Received: 14 September 2023

DOI: 10.1002/aro2.44

REVIEW
- -
Revised: 8 November 2023 Accepted: 28 November 2023

Advancements in artificial intelligence technology for

improving animal welfare: Current applications and
research progress

Li Zhang1 | Wenqiang Guo1 | Chenrui Lv1 | Meng Guo1 | Mei Yang2 |

Qiuyue Fu1 | Xiaomeng Liu1

1
Department of Animal Nutrition and Feed
Science, College of Animal Science and Abstract
Technology, Huazhong Agricultural University, The integration of Artificial Intelligence (AI) in various sectors has led to
Wuhan, China
2
significant advancements, with the animal industry being no exception. This
Southwest University of Science and
Technology, Mianyang, China
review aims to investigate the benefits, limitations, and future prospects of
AI technology in improving animal welfare. First, it examines the role of AI in
Correspondence understanding animal behaviors and emotions, providing deeper insights
Li Zhang. into their well‐being and sources of stress. Next, the paper explores how AI
Email: [email protected] can revolutionize animal nutrition through innovative algorithms and data
analytics. The health aspect emphasizes the ability of AI to identify and
manage illnesses through intelligent systems. This review also highlights
the application of AI in improving animal living conditions, with a focus on
environmental management and automated cleaning and disinfection sys-
tems. In conclusion, the review emphasizes AI‐driven techniques for early
prediction, close monitoring, and accurate diagnosis of animal diseases,
ensuring healthier and more sustainable livestock management. By
leveraging its advantages, addressing limitations, and exploring future di-
rections, AI has the potential to significantly enhance animal welfare, sus-
tainable agriculture, and veterinary practices.

Keywords
animal behavior recognition, animal emotion recognition, animal welfare, artificial intelligence,
intelligent health systems

INTRODUCTION moral obligations associated with our treatment of ani-

mals [1–3]. Whether they are farm animals, companion
Animals play a crucial role in our ecosystem and coexist pets, wildlife, or laboratory subjects, ensuring their
with us, forming an intricate web of life [1]. As sentient welfare has become an imperative in our modern times
beings, they deserve not only respect but also our [4, 5].
vigilant attention to their welfare. The importance of Concurrently, the field of Artificial Intelligence (AI)
animal welfare has gained widespread recognition in has witnessed unprecedented advancements, encom-
recent years as we grapple with the ethical and passing machine learning, deep learning, and neural

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided

-
the original work is properly cited.
© 2023 The Authors. Animal Research and One Health published by John Wiley & Sons Australia, Ltd on behalf of Institute of Animal Science, Chinese Academy of
Agricultural Sciences.

Anim Res One Health. 2024;2:93–109. wileyonlinelibrary.com/journal/aro2 93

94
- ZHANG ET AL.

