence and absence of SNPs in the cases and controls respectively, to identify disease protective effects associated with the controls.

It will be appreciated that in order to compute the welfare trajectory, and provide the output signals, the data processing arrangement 102 considers not only genotype of each of the plurality of animals, observations and tests carried out on each of the plurality of animals, detailed information of various medication and drugs that are given to each of the plurality of animals, on-going observations as the medications are applied to each of the plurality of animals , but also the temporally logged sensor signals obtained by the sensor arrangement 104 (such as, food intake, sun time, hoof and so forth).

Furthermore, it will be appreciated that the aforesaid Internet®-based system 100 allows for efficiently monitoring the farming environment 200 by way of the sensor arrangement 104, and also allows for substantially reducing a magnitude of computational tasks by way of the data processing arrangement 102, in a manner that such tasks can be executed by the software product on modest hardware such as laptop computers, tablet computers, desktop computers, smartphones, smartwatches, and the like. In other words, the data processing arrangement 102 can be implemented by way of the aforesaid equipment, in a cost-effective and user-friendly manner.

Optionally, the output signals are used to control at least one of:

(i) a type and/or a quantity of feed provided to the animals;

(ii) a time when feed is provided to the animals;

(iii) additional feed supplements and/or one or more drugs to be administered to the animals;

(iv) selective heating or cooling to be provided to the animals; and (v) pathogen reducing processes to be applied to the farming environment 200. Notably, the term output signal relates to a control signal that is to be implemented within the farming environment 200, for providing individualized support (namely, precision support) to the livestock animals. Specifically, the output signal is generated by the data processing arrangement 102, based upon the welfare trajectory to be used in providing individualized husbandry of each animal. More specifically, the output signal, when implemented, allows for the Internetbased system 100 to perform at least one physical function within the farming environment 200, based upon the computed welfare trajectory. It will be appreciated that the at least one physical function is not limited to the aforesaid functions (i), (ii), (iii), (iv) and (v) only, and could also include other physical functions such as opening/closing barriers within the farming environment 200 to assist in movement of the livestock animals, adjusting lighting within the animal stalls, and the like.

Optionally, the data processing arrangement 102 provide in operation the output signals to at least one actuator and/or at least one valve for implementing the at least one physical function in respect of the livestock animals within the farming environment 200. Alternatively, optionally, the data processing arrangement 102 provides in operation the output signals to the sensor arrangement 104, and the sensor arrangement 104 is configured (namely, is operable) to provide the output signals to the at least one actuator and/or the at least one valve for implementing the at least one physical function in respect of the livestock animals within the farming environment 200.

In an example, when the welfare trajectory computed by the software product describes that a given diet is to be administered to a given animal thrice a day at times ΤΙ, T2 and T3, the output signals for controlling the type and/or the quantity of feed provided to the animals may be provided by the data processing arrangement 102 to the sensor arrangement 104. The sensor arrangement 104 may be operatively coupled to at least one actuator of an animal feeding mechanism (such as a feeding tray). In operation, the sensor arrangement 104 may employ the output signals to control operation of the at least one actuator of the animal feeding mechanism, to administer the given diet to the given animals at the times ΤΙ, T2 and T3.

In another example, when the welfare trajectory computed by the software product describes that a given animal should have a minimum 12 hours of physical activity in a day and be administered a given drug once a day, a first output signal may be used to control the physical activity of the given animal and a second output signal may be used to control administration of the given drug. In such an example, the sensor arrangement 104 may provide the first output signal to at least one actuator associated with gates and/or barriers within the farming environment 200 to keep the gates and/or the barriers closed in a manner that the given animal is unable to retire to its stall until the minimum 12 hours of physical activity in the day elapse. Furthermore, in such an example, the sensor arrangement 104 may provide the second output signal to at least one actuator associated with a drug dispensing mechanism arranged within the farming environment 200, to administer the given drug to the given animal, once a day (for example, when the given animal is sleeping).

Optionally, the Internet®-based system 100 includes an avatar testing arrangement (not shown) including a test animal into which genotype information from a given animal is imported, and combinations of drugs and/or feed supplements are tested to determine whether or not toxicological problems are likely to arise if a given combination of drugs and/or feed supplements are administered to the given animal. Notably, the avatar testing arrangement allows for the Internet-based system 100 to evaluate treatments before they are applied to the livestock animals, to allow for reducing possibility of undesired health complications within the livestock animals due to such treatments. As an example, a given combination of drugs and/or feed supplements may be tested to determine that toxicological problems are likely to arise, for example, with a certainty greater than or equal to 60 percent, if the toxicological problems arise for a majority of test animals under the aforesaid investigation.

In an embodiment, the avatar testing arrangement is implemented by way of hardware (for example, such as computing devices, medical devices, and the like), software (for example such an avatar test simulation application), firmware, or a combination of these.

Referring to FIG. 4, illustrated is a process diagram 400 illustrating steps that are implemented by the avatar testing arrangement, in accordance with an embodiment of the present disclosure. In the process diagram 400, at step 402, the avatar testing arrangement imports genotype information from the given animal into the test animal, and at step 404, the avatar testing arrangement tests the combinations of drugs and/or feed supplements to determine whether or not toxicological problems are likely to arise if a given combination of drugs and/or feed supplements are administered to the given animal. In such a scenario, the avatar testing arrangement is operable to experimentally determine any toxic effects that may be caused by various combinations of food supplements and medications, upon such combinations of food supplements and medications being administered to the given animal. Notably, the term given animal relates to an animal to which the given combination of drugs and/or feed supplements are to be eventually administered for improving its overall health and well-being whereas the term test animal relates to an animal upon which the given combination of drugs and/or feed supplements are to be tested, prior to their administration to the given animal. The test animal may be introduced with genotype information (such as, genes, DNA) of the given animal for which the combinations of drugs and/or feed supplements are to be tested. The test results and effects are measured and observed, via the avatar testing arrangement, to predict an effect thereof on the given animal. In an example, cancerous tissues from the given animal (for example, such as a cow) may be introduced into a body of the test animal (for example, such as a mouse). Subsequently, in such an example, different drugs and combinations thereof may be tested on the mouse and results thereof may be observed and analysed, via the avatar testing arrangement. The drugs and combinations that have a positive effect on the mouse may be administered to the cow for treatment thereof.

Optionally, the avatar testing arrangement uses the test animal implemented as a fruit fly. It will be appreciated that the fruit fly (namely, Drosophilia) are used in research as it is easily cultured and has a short generation time. As a result, mutant test animals are readily obtainable making admisitration (administration?) of foreign genes easier. In an example, a tumor biopsy is obtained from a given livestock animal, and the entire exome is sequenced and analyzed to capture information of SNPs present therein. A tumor network is formulated based on the exome sequencing and SNP selection. Furthermore, mutations comprising tumor network are engineered into fruit fly (Drosophila) avatars. In an example, Fruit fly (Drosophila) avatars that recapitulate key features of cancer, may be used to study genome of an animal suffering from cancer and provide an effective approach for developing novel targeted therapies. Such engineered Drosophila avatars are used for screening a combination of regulatory approved drugs (including noncancer drugs). Based on results from drug screening, a treatment strategy, for example, based on combination therapy, is recommended to the given livestock animal. Beneficially, such tests reduce a potential formulary of 200 drugs to combinations of around 20 drugs.

Referring to FIG. 5, illustrated are steps of a method 500 of (for) using an Internet®-based system (for example, such as the Internet-based system 100 of FIG. 1) that provides individualized support to livestock animals in a farming environment, in accordance with an embodiment of the present disclosure. In the method 500, the Internet®-based system includes a data processing arrangement that receives in operation temporally logged sensor signals from a plurality of sensors that are spatially distributed within the farming environment, wherein the data processing arrangement includes an array model, for example including a multi-dimensional array model, of the data processing arrangement against which the temporally logged sensor signals are compared, wherein the data processing arrangement provides output signals that control operation of the farming environment, and wherein the data processing arrangement executes a software product that in execution analyzes the temporally logged sensor signals in respect of the array model and generates the output signals. At a step 502, a software product is arranged to perform inferences from the multi-dimensional array model, the inferences being based on at least a subset of the temporally logged sensor signals and a genotype determination by DNA sequencing of each animal hosted within the farming environment. At a step 504, environmental conditions experienced by each animal, including food intake for each animal are sensed using the sensor arrangement. At a step 506, the array model is populated with at least one of drug characteristics, food supplement characteristics, husbandry strategies depending on animal health complications, disease characteristics of each breed (genotype) of animal, animal genotype data, and Single Nucleotide Polymorphism (SNP) data. At a step 508, the software product is used to compute a welfare trajectory to be used in providing individualized husbandry of each animal.

Optionally, the method 500 includes arranging for the SNP data to include single nucleotide polymorphisms characterizing each animal, determined by using Polymerase Chain Reaction (PCR) to read genetic tissue samples derived for each animal.

Optionally, the method 500 includes arranging for the sensor arrangement to include a plurality of sensors that are spatially distributed within the farming environment and are coupled in communication with the data processing arrangement by using a wireless dynamically reconfigurable communication network.

Optionally, the method 500 includes arranging for the sensor arrangement to include tags that are attached to the livestock animals and that monitor and log environmental conditions experienced by each animal and movement performed by each animal.

Optionally, the method 500 includes arranging for the wireless dynamically reconfigurable communication network to be implemented as a peer-to-peer network.

Optionally, the method 500 includes arranging for the Internet®-based system to collect in operation one or more pathogens present in the farming environment, perform genotype sequencing of the one or more pathogens to characterize the one or more pathogens, and employ the characterization of the one or more pathogens as an input parameter to the software product executed in the data processing arrangement to use in performing its search.

Optionally, the method 500 includes using the output signals to control at least one of:

(i) a type and/or a quantity of feed provided to the animals;

(ii) a time when feed is provided to the animals;

(iv) selective heating or cooling to be provided to the animals; and (v) pathogen reducing processes to be applied to the farming environment.

Optionally, the method 500 includes arranging for the Internet®-based system to include an avatar testing arrangement including a test animal into which genotype information from a given animal is imported, and combinations of drugs and/or feed supplements are tested to determine whether or not toxicological problems are likely to arise if a given combination of drugs and/or feed supplements are administered to the given animal.

Optionally, the method 500 includes arranging for the avatar testing arrangement to use the test animal implemented as a fruit fly.

Optionally, the method 500 includes arranging for the sensor arrangement to include a processing module that pre-processes the temporally logged sensor signals, using at least one artificial intelligence algorithm.

Optionally, the present disclosure provides a computer program product comprising a non-transitory computer-readable storage medium having computer-readable instructions stored thereon, the computer-readable instructions being executable by a computerized device comprising processing hardware to execute the method 500. Optionally, the computer-readable storage medium comprises one of a floppy disk, a hard disk, a high capacity read only memory in the form of an optically read compact disk or CD-ROM, a DVD, a tape, a read only memory (ROM), and a random access memory (RAM).

The present disclosure provides the aforesaid Internet-based system, method of using the Internet-based system, and computer program product to provide individualized support to livestock animals. The Internet-based system can be implemented by way of readily available off the shelf' hardware that are often mass-produced, and can therefore be easily employed within the farming environment in a cost-effective manner. Furthermore, the Internet-based system allows for substantially reducing a magnitude of computational tasks by way of the data processing arrangement. As a result, the computational tasks described herein can be executed on modest hardware such as laptop computers, tablet computers, desktop computers, smartphones, smartwatches, and the like. Moreover, the Internet-based system and the method allow for accurately and reliably providing individualized support to livestock animals in the farming environment.

Modifications to embodiments of the present disclosure described in the foregoing are possible without departing from the scope of the present disclosure as defined by the accompanying claims. Expressions such as including, comprising, incorporating, have, is used to describe and claim the present disclosure are intended to be construed in a nonexclusive manner, namely allowing for items, components or elements not explicitly described also to be present. Reference to the singular is also to be construed to relate to the plural.

ANNEX

GLOSSARY

Constraint resolution means establishing substantially all valid combinations of variables satisfying substantially all constraints of a given system. Optionally, all valid combinations of the variables satisfying all the constraints of the given system are established, namely computed, wherein, in an optional case, valid Cartesian sub-spaces of states or combinations satisfy a conjunction of all system constraints for all interconnected variables. The valid Cartesian sub-spaces may comprise Cartesian planes. A point in such Cartesian planes can be represented as tuples (or a list) of 'n' real numbers, wherein 'n' can be dimensions associated with the Cartesian plane. It will be appreciated that when the Cartesian sub-spaces are associated with Cartesian planes, the variables and the constraints corresponding to the variables can comprise more than 3 values associated with abscissa (x-axis), ordinate (y-axis) and applicate (z-axis) of the Cartesian co-ordinate system.

The term optimizing means applying a heuristic selection of combinations within a set of valid combinations.

The term a system spanned by variables on finite domains and/or intervals indicates that each variable of a given system consists of a finite set of elements or state values (for example, logical truth values) or a finite set of intervals.

The term an addressable solution space indicates that substantially all valid combinations are explicitly represented.

The term a Cartesian sub-space is a compact representation of one or more valid combinations, wherein all combinations are derivable/calculable as a Cartesian product of elements or state values for each variable. It will be appreciated that when the Cartesian subspace comprises Cartesian planes, the Cartesian product of elements or state values of each variable may be associated with products of more than the 3 values corresponding to the Cartesian coordinates x, y and z.

The term system constraints refers to relations (namely propositional functions) for variables defined for a given system.