networks, demonstrating remarkable potential in researchers can enhance the accuracy and efficiency
numerous domains [6, 7]. One of the most promising and of individual identification in animals [10]. This tech-
ethically significant applications of AI is its integration nology holds promise for various applications in con-
into the realm of animal welfare [8]. By leveraging AI servation, research, and wildlife management,
technologies, we have the opportunity to revolutionize ultimately contributing to the well‐being and conserva-
the monitoring, management, and enhancement of ani- tion of animal populations.
mal lives across various contexts [9]. In addition to visual algorithms, collecting relevant
This comprehensive review aims to delve into the data on animal behavior through sensors and training
synergistic relationship between AI and animal welfare, AI algorithms based on that data is also commonly used
shedding light on the multifaceted applications of AI in to improve animal welfare and farm management effi-
this field. Specifically, we explore how AI is employed to ciency [39]. B. Ji et al. developed a machine learning
address key aspects of animal welfare, including framework by collecting behavior, health, and produc-
nutrition, habitat improvement, health care, and tivity data from 80 cows on a dairy farm over a period of
emotional recognition (refer to Table 1 and Figure 1). 5 years (Figure 2A.) [11]. They predicted daily milk
By elucidating these applications, we intend to em- production, composition (fat and protein content), and
phasize the profound significance of AI in reshaping our milking frequency for each cow over the subsequent
approach to animal welfare and fostering a more 28 days. The experimental results demonstrated an
compassionate and informed coexistence between accuracy rate of over 80% for this learning framework.
humans and animals. Through this exploration, we The authors aim to integrate this framework with robotic
aspire to inspire further research and innovation, ulti- milking systems to determine healthier milking strate-
mately advancing the cause of animal welfare in our gies for cows. By leveraging sensor data and machine
ever‐changing world. learning, researchers can gain valuable insights into
animal behavior and well‐being, leading to improved
management practices and enhanced animal welfare
THE IMPACT OF AI TECHNOLOGY ON [39]. These technological advancements offer prom-
ANIMAL BEHAVIOR AND EMOTION ising possibilities for optimizing various aspects of ani-
RECOGNITION mal care and husbandry.
In challenging environments, accurately identifying
Animal behavior recognition animal behavior using visual recognition and activity
sensors can be difficult [40]. However, Manteuffel
AI technology is revolutionizing our ability to recognize addressed this issue by employing radar sensors,
and understand animal behavior, offering numerous which proved effective in recognizing parturition‐related
benefits for animal welfare, conservation, research, and activities in sows (Figure 2B) [12]. By analyzing the data
farm management [34]. Traditional methods of visual from 15 sows and utilizing techniques such as daily
recognition have limitations, but AI‐powered solutions activity differences and accumulation, a simple classi-
enhance accuracy and efficiency in identifying individ- fier was implemented to detect deviations in activity
ual animals and monitoring their behavior [35]. These levels. The classifier exhibited a sensitivity of 92%, a
advancements hold great promise for improving animal specificity of 80%, and an area under the receiver
well‐being and the management of animal populations operating characteristic curve of 95%. This novel
[36]. approach achieved the classification and identification
While visual recognition has been widely used for of animals in different states through the analysis of
animal identification, it often requires human assistance daily activity level differences.
and has limitations in terms of accuracy and long‐term In the field of livestock farming, future poultry pro-
recognition [37, 38]. Recently, there has been a focus duction could potentially optimize operations through
on improving recognition accuracy and efficiency the use of intelligent autonomous systems, remotely
through the introduction of AI. Efficient individual managed by human workers and intervening only when
recognition can help quickly obtain individual informa- necessary [41]. Park et al. proposed a system that
tion to ensure animal welfare. In a recent study, deep effectively ensures the welfare of poultry animals [13].
learning techniques were applied to develop a novel This system involves sensors and robots for round‐the‐
face recognition model based on convolutional neural clock management of the poultry house. A multifunc-
networks (CNNs) for identifying individual giant pandas tional robot performs most of the daily tasks, such as
[10]. The model successfully recognized 95% of the removing dead chickens and monitoring flock behavior
giant panda individuals in the validation dataset. Even to ensure the growth and welfare of the chickens.
under simulated field conditions with decreased photo The livestock density is a crucial factor that affects
data quality, the model still accurately identified large‐scale animal production and animal welfare [42].
over 90% of giant panda individuals. By leveraging However, the current artificial counting methods used in
deep learning and advanced recognition models, the Muscovy duck farming industry are inefficient,
TABLE 1 Specific applications of AI in improving animal welfare.

Field of application Function Reference

The impact of AI technology on animal behavior and Animal behavior recognition A novel face recognition model based on convolutional [10]
emotion recognition neural networks for recognizing individual giant pandas
was developed by applying deep learning techniques.
Predicting a cow's daily milk yield, composition, and milking [11]
frequency by utilizing sensor data and machine learning.
Identifying farrowing‐related activity in sows using radar [12]
sensors.
Intelligent robotic system applied to round‐the‐clock [13]
management of poultry houses.
Smart duck counting via AI detection of image data. [14]
Animal emotion recognition Analyzing domestic shorthair cat facial image data by [15]
machine learning models for automated pain recognition
CURRENT APPLICATIONS AND RESEARCH PROGRESS

in domestic cats.
Intelligent detection of equine pain signals and diagnosis by [16]
training vision algorithms with automatic computational
classifiers via machine learning models.