The term interconnecting variables indicates variables present in at least two relations.

The term link variable means a variable generated by a method according to the present disclosure and added to a given relationship with a unique index, wherein the unique index identifies one corresponding Cartesian sub-space.

The term interconnected valid Cartesian sub-spaces means valid Cartesian sub-spaces with at least one common variable associated therewith.

The term external variables means variables that are to be used by or being accessible from an environment during a runtime simulation. The term external variable is used herein interchangeably as external state variable.

The term internal variables or interim variables means variables that are not to be used by, or are not to be accessible from an outer environment during a runtime simulation.

The term duster means an accumulation of states, or a list of state vectors associated with known attributes. The state variables are subsets of domain of static array system model and/or external variables.

Moreover, the embodiments are capable of performing real-time processing. Furthermore, real-time means in practice while a user of the system waits for results of computations that are delivered in a time scale of tens of seconds, or within several minutes, even when largescale constraint problems are being computed and resolved by the system. It will be appreciated that , in practice, personalized (contextsensitive) recommendations from hyper-dimensional data feeds may be provided in real time on a wearable, mobile or loT (Internet of Things”) device. The aforementioned hyper-dimensional data feeds provide hyperdimensional data that are stored in an array system model, wherein the array system model may represent constraints as well as other types of knowledge associated with each valid combination of the array system model, namely one or more object functions, all of which must interface with an environment provided to a given in a simple interactive way, via a user interface.

The term tractable-time means in practice a time, of the polynomial order (i.e. n², n³, n⁴ and so on), required by a computing arrangement for computation of a large-scale constraint problem.

In overview, embodiments of the present disclosure employ in operation a multi-dimensional system model (namely, an array system model), for performing data processing using a computing arrangement of a control apparatus. The terms multi-dimensional system model and array system model are hereinafter used interchangeably in the description. Moreover, the computing arrangement is capable of performing real-time processing. Notably, real-time” means that results of computations are delivered in a time scale of tens of seconds, or within several minutes, even when large-scale constraint problems are being computed and resolved by the computing arrangement. It is thereby feasible, in practice, to provide personalized (context-sensitive) recommendations from hyper-dimensional data feeds in real time on a wearable, mobile or loT (Internet of Things”) device. The aforementioned hyper-dimensional data feeds provide hyper-dimensional data that are stored in the array system model, wherein the array system model may represent constraints as well as other types of knowledge associated with each valid combination of the array system model, namely one or more object functions, all of which must interface with an environment provided to a given in a simple interactive way, via a user interface.

Furthermore, the control apparatus includes a user interface for interacting with a user of control apparatus for controlling operation of the control apparatus, a data processing arrangement that is operable to receive the one or more data inputs and to output the one or more control outputs and/or one or more analysis output and/or one or more recommendation outputs, and the computing arrangement that supports automatic modelling, analysis and real-time inference processing on multi-dimensional system models, can be implemented by way of a wide range of computational hardware. Such computational hardware includes, signal processing and embedded controllers to mobile devices (for example, smart watches, smartphones and tablets), standard computers (for example, personal computers or laptop computer), graphics processing units (GPUs), distributed computers with parallel processing capabilities, and so forth. The data processing using such computing hardware of the control apparatus, is capable of enabling a wide range of innovative decision support tools, such as clinical decision support systems, to be realized in practice.

Optionally, the multi-dimensional system models are constraint problems expressed in terms of truth tables with MN combinations, wherein each combination assumes either a truth-value true (valid) or a truth-value false (invalid). In such a case, the multi-dimensional system models assume that N variables are involved, wherein each variable has M elements. In general, each valid combination in a solution space computed in embodiments of the present disclosure may have one or more associated attributes or object functions, for example a price. In a special case of an embodiment of the present disclosure, all combinations may be valid, namely without any constraints on the system model being employed when computing results.

It will be appreciated that computing M^N combinations using contemporary known computation methods will result in a combinatorial explosion in a contemporary computing device, that would result in an unacceptably long computation time for providing results to users via a user interface. Therefore, it is not currently a trivial task to solve large constraint problems with a multitude of variables. Nevertheless, embodiments of the present disclosure make it possible to unify seemingly contradictory requirements for completeness (all combinations must be accessible to ensure logical consistency) with compactness of representation and real time inference processing even with complex combinatorial applications on relatively low power computational devices (for example, as aforementioned).

Furthermore, it will be appreciated that the control apparatus is practical and useful, and optionally, compact and portable. As mentioned previously, the control apparatus includes data processing arrangements that are operable to execute software products that are able to provide solutions to medical problems and other types of technical control problems, without resulting in a combinatorial explosion that results when multi-dimensional tasks are being addressed. As it will be understood from the following description, embodiments of the present disclosure employ an advantageous form of data representation, referred to as the array system model or the multi-dimensional system model.

While the array system model is an optimal tool for complex constraint problems described in the foregoing, it will be appreciated that embodiments of the present disclosure are not limited to addressing medical related problems; for example, embodiments of the present disclosure can be used in safety critical power stations (for example, nuclear power plant, arrays of wind turbines, arrays of ocean wave energy converters and so forth), for supervising oil well equipment, for chemical plant, for airborne radar systems, for railway network management, for automatic driverless vehicle systems, and similar.

Thus, embodiments of the present disclosure concern a method of generating the array system model useful for interrogating and/or configuring and/or optimizing and/or verifying a logical system spanned by variables on finite domains and/or intervals, wherein the method comprises: (i) generating and storing, in a memory or a storage medium of the computing arrangement, an addressable solution space for a set of external variables, wherein the addressable solution space is expressed in terms of all valid Cartesian sub-spaces of states or combinations for the set of external variables with interconnected valid Cartesian subspaces being addressable as valid combinations of indices of link variables and/or core link variables; and (ii) arranging for the solution space to satisfy a conjunction of all, or substantially all, relations of the set of external variables, in order to establish a system model in which all, or substantially all, valid solutions are stored as nested relations.

In embodiments of the present disclosure, there are encountered raw data feeds (one or more inputs provided to the system) that are complex and multi-dimensional; such raw data feeds are, for example, derived, at least in part, from sensor arrangements. However, there arises a need to transform such complex and multi-dimensional raw data feeds into transformed to useful actionable insight, wherein real-time inferencing is required to be performed and personalized, and wherein there is generated context-specific recommendations or advice. In practice, for embodiments of the present disclosure, there are distinct advantages to being able to compute across such raw data on data collection hardware itself that generates the raw data in operation (for example, a smartwatch, a mobile phone or a remote sensing platform), as such a manner of operation negates requirements for large data transfers that are a potential target of data interception; moreover, such large data transfers potentially consume expensive resources in terms of both network bandwidth and power on small, battery powered mobile devices.

Embodiments of the present disclosure are operable to employ, for their variables and constraints, a semantically normalized knowledge graph (namely, 'knowledge graph'}. Moreover, such knowledge graphs are beneficially used in the embodiments to represent all available information from a variety of public and other data sources containing information associated with variables, relationships and constraints operating on a given complex system.

Such knowledge graphs are optionally based on a master multi-relational ontology, which includes a plurality of individual assertions, wherein an individual assertion comprises a first concept, a second concept, and a relationship between the first concept and the second concept, wherein at least one concept in a first assertion of the plurality of individual assertions is a concept in at least a second assertion of the plurality of assertions.

For each pair of related concepts, there is beneficially a broad set of descriptive relationships connecting the related concepts, for example expressed in a logical and/or probabilistic as well as linguistic manner. As each concept within each pair is potentially paired (and thus related by multiple descriptive relationships) with other concepts within a given ontology, a complex set of logical connections is formed. A corresponding superset of these connections provides a comprehensive knowledge graph describing what is known directly and indirectly about an entirety of concepts within a single domain. The knowledge graph is also optionally used to represent knowledge and relationships between and among multiple domains and derived from multiple original sources.

In another beneficial embodiment of the present disclosure, a semantic distance or relatedness of concepts in a specific context is calculated. Such probabilistic semantic distance metrics are susceptible to being represented as relationships between two concepts in the semantically normalized knowledge graph and used to determine a degree of connectedness of concepts above, below or between selected thresholds in a context of a specific domain or corpus.

In these aforementioned embodiments of the present disclosure, the specification of a given subset of the knowledge graph to be derived for the array system model optionally includes a selection of two or more concepts or types of concepts from a plurality of assertions of a master multi-relational ontology, applying one or more queries to two or more concepts or concept types to yield a subset of individual assertions from the plurality of assertions, wherein the queries identify one or more individual assertions from the plurality of individual assertions of the master multi-relational ontology. Specifically, the master multi-relational ontology connects the two or more concepts directly or indirectly. In a context of complex domains such as healthcare examples described in the foregoing, such derived knowledge graphs potentially contain millions of concepts, each of which has multiple properties (namely, variables) with multiple potential values, and each of which may have up to tens of thousands of direct or derived logical constraints.

In describing embodiments of the present disclosure, a term logical system is used to mean a complete system, alternatively a sub-system that is a part of a larger system. When used to refer to a sub-system, variables associated with other sub-systems are treated as being external variables.

In embodiments of the present disclosure, for example implemented as a control apparatus employing data processing hardware, all invalid states or combinations violating constraints of a given system are excluded from relations that are employed in operation of the multidimensional system model. Such exclusion of invalid states or combinations is beneficially performed when the system model is generated by a method pursuant to the present disclosure; in other words, in embodiments of the present disclosure, the invalid states or combinations are excluded from computations whenever identified to enable more rapid computation of useful results to be achieved. In practice, a state of contradiction or inconsistency is present in a system if just one relation of the system has no valid combination or state. Conversely, the system is regarded as being consistent if at least one state or combination of states is valid; namely, one state or a combination of states satisfies all system constraints. At an instance, when generating a given system model, just one relation of a system is found to have no valid combination or state, then that whole system is in a state of contradiction or inconsistency and is excluded for achieving enhanced computational efficiency.

Optionally, the method includes operating the multi-dimensional system model to have a plurality of system model states, and to change state from a given preceding system model state in among the system model states to a subsequent system model state among the system model states, depending upon a computed solution to the given preceding system model state and operative input data applied to the multidimensional system model.

Furthermore, a process of colligating relations (that is, combining relations to arrive at a more complex sub-system or system) is elucidated in detail. It will be appreciated that on each level of a process of colligation, inconsistencies or contradictions are identified in embodiments of the present disclosure, and will, thus, result in exclusion of the colligated sub-system or system. Thus, when a generating process has been completed in embodiments of the present disclosure, the system will be consistent, as manifested by all relations having at least one valid Cartesian sub-space.

In the present disclosure, the term system is used to refer to an entire system of variables or, alternatively, to a part of the entire system of variables, for example as aforementioned. With reference to a specific application (for example healthcare), the system provides a representation of a complete set of available domain knowledge upon which real-time reasoning or inferencing can be performed using embodiments of the present disclosure to provide useful, actionable controls, insights and recommendations using decision support tools incorporating an array system model (for example, for selecting a best available therapy for a specific patient at a point of care; for example, for selecting a best available selection of replacement component parts to be used when repairing an item of machinery). There is thereby provided an interaction between the array system model and an environment that is carried out by a state vector representing states of all variables involved, including physical measurements as well as decision parameters. Thus, in example embodiments of the present disclosure, variables involved can include sensor signals acquired using physical sensors, and decision parameters can be outputs that are used to control operating states of various apparatus, for example in a hospital, in an industrial plant, in a vehicle, in an energy power plant, and so forth.

In embodiments of the present disclosure, the given system is completely defined in that every combination under the system is either valid or invalid with respect to each of the system constraints relevant to use of the multi-dimensional system model and preferably with respect to absolutely each of the system constraints. Thus, the term 'system', used about the entire system of variables, indicates that the entire system is completely defined with respect to all system constraints relevant to the use of the system model, and optionally with respect to absolutely all system constraints. When the system of variables is not completely defined in above sense of this term, then only that part of the system which is in fact completely defined is covered by the term system pursuant to the present disclosure. The term substantially indicates a system in which process of colligation has not been completed, and where a runtime environment must be adapted to perform certain tests for consistency; for example, substantially all refers to at least 90%, more optionally at least 95%, and most optionally at least 99%.

As aforementioned, the system constraints are optionally determined by conjugating one or more relations, wherein each relation represents valid Cartesian sub-spaces of states or combinations on a given subset of variables. The conjugation of the one or more relations comprises calculating Cartesian sub-spaces satisfying the combined constraints of the one or more relations. If no relations have common variables, no further action is required to conjugate the relations in embodiments of the present disclosure.

According to an important optional feature of the invention, all relations with at least one common variable are colligated. The colligation comprises conjugating the constraints of two or more relations that are connected by having common variables therebetween to establish one or more Cartesian sub-spaces satisfying combined constraints of the two or more relations. Furthermore, the colligation of two or more relations will normally be performed by joining the two or more relations up to a predetermined limit. Such joining comprises an operation of replacing a set of relations with a single relation satisfying combined constraints of the set of relations.

The set of relations is not limited to two relations, but can in general be any finite number of relations. In an example embodiment of the present disclosure, a case where three or more relations are joined is typically decomposed into a number of pairwise joins; this pairwise joining optionally comprises a predetermined strategy or this pairwise joining is optionally in a random order. Moreover, joining of relations will typically reduce the number of relations, and the result will be one or more relations with common link variables. Moreover, the linking of the relations consists of adding link variables and adding one or more calculated relations representing constraints on the link variables.