AI technology ensuring animal nutrition and health AI revolution in animal nutrition An intelligent learning model for collecting cow feed intake in [17]
automatic feeding systems to develop feeding
schedules.
An intelligent feeding system for controlling pet obesity. [18]
Advancing animal health through intelligent systems Predicting learning models for cow eye temperature, milk [19]
yield, and quality.
An intelligent diagnostic system for equine lameness. [20]
Intelligent systems for animal health monitoring, animal [9]
tracking, milk quality and supply chain, and feed
monitoring and safety.
A machine learning model for a noninvasive video biometric [21]
system for intelligent monitoring of cow health.

The role of AI in revolutionizing animal living The revolutionary application of AI in animal living Intelligent real‐time forecasts for CO2, SO2, NO2, PM2.5, [22]
environments environment management and PM10 levels.
Monitoring system for temperature, carbon dioxide, and [23]
wind speed in layer coops.
Automatic monitoring and an adjustment system for [24]
livestock and poultry farming environment based on
(Continues)
-
95
 96
-

T A B L E 1 (Continued)

Field of application Function Reference

WiFi, LoRa wireless communication, and IoT

technologies.
The application of automated cleaning and disinfection Rapid disinfection strategy for livestock farms based on [25]
systems in animal facilities remote control of intelligent robots.
Magnetic RFID‐based navigation‐guided intelligent robots [26]
target the full range of disinfection strategies for large
livestock farms.

Advancements in AI‐driven animal disease prediction, Animal disease prediction and monitoring A fog‐centered, IoT‐based smart health care support system [27]
monitoring, and diagnosis to monitor and control swine flu virus outbreaks.
A fog computing‐assisted end‐to‐end animal behavioral [28]
analysis and disease monitoring platform for remote
farms with limited Internet connectivity.
An IoT‐based “livecare” framework for automated [29]
monitoring of the health of cows on large dairy farms.
An IoT‐based animal social behavior sensing framework for [30]
modeling the spread of mastitis in dairy cows and
inferring the risk of mastitis infection in dairy cows.
Diagnosis of animal diseases An automated swine welfare indicator assessment system [31]
for the diagnosis of ear and tail lesions in slaughterhouse
pigs.
Convolutional neural networks to improve the diagnostic [32]
accuracy of expert systems for fast and concise animal
disease diagnosis.
Computational models developed using penalized linear [33]
regression, random forests, gradient boosters, and
neural networks for helping farmers take evidence‐
based interventions before anticipated stressful
environmental conditions occur.
An artificial intelligence‐based contactless biometric system [3]
for automated assessment of animal welfare on farms
and during transportation.
ZHANG
ET AL.
CURRENT APPLICATIONS AND RESEARCH PROGRESS
- 97

F I G U R E 1 The application of AI in
animal welfare.

F I G U R E 2 The impact of AI technology on animal behavior and emotion recognition. (A) This machine learning framework is integrated
with a robotic milking system to predict the milk production, milking frequency, and milk quality of each cow, thus determining healthier milking
strategies for cows [11]. (B) By collecting daily activity variations of sows through radar sensors, it enables monitoring of increased sow activity
and prediction of sow parturition [12]. (C) The Horse Grimace Scale monitors pain responses in horses by analyzing their ear, eye, and muzzle
features [16].
98
- ZHANG ET AL.

labor‐intensive, less accurate, and prone to repetitive this issue, the present study designed an emotion
counting and omissions. In a study by K. Jiang et al., recognition system for dogs [53]. Specifically, they
they addressed this issue by incorporating an attention initially trained this model using 13,809 annotated im-
mechanism into the deep learning algorithm YOLOv7, ages of canines. By observing limb movements such as
resulting in an improved version called CBAM‐YOLOv7 the angle between the tail and spine, the height of the
[14]. The experimental results demonstrate that CBAM‐ head lift, and the curvature of the limbs, they were able to
YOLOv7 achieves higher accuracy and slightly im- determine the emotional state of the dogs. Subse-
proves the recall rates, [email protected] and [email protected]:0.95, quently, they validated the model using 400 full‐body
compared to SE‐YOLOv7 and ECA‐YOLOv7, respec- images of dogs representing four emotions: anger,
tively. The evaluation metrics of CBAM‐YOLOv7 show fear, happiness, and relaxation. The research findings
greater improvement. This indicates that the proposed demonstrated that this emotion recognition approach
intelligent duck counting method in this study is feasible could intuitively identify different emotional states in
and can contribute to the development of intelligent and dogs, achieving an accuracy rate ranging from 60% to
reliable automated duck counting. This model enables 70%. This effectively safeguards the animal welfare of
real‐time monitoring of the population density of Mus- canines during training [53].
covy ducks in dense groups, thereby promoting the Visual algorithms have shown potential not only in
development of the smart farming industry. pets but also in assessing the degree of pain in horses.
Lencioni et al. developed a visual algorithm based on
the automatic calculation classifier of the Horse
Animal emotion recognition Grimace Scale (HGS) (Figure 2C) [16]. By training the
algorithm using machine learning techniques, they
AI applied to animal emotion recognition can provide were able to monitor the pain responses of horses more
valuable insights and methods for animal behavior accurately and in real time. This provides earlier diag-
research and welfare [43]. By utilizing advanced image nosis and more effective treatment for affected animals.
processing and machine learning techniques, we can When categorizing pain into three levels: absent,
accurately analyze and identify the emotional states of moderate, and obvious, the overall accuracy was
animals [5]. This application helps us better understand 75.8%. When classifying between two categories, pain
their emotional experiences and needs, thereby absent and pain present, the overall accuracy reached
improving their quality of life and welfare [44]. Moreover, 88.3%.
it can play a crucial role in animal conservation by
helping monitor their health, identify potential disease
risks, and take necessary intervention measures in a AI TECHNOLOGY ENSURING ANIMAL
timely manner [45]. Through the application of AI in an- NUTRITION AND HEALTH
imal emotion recognition, we can comprehensively focus
on and protect the emotional needs of animals [46]. AI revolution in animal nutrition
Research on determining the emotional states of
animals has been a topic of great interest among ani- AI has revolutionized animal nutrition by enabling
mal welfare researchers [47]. One study suggests that practical applications that have transformed the man-
the internal emotional states of domestic cats can be agement and optimization of animal diets [17]. One
captured through facial morphological changes asso- notable application involves the development of AI‐
ciated with pain [15]. By employing a machine learning driven feeding systems. These systems utilize AI al-
model using CNNs (ResNet50) and geometric land- gorithms to accurately compute and dispense precise
mark analysis based on a specific species' facial action quantities of feed tailored to the specific nutritional re-
coding system (i.e., catFACS), researchers analyzed quirements of individual animals [54]. By analyzing
facial images of domestic shorthair cats at different time data such as animal weight, body condition, and
points corresponding to varying levels of pain during growth rate, these algorithms generate customized
ovariohysterectomy. The automatic recognition of pain feeding plans [55]. These plans not only promote
achieved an accuracy rate of over 72%, indicating the optimal growth and health but also reduce waste and
potential application of this intelligent recognition strat- mitigate the risks associated with overfeeding or un-
egy in animal emotion recognition. derfeeding [56].
Dogs, as significant companion animals, typically The accurate collection of feed intake data is crucial
require training to adapt to various environments. During in the creation of an automated feeding system [57].
the training process, the emotional state of the canines Analyzing this data enables meticulous statistical
plays a crucial role in determining the effectiveness of the analysis, which is currently lacking in most commercial
training [48–50]. Excessive punitive training not only dairy farms [58]. One recommended approach sug-
leads to decreased canine morale but also severely gests using visual algorithms in conjunction with RGBD
compromises their animal welfare [51, 52]. To address cameras. These cameras capture red, green, blue, and
CURRENT APPLICATIONS AND RESEARCH PROGRESS
- 99