In an embodiment of the present disclosure, any relation with nonconnecting variables as well as connecting variables is extended by adding a unique link variable with a unique index identifying each valid Cartesian sub-space on either the non-connecting variable or the connecting variables. In such situations, it is often advantageous to split a given relation into two relations, wherein one relation pertains to the non-connecting variables and the link variable, and the other relation pertains to the connecting variables and the link variable.

In relation to embodiments of the present disclosure, a term completeness of deduction indicates that all logical consequences are required to be deduced for one or more variables. Moreover, embodiments of the present disclosure, the completeness of deduction relates to all logical consequences on all variables, but as indicated above, the embodiments of the present disclosure are not limited to computing all logical consequences.

When the colligation process is completed, relations for isolated variables are optionally split into a plurality of smaller interconnected relations with the isolated variables are expanded to form (namely tuples). It is to be understood that such a representation is potentially more compact than compressed Cartesian arguments, and will make it possible to associate object functions to each single combination of the defining variables.

When the array system model is to be used for optimization or learning, one or more object functions, for example pricing functions, are optionally incorporated into the array system model. An object function of a given subset of variables, wherein the object function derives characteristics of a given subset of variables, and is linked to a complete solution space by deducing constraints imposed by the object function on each link variable connected to the given subset of variables. After the array system model has been generated by the method pursuant to the present disclosure, object functions can provide information between a set of variables and a set of object function values, for example cost, price, risk or weight.

As an example in healthcare, given a patient's set of co-morbidities and co-prescriptions, it is potentially contemporarily not possible to select a drug for a particular disease from any of available options that does not present some significant risk of interactions or side-effects arising. In such a case, it is necessary to choose a best available drug, which reduces, for example minimizes, a likelihood and/or severity of any of these potential interactions or side-effects. Such a reduction, for example minimization, can be achieved by accepting a partially incomplete deduction (with, for example, a single invalid variable), and then using object functions as described below to evaluate and optimize the likely outcomes, such as potential patient benefit, treatment cost and side-effect risk.

If a set of object function values does not have a natural order, in contrast, for example with numbers, an arbitrary order can be assigned to the set of object function values.

Characteristics of the object function are susceptible to being determined; moreover, constraints on the link variables deduced on each combination of the given variables can be determined, wherein the result is represented as a relation on the object function, the given variables, and the link variables. These characteristics are optionally values of the object function given by functional mapping of a set of independent variables or a set of constrained variables. The mapping can also be a general relation yielding one or more object function values for each combination of the variables.

Embodiments of the present disclosure provide a method of interrogating and/or configuring and/or optimizing and/or verifying and/or controlling a system spanned by variables on finite domains, wherein the method comprises:

(i) providing an array system model in which substantially all valid solutions in the system are stored as nested arrays representing valid Cartesian subspaces on all external variables, with all interconnected valid Cartesian subspaces being addressable as valid combinations of indices of link variables; and (ii) deducing any sub-space, corresponding to an input statement and/or query, of states or combinations spanned by one or more variables of the system represented by the nested arrays by deriving the consequences of a statement and/or an query by applying the constraints defined by the statement and/or query to the system model.

In respect of embodiments of the present disclosure, deducing refers to deriving or determining logical inferences or conclusions, for example all inferences or conclusions, from a given set of premises, namely all the system constraints.

In respect of embodiments of the disclosure, the term query refers to a question for which the array system model is operable to provide answers, for example, a question regarding a particular combination of sensor signal values, but not limited thereto, subject to defined conditions. An exemplary question concerns one or more valid combinations of a given set of variables satisfying the system constraints and, optionally, also satisfying an external statement. An external statement may be a number of asserted and/or measured states and/or constraints from the environment. Moreover, a deduction of any subspace of states or combinations is performed on a given subset of one or more variables either without or colligated with asserted and/or measured states and/or constraints from the environment.

An interaction between the system represented by the array system model and the environment is suitably performed by means of a state vector (SV) representing all valid states or values of each variable.

Thus, an input state vector (SV1) is employed to represent the asserted and/or measured states from the environment, whereas an output state vector (SV2) is used to represent one or more deduced consequences on each variable of the entire system when the constraints of SV1 are colligated with all system constraints in the array system model.

Optionally, the multidimensional system model includes static constraints, clusters of accumulated states, and dynamic rules which represents valid transitions between valid states.

In a preferred embodiment of the present disclosure, each invalid variable may be either discarded from the environment (SV1) or may be deduced as a consequence (SV2). Furthermore, optionally, variables defined as output variables are allowed to change a state without causing a contradiction. Moreover, deduction may be optionally performed by consulting one or more relations and/or one or more object functions at a time by colligating a given subset of variables in a relation with given subsets of states in a state vector and then there is deduced therefrom possible states of each variable.

In embodiments of the present disclosure, clustering and dynamic properties are employed in operation of the array system model. Such clusters represent a list of state vectors associated with known attributes. States of the cluster are determined from external variables (EV) and/or internal state variables that span the array system model. Relationships between the states of the clusters and state variables are defined by a cluster relation. For example, a given cluster relation has three state variables: a state of the cluster, and variables VI and V2. In operation of embodiments of the present disclosure, there may be a logical OR between rows in a relation table (namely, as in a disjunctive form). Alternatively, the cluster relation is a relation between the states of the clusters and state variables, wherein states of clusters are input and state variables are output. The cluster relations reduce a hyper-dimensional space, having millions of parameters, to a corresponding multidimensional array system model. When the states of the external variables are known, processing of the cluster relation in run time may be described as including steps as follows:

(i) comparing external measurements with the states of cluster in the cluster relation and identify corresponding matching rows;

(ii) deducing values of the output state variables VI and V2; and (iii) deducing the constraints on all other state variables by a state propagation in the array system model.

Completeness without colligation can be ensured as the given cluster may be only a part of one relation and therefore considered as an isolated variable in the multi-dimensional and complete array system model.

Exemplary applications of clustering include, precision medicine, stateevent processing and many other exceptionally complex applications.

A consultation of a relation is beneficially performed by colligating, for example joining, the relation and states of variables present in the relation. The consultation provides a result that can be a projection (namely, a union of all elements) on each variable of the colligated relation, or the result can be the colligated relation. The colligation is optionally a joining, however, it will be appreciated that the consultation of each relation is not limited thereto. In an example embodiment of the present disclosure, two or more variables are colligated in parallel; projections on two or more variables are similarly performed in parallel. However, it will be appreciated that embodiments of the present disclosure are not limited to such parallel implementation, and the embodiments are optionally susceptible to being implemented sequentially.

In an embodiment of the present disclosure, completeness of deduction is obtained by consulting connected relations, until no further consequences can be deduced on any link variable. Such an operation is termed state propagation. Moreover, such a state propagation comprises consulting two or more relations in parallel, namely concurrently.

The parallel execution of the state propagation may be implemented on one or more GPUs (Graphics Processing Units) or hardware designed for such parallel execution. The interaction between the array system model and the environment by the state vector may be carried out by simple operations that are suitable for a hardware implementation on devices such as embedded control systems, Internet of Things (loT) sensors or Field Programmable Gate Arrays (FPGAs).

An important feature of configurations and/or optimizations employed in embodiments of the present disclosure is that states of contradiction can be identified, namely when no valid states or values are deduced when consulting, namely investigating or checking, at least one relation. Such identification of contradictions and an elimination of a need to perform computations in connection with the contradictions, enable methods of the present disclosure to reduce computational resources required for performing complex hyper-dimensional computations.

The array system model (referred to as ASM in the following) is a compact and complete representation of all valid combinations and associated object functions of constraint problems on finite domains or intervals. The ASM is used to represent a person, an apparatus, a facility, a factory or similar system. A solution space of valid states or combinations is beneficially represented geometrically in terms of nested data arrays, and the ASM is simulated very efficiently in operation by simple operations on these arrays using CPUs (central processing units), GPUs (graphics processing units) or hardware devices designed for this specific use.

Major data flows required for performing ASM modelling include input data, for example a user-defined specification of system constraints in terms of a set of rules or relations pertaining to a given set of variables. Thus, the ASM modelling is implemented in a six-step procedure, wherein the six-step procedure includes STEP 1 to STEP 6 as follows:

STEP 1: Compile variables and relations

Each user-defined variable and each relation is compiled into the internal array representation. At this stage STEP 1, the relations are considered as independent items.

STEP 2: Colligate relations, verification of system

The solution space of the entire system is determined by colligating interconnected relations (constraint elimination). The system is simultaneously tested for logical consistency and redundancy. Embodiments of the present disclosure relate inter alia to a more efficient colligation strategy.

STEP 3: Minimize and link complete solution space

The complete solution space can be, for example, minimized and restructured in order to meet requirements in a runtime environment. Examples include: minimizing memory footprint to enable operation on a wearable device; splitting the array system model into multiple instances for parallel processing hardware; adding object functions on combinations of selected variables; and adding dynamic constraints in terms of relations as well as states to enable real-time response to signals from loT or wearable sensors.

Step 4: Link object functions

Optionally, the relations may be extended with further attributes, when the valid combinations satisfying the system constraints are associated with values or object functions to be optimized or used for specific applications, such as, for example, a price or soft constraints such as side-effect risk and severity with further values than just true or false.

Step 5: Cluster states and duster relations

Optionally, clustering is performed to reduce the hyper-dimensional space, potentially with millions of parameters, to the multi-dimensional ASM for performing decision support. Examples include: millions of genomic phenotypic and clinical variables that are condensed/reduced to a few hundred variables, which is utilized by decision support system. Clustering is based on cluster states (i.e. states of clusters} and cluster relations.

Step 6: State-event relations

Optionally, state-event relations utilize external events to describe the change from one state to another. Clustering is based upon internal state variables representing the conditions for change of state.

At this stage, the process of ASM modelling is finished. The entire solution space is now susceptible to being addressed by coordinate indexing and other simple operations on the nested arrays.

Each item of the state vector SV represents the state (namely the valid values) of an associated variable. For example, in respect of the input state vector SV1, one or more variables are bounded due to external measurements or assertions. Moreover, the input state vector SV2 represents the resulting constraints on all variables. The properties of the ASM are summarized as follows: (i) a run-time execution on the ASM is performed with completeness of deduction in real-time, namely with predictable use of processing time and memory. The ASM technology is therefore suitable for use in embedded decision support or for use in control systems on small computer devices: and (ii) the ASM representation is compact and complete. Embedded applications of embodiments of the present disclosure are required to fulfill all requirements for compactness, completeness and real-time capability with limited computing resources, even on large system models.

SIMPLE COLLIGATION STRATEGY (ADB)

Optionally, a generation of the ASM technology (to be abbreviated to ADB in the following) is based on a simple colligation strategy by pairwise joins of relations and then linking isolated variables whenever possible, the relations are operable to share variables. Moreover, the colligation graph is an illustration of a structure of interconnected relations, wherein nodes represent relations and arcs represent common variables of two of the relations.

A first colligation step is to compile each relation, namely to determine valid combinations of each relation. It will be appreciated that all invalid combinations are eliminated from each of the relations. Moreover, the valid combinations are expressed in terms of Cartesian sub-spaces; however, it will be appreciated that other coordinate spatial reference frames may be optionally employed for implementing embodiments of the present disclosure.

A second colligation step is to colligate the relations to determine the solution space of the conjunction of all relations. It is now possible to perform inference processing by performing simple array operations. The state vector is the important link between the compiled (colligated) array system model and the environment. The output state vector is deduced by consulting the complete solution space. The state of each variable is deduced by computing the union of elements from the two valid Cartesian sub-spaces. In general, the colligation process is carried out by pairwise joins of the relations, and after each join isolated variables are separated (assuming at least two isolated variables) into new relations connected by common link variables representing the valid Cartesian sub-spaces. The state vector is deduced by consulting one relation at a time, until no further constraints are added to each variable (state propagation).

Thus, the state propagation on a tree structure of interconnected relations (by the valid states of the common link variables) ensures completeness of a given deduction; in other words, all constraints on all variables are deduced in embodiments of the present disclosure. It will be appreciated that completeness of deduction is not possible by state propagation on the array representation of user-defined relations. Thus, there arises a need to colligate all interconnected relations in advance, even on such a very simple cyclic structure with only a single variable connecting each relation pair.

The simple colligation strategy (namely ADB) technology described in the following is susceptible to being summarized as follows:

1. A given process of joining relations with common variables and linking isolated relations on isolated variables is potentially impossible to implement in practice on large sets of relations due to a possible blow-up in size of a corresponding joined result (namely, is computationally impossible to achieve in practice using contemporary computing hardware). Such requirement for huge computational resources is an insurmountable and constant issue arising on account of a complexity of constraints in a range of practical technical fields of use of intelligent data processing systems in fields such as healthcare and life sciences.

2. If the process of joining relations can be completed, a binary output that is thereby achieved does satisfy requirement for completeness, but does not satisfy other requirements for embedded solutions, namely:

a. A representation thereby derived will not be as compact as possible, and potentially must be reduced in size, for example minimized in size, to meet specific hardware requirements for achieving size and real-time capability.

b. A complete solution space will not necessarily be accessible by parallel processing hardware using simple instructions, for example using GPUs.

c. The complete solution space must be addressable in order to include object functions. A compressed data format (namely, nested Cartesian arguments) may not be a suitable representation for variables defining an object function; relations for these variables are beneficially in an expanded form representing all valid tuples rather than Cartesian arguments.

d. Relations without any constraints (tautologies) are potentially also a part of static constraints of a system model interacting with dynamic constraints from an environment.