depth information to impeccably gather individual feed driven intelligent feeding systems. It also created a vir-
intake data [17]. The author has developed a transfer tual database incorporating various variables such as
learning (TL) model based on CNNs. This model in- the pet's growth, temperature, heart rate, eating habits,
corporates a CNN model trained for two feed types, a activity, sleep, and urine pH level. Researchers trained
multilayer perceptron, and a CNN model trained for two and validated supervised machine learning models,
feed types along with classification data (Figure 3A). including linear regression, Gaussian process regres-
These models collaborate with the RGBD camera to sion, narrow neural networks, linear support vector
analyze a series of images capturing the weight of machines, and fine trees, using this database. A com-
diverse feed types collected under diverse configura- parison of different models revealed that the interactive
tions and lighting conditions within a research‐based linear regression model exhibited the best efficiency in
dairy farm. Rigorous experimentation has demon- analyzing animal welfare‐related data. Further integra-
strated the effectiveness of the TL model, resulting in a tion of this model into hardware and the IoT can lead to
maximum error of 0.14 kg per meal and a root mean the development of smart feeding products that address
square error of 0.19 kg [17]. This model holds great issues such as pet obesity, thus safeguarding the wel-
potential for accurately quantifying individual feed fare and safety of companion animals [18].
intake in barn‐kept cows, providing invaluable guidance
for tailored feeding plans.
With the rapid development of artificial intelligence, Advancing animal health through
Internet of Things (IoT), and digital information pro- intelligent systems
cessing technologies, pet feeding systems are no longer
limited to simple adjustments in food quantity [59, 60]. AI is revolutionizing our comprehension of animal wel-
Nowadays, an increasing number of pet feeding devices fare and playing a crucial role in monitoring and main-
focus on the animal welfare of companion animals like taining of animal health [61]. Noninvasive AI‐based
cats and dogs. One study developed an automatic detection techniques curtail human intervention while
model for estimating pet food amounts [18]. This model proficiently and precisely amassing data, thus estab-
analyzed and compared different machine learning al- lishing a novel benchmark for augmenting animal wel-
gorithms used in pet care applications by utilizing data‐ fare [62].

F I G U R E 3 AI technology ensuring animal nutrition and health. (A) A transfer learning model based on convolutional neural networks is
used to determine the feeding behavior of cows using images from within the dairy farm [17]. (B) A computer vision approach utilizing images
and infrared thermal videos is employed to predict the eye temperature of cows, milk yield, and milk quality, enabling health monitoring of
cows [19]. (C) AI interprets the limb movements of horses through visual images to distinguish between healthy and lame horses [20]. (D) An
intelligent ecosystem combining technologies such as smart systems, Internet of Things, and blockchain is implemented to promote the dairy
industry and improve animal welfare [9].
100
- ZHANG ET AL.