PARALLEL COLLIGATION

Initially, relations are joined pairwise using an approach as described in published patent documents WO 09948031A1 and WO 2001022278A1. Moreover, isolated variables (namely, variables only present in their corresponding single relations) are separated and linked into new relations. A trivial case of parallel colligation is to join all relations into a single relation (wherein such an approach is suitable for smaller problems) or into a tree structure of interconnected relations with isolated variables (wherein such an approach is suitable for larger problems), and the colligation is thus thereby completed. In respect of aforementioned larger problems, it is not potentially feasible to use known joining methods due to a size of the joined result arising from such joining methods. It is thereby beneficial to introduce a parallel colligation of smaller parts of the system, wherein:

(I) a parallel join of a relation pair is performed (ii) a parallel colligation of variable groups is performed

PARALLEL JOIN OF RELATION PAIRS

In an internal array representation in compressed form relations are represented by 5 and 3 Cartesian arguments. Such small relations are susceptible to being joined in different ways. However, in respect of large relations, such an approach would cause a combinatorial explosion of possible argument intersections, which would be very expensive in terms of central processing unit (CPU) resources and data memory to compute in a practical example. Thus, pursuant to embodiments of the present disclosure, it is therefore beneficial to use a much more efficient methodology for colligating a smallest possible subsystem spanned by just a single variable step of a join algorithm as a result of expanding the local intersections of each variable to the matching indices of arguments in the joined relation. This indexing procedure is highly efficient and does not benefit from being implemented by employing parallel data processing. Thus, it will be appreciated from the foregoing that the number of arguments in the joined result, and data memory requirements, for computing and storing results, can be predicted from said indexing procedure. The local results of each colligated variable are now expanded to the attributes of the joined relation using the associated indices.

COLLIGATION STRATEGY FOR VARIABLE GROUPS

Initially, relations are joined and compressed pairwise using an approach as described in published patent documents WO 09948031A1 and WO 2001022278A1 (namely, as per Step 1 in the foregoing). Isolated variables (only present in a single relation) are separated and linked into new relations. A trivial case is to join all relations into a single relation (suitable for small problems) or the tree structure of connected relations with isolated variables (suitable for larger problems), and the colligation is thus thereby completed. On large problems, it is not potentially feasible to join all relations due to the size of the joined result. It is thereby beneficial to introduce the colligation of relations on selected variable groups.

In an embodiment, the number of Cartesian arguments in the relations is very large, and it is not possible to join the relations. A corresponding workflow for colligation in respect of groups of variables shared by same given relations is:

Step 1: Determine distinct variable groups shared by two or more relations: all variables shared by same relations are grouped. An aim in the step 1 is to find distinct groups (namely, with no overlap), and therefore there is performed a merging of the small group into the larger one.

Step 2: Split relations on each variable group: all relations share the variable group. A copy of the relations on these variables and the associated link variables is made.

Step 3 and 4: Join and link relations on each variable group: joining the relations on the variable group yields a relation with the following variables. Next, the variable group is isolated and a new link variable indexing each Cartesian argument is thereafter added. There is thereby generated a result that is a relation.

Step 5: Substitute variable groups in original relations with the associated link variables. The relation defines the relationship between variables.

Step 6: Colligate relations on link variables. The original relations are now defined on the link variables of isolated relations. These results are also colligated by join and, if possible, to isolate variables. Assuming that it is required to join to provide a single relation, there is thereby provided a relation.

MINIMIZE AND LINK COMPLETE SOLUTION SPACE

Furthermore, there is thereby now completed the colligation process yielding the addressable complete solution space. All invalid combinations are eliminated (with a state of contradiction as a special case). A final task is to prepare the model for embedded applications, namely to seek to minimize a size of the binary file (to achieve compactness) and to optimize a run-time performance in respect of specific hardware, whether with or without parallel processing capabilities, for example multi-core GPUs are susceptible of providing parallel processing functionality. Each individual relation is potentially split into more relations in two different ways, depending upon a size of an output to be generated and upon whether or not there is use made of parallel processing hardware.

Option 1: Split core relation into pairs and split model for parallel processing A given relation is extended with a link variable (LINK) indexing the Cartesian arguments (in a compressed form) or tuples (in an expanded form) of the given relation with variables VARI, VAR.2, ... VARn. The given relation is then split into n derived relations on (VARI, LINK), (VAR2, LINK), (VARn, LINK), respectively; n is an integer of value 2 or greater (namely, a plurality). Such a method will always be used on a core relation of a complete array system model, whenever the model is to be split and distributed for parallel processing. In a runtime environment, it is thereby feasible to ensure completeness of deduction by a simple state propagation of a state vector. Option 2: Split relation into tree structure of interconnected relations: the original relation is split into a tree structure of relations (represented in bold boxes). There is employed a method as follows:

Step 1: Find smallest derived relation on N variables in the Step 1, the smallest number of Cartesian arguments (or, alternatively, tuples in expanded form) is on the variables.

Step 2, 3: Add new link variable and isolate relation in the Steps 2 and

3. The relation on variables is extended with a link variable and then isolated (namely stored) for the binary output file.

Step 4: Update relation R: remove VARI, VAR2 and substitute with link variable in the method. The aforementioned Steps 1 to 4 of the method are executed recursively to yield a list of relations and finally a relation which is not split (namely, representing a root of the aforementioned tree).

BUILDING DECISION SUPPORT SYSTEMS

The construction of the Array System Model and Decision Support Application is a multi-stage process involving the following steps:

Step 1 - Mining of Source Data and Semantic Normalization:

Step 2 - Compilation & Validation of Array System Model

Step3 - Accessing Array System Model on mobile/wearable device via Runtime API using the User's Input State Vector.

The array system model is converted by the Array System Model compilerinto a verified and normalized structure, that can be represented in a 428 KB file, which is an amount of memory the model consumes when loaded into an Array Runtime API on a given user's mobile device, for example a smart phone or a smart watch. A significant proportion of this memory (namely, over 60% thereof) is simply used for storing names of drugs being considered in the computation, as well as diseases and foods; such data is potentially further optimized, if necessary, so that the Array System Model requires even less computing resources in operation. The Array System Model provides an analytical and predictive substrate to power a personalized decision support app (namely, application software) on the given user's mobile device. This substrate enables the Runtime API running directly on the user's mobile device (smart watch, phone, or tablet) to use the Array System Model to perform logical inferences on the data and deduce all the consequences of a given parameter for a selected data set. Embodiments of the present disclosure are operable to provide a decision support system for performing aforementioned analyses within a predictable and very short time; for example, a proprietary Google Nexus 7 tablet computer running an Android software platform is capable of implementing analyses within five to ten milliseconds. Such computational performance is provided with a constant and low memory footprint (namely, around 430kBytes in practice), and is guaranteed to find all the potential adverse consequences given by constraints imposed by a given user's input state vector.