In the field of animal welfare, noninvasive methods Additionally, the design utilizes gravity sensors to
assume paramount importance for their capacity to automate feeding functions, monitor weight, and ensure
minimize human intrusion while efficiently and accu- regular removal of pet waste [65].
rately acquiring data. Fuentes et al. have pioneered a The overall health status of companion animals is a
noninvasive computer vision approach employing RGB top concern for pet owners. However, most people lack
cameras and infrared thermal videos (IRTV) for data knowledge about pet‐related diseases beyond pro-
acquisition (Figure 3B) [19]. They have also employed fessionals [66]. As a result, the behaviors of many
machine learning models to predict eye temperature, companion animals often hint at potential diseases or
milk yield, and quality. In their study involving 102 dairy fragile health conditions, but owners are unable to take
cows, they demonstrated the model's remarkable ac- timely and effective measures [67]. To address this
curacy (R = 0.96) and its potential for application in issue, a study proposed an integrated framework for an
both animal welfare and the modeling of biological and intelligent pet care system to monitor the health con-
non‐biological pressures. ditions of household pets [68]. The framework utilizes
Feuser et al. have developed a diagnostic software Internet of Medical Things technologies and remote
tool that leverages AI technology to promptly identify monitoring devices such as cameras and electrocar-
lameness in horses and mitigate its impact on their well‐ diogram sensors to gather health information about the
being (Figure 3C) [20]. This software can differentiate pets. This information is then transmitted through the
between lame and healthy horses while identifying the FogBus fog computing platform and analyzed using a
affected limbs. By analyzing 22 horses with unilateral time series artificial neural network predictive model to
forelimb or hindlimb lameness and comparing them to a measure real‐time health vulnerability probabilities
control group of eight healthy horses, they have (PoHV) and time‐sensitive measurements (TSM) for
developed a program based on pose estimation household pets. PoHV is used to assess the health
capable of discerning different types of lameness and status index of animals, while TSM provides analytical
healthy horses. This technology holds promise for results on the development and health conditions of
studying limb and body movement in both human and animal diseases over time and space. Finally, the data
veterinary sciences without requiring any attachments are visualized and presented to caregivers and pet
to the subjects. owners using self‐organizing mapping technology [68].
Gehlot et al. have explored the potential of AI tech- The system has been experimentally validated on
nology to significantly enhance animal welfare in dairy American bulldogs, achieving high accuracy, precision,
cows (Figure 3D) [9]. They have discussed intelligent sensitivity, and f1‐measure values all above 93% for
systems that incorporate sensory, monitoring, and disease monitoring.
analytical techniques for diverse applications, including The development of noninvasive video‐based bio-
animal health monitoring, animal tracking, milk quality metric systems and the application of machine learning
and supply chain, feed monitoring, and safety. Further- techniques have opened up possibilities for compre-
more, they have proposed a flexible architecture hensive animal health monitoring and automated wel-
adaptable to research and development advancements fare assessment [69]. Recent studies have mentioned
in data collection, in addition to applications involving the extraction of data from cows through noninvasive
edge devices, robots, drones, and blockchain, thus video collection and the use of biostatistical analysis
establishing intelligent ecosystems [9]. Their developed techniques to develop two machine learning models
system, bolstered by technologies like the IoT, facilitates [21]. The first model focuses on biostatistical regression
the creation of a smart ecosystem for dairy farming, for parameters such as somatic cell count, body weight,
making noteworthy contributions to improving animal rumination, and feed intake. The second model utilizes
welfare in the dairy industry. facial feature mapping to predict the age of the animals.
The rapid economic development in China has led These models were tested in a robotic dairy farm, and
to a significant growth in the pet market [63]. However, the results showed high accuracy and low error rates,
challenges remain regarding welfare standards and indicating potential applications in evaluating milk pro-
care strategies for companion animals in China [64]. ductivity, quality traits, and welfare in dairy farms.
There is an urgent need for improved products and
policies to enhance conditions for pets. In response to
this, Xu et al. proposed a design for an intelligent collar THE ROLE OF AI IN REVOLUTIONIZING
and accompanying facilities [65]. This design integrates ANIMAL LIVING ENVIRONMENTS
the collar with Alibaba Cloud services platforms and the
MQTT IoT access protocol to enable real‐time collec- The revolutionary application of AI in
tion of pet activity levels, heart rate, and location data. animal living environment management
The collar incorporates sensors such as a JY901 gy-
roscope, MAX30102 heart rate sensor, and Antrack- AI has revolutionized livestock and poultry manage-
erA9G positioning system to gather this information. ment, exerting a significant influence on animal health
CURRENT APPLICATIONS AND RESEARCH PROGRESS
- 101

and productivity. Environmental parameters, such as the actuators inside the housing, enabling automatic
temperature, humidity, and air quality, play crucial roles monitoring and adjustment of the livestock and poultry
in these settings [70]. AI‐driven environmental moni- farming environment. The system management plat-
toring systems utilize sensors and data analytics to form is built using Java and C# programming lan-
continuously assess these conditions [71]. When de- guages, providing comprehensive management and
viations from the optimal range occur, AI systems can control of multiple zones within the housing. The au-
automatically adjust ventilation, heating, or cooling pa- thors aim to ensure animal welfare throughout the
rameters to create a more comfortable and healthy farming process while simultaneously maintaining the
environment for the animals [13]. This not only en- quality of animal growth and economic benefits.
hances animal welfare but also improves production
outcomes by reducing stress‐related issues and dis-
ease prevalence. The application of automated cleaning
Previously, environmental monitoring in animal and disinfection systems in animal
farming mainly focused on toxic gases like ammonia facilities
[72]. However, in response to growing concerns about
animal welfare, monitoring methodologies for gases Automated cleaning and disinfection systems, often
such as carbon dioxide and sulfur dioxide have utilizing robotic technology, have become essential for
emerged. These approaches aim to prevent livestock maintaining sanitary conditions in livestock and poultry
from inhabiting unhealthy environments. A research facilities [73]. AI‐driven robots, equipped with sensors
study introduces a method for constructing a model and cameras, efficiently navigate the environment and
capable of predicting four distinct gases: carbon diox- identify areas requiring cleaning and disinfection.
ide, sulfur dioxide, nitrogen dioxide, and atmospheric These robots effectively remove organic matter, such
particulate matter [22]. The study combines linear as manure, and accurately apply disinfectants as
regression, support vector regression, and gradient necessary. Automating this critical aspect of animal
boosting decision tree (GBDT) ensemble models to housing not only reduces the workload of farm
formulate an air quality index prediction model. Various personnel but also ensures a consistently clean and
sensors designed for air quality analysis are also inte- pathogen‐free environment [25]. Consequently, it con-
grated. Models with lower errors were selected based tributes to disease prevention, improves animal health,
on performance evaluations. By incorporating algo- and enhances overall hygiene standards.
rithms from AI systems and integrating observational To address the challenge of disinfection in livestock
data to rectify inaccuracies, real‐time forecasting of the farms, a study aims to develop an intelligent strategy by
levels of five air pollutants (carbon dioxide, sulfur di- creating smart robots equipped with disinfection
oxide, nitrogen dioxide, PM2.5, and PM10) can be sprayers (Figure 4B) [25]. This robot consists of an
achieved. This advancement presents opportunities for automated vehicle, a disinfectant spraying unit, a
enhancing the living conditions of animals. monitoring unit, and a controller unit. It supports both
Li and his team have designed an intelligent moni- automatic and remote operation and can follow desig-
toring system specifically for stacked chicken coops nated routes. The use of this automated smart robot
[23]. The sensor distribution topology, tailored to the effectively resolves the issues associated with slow and
complex structure of the henhouse, enables real‐time ineffective manual disinfection, thereby upholding
monitoring of temperature, CO2 levels, and wind overall hygiene standards for livestock.
speed in different layers of the chicken coop. The data Another study adopts a similar approach by
are analyzed and organized on a cloud server. Through employing a robot guided by magnetic‐RFID navigation
a smartphone app, users can instantly check the envi- (Figure 4C) [26]. This robot comprises a mobile platform,
ronmental conditions of the chicken coop. This system a disinfection spraying