Optionally, the computing arrangement includes at least one of: a computing device and a distributed arrangement including a plurality of computing devices.

Optionally, the sub-models are distributed over a plurality of computing devices that are mutually coupled together in operation via a data communication network.

Optionally, the method includes generating and storing, in a data memory or data storage medium of the computing arrangement, an addressable solution space defining all valid transitions between all valid states.

Optionally, the method includes computing the state of the entire system model in real-time by consulting one or more sub-systems and/or relations at a time by deducing possible states of each variable and propagating one or more bound link variables to connected one or more relations until no further constraints can be added to the state vectors.

Optionally, the control apparatus is configured to be employable for controlling one or more of:

(i) industrial production facilities;

(ii) agricultural production facilities;

(iii) healthcare providing facilities;

(iv) drug discovery systems;

(v) smart metering arrangements;

(vi) autonomous and self-drive vehicle driving arrangements;

(vii) in intelligent drones for surveillance use;

Cviii) in airborne radar systems; and (ix) in intelligent apparatus for assisting medical surgery and/or treatment.

Claims

1. An Internet® (Io T)-based system that provides individualized support to livestock animals in a farming environment, wherein the Internet-based system includes a data processing arrangement that receives in operation temporally logged sensor signals from a sensor arrangement that is spatially distributed within the farming environment, wherein the data processing arrangement includes an array model of the data processing arrangement against which the temporally logged sensor signals are compared, wherein the data processing arrangement provides output signals that control operation of the farming environment, and wherein the data processing arrangement executes a software product that in execution analyzes the temporally logged sensor signals in respect of the array model and generates the output signals, characterized in that:

(a) the software product is configured to perform a multi-dimensional solution search in the array model implemented as a multi-dimensional array model, the search being based on at least a subset of the temporally logged sensor signals and a genotype determination by DNA sequencing of each animal hosted within the farming environment;

2. An Internet®-based system of claim 1, characterized in that the SNP data includes single nucleotide polymorphisms characterizing each animal determined by using Polymerase Chain Reaction (PCR) to read genetic tissue samples derived for each animal.

3. An Internet®-based system of claim 1 or 2, characterized in that the sensor arrangement includes a plurality of sensors that are spatially distributed within the farming environment and are coupled in communication with the data processing arrangement by using a wireless dynamically reconfigurable communication network.

4. An Internet®-based system of claim 3, characterized in that the sensor arrangement includes tags that are attached to the livestock animals and that monitor and log environmental conditions experienced by each animal and movement performed by each animal.

5. An Internet®-based system of claim 3 or 4, characterized in that the wireless dynamically reconfigurable communication network is implemented as a peer-to-peer network.

6. An Internet®-based system of any one of the preceding claims, characterized in that the Internet®-based system collects in operation one or more pathogens present in the farming environment, genotype sequences the one or more pathogens to characterize the one or more pathogen, and employs the characterization of the one or more pathogens as an input parameter to the software product executed in the data processing arrangement to use in performing its search.

7. An Internet®-based system of any one of the preceding claims characterized in that the output signals are used to control at least one of:

(i) a type and/or a quantity of feed provided to the animals;

(ii) a time when feed is provided to the animals;

(ill) additional feed supplements and/or one or more drugs to be administered to the animals;

8. An Internet®-based system of claim 7, characterized in that the Internet®-based system includes an avatar testing arrangement including a test animal into which genotype information from a given animal is imported, and combinations of drugs and/or feed supplements are tested to determine whether or not toxicological problems are likely to arise if a given combination of drugs and/or feed supplements are administered to the given animal.

9. An Internet®-based system of claim 8, characterized in that the avatar testing arrangement uses the test animal implemented as a fruit fly.

10. An Internet®-based system of any one of the preceding claims, characterized in that the sensor arrangement includes a processing module that pre-processes the temporally logged sensor signals, using at least one artificial intelligence algorithm.

11. A method of (for) using an Internet® (Io 7)-based system that provides individualized support to livestock animals in a farming environment, wherein the Internet-based system includes a data processing arrangement that receives in operation temporally logged sensor signals from a sensor arrangement that is spatially distributed within the farming environment, wherein the data processing arrangement includes an array model of the data processing arrangement against which the temporally logged sensor signals are compared, wherein the data processing arrangement provides output signals that control operation of the farming environment, and wherein the data processing arrangement executes a software product that in execution analyzes the temporally logged sensor signals in respect of the array model and generates the output signals, characterized in that the method includes:

(a) arranging for the software product to perform a multi-dimensional solution search in the array model implemented as a multi-dimensional array model, the search being based on at least a subset of the temporally logged sensor signals and a genotype determination by DNA sequencing of each animal hosted within the farming environment;

12. A method of claim 11, characterized in that the method includes arranging for the SNP data to include single nucleotide polymorphisms characterizing each animal, determined by using Polymerase Chain Reaction (PCR) to read genetic tissue samples derived for each animal.

13. A method of claim 11 or 12, characterized in that the method includes arranging for the sensor arrangement to include a plurality of sensors that are spatially distributed within the farming environment and are coupled in communication with the data processing arrangement by using a wireless dynamically reconfigurable communication network.

14. A method of claim 13, characterized in that the method includes arranging for the sensor arrangement to include tags that are attached to the livestock animals and that monitor and log environmental conditions experienced by each animal and movement performed by each animal.

15. A method of claim 13 or 14, characterized in that the method includes arranging for the wireless dynamically reconfigurable communication network to be implemented as a peer-to-peer network.

16. A method of any one of claims 11 to 15, characterized in that the method includes arranging for the Internet®-based system to collect in operation one or more pathogens present in the farming environment, perform genotype sequencing of the one or more pathogens to characterize the one or more pathogens, and employ the characterization of the one or more pathogens as an input parameter to the software product executed in the data processing arrangement to use in performing its search.

17. A method of any one of claims 11 to 16, characterized in that the method includes using the output signals to control at least one of:

(I) a type and/or a quantity of feed provided to the animals;

(ii) a time when feed is provided to the animals;

18. A method of claim 17, characterized in that the method includes arranging for the Internet®-based system to include an avatar testing arrangement including a test animal into which genotype information from a given animal is imported, and combinations of drugs and/or feed supplements are tested to determine whether or not toxicological problems are likely to arise if a given combination of drugs and/or feed supplements are administered to the given animal.

19. A method of claim 18, characterized in that the method includes arranging for the avatar testing arrangement to use the test animal implemented as a fruit fly.

20. A method of any one of claims 11 to 19, characterized in that the method includes arranging for the sensor arrangement to include a processing module that pre-processes the temporally logged sensor signals, using at least one artificial intelligence algorithm.

21. A computer program product comprising a non-transitory computerreadable storage medium having computer-readable instructions stored thereon, the computer-readable instructions being executable by a computerized device comprising processing hardware to execute the method of any one of claims 11 to 20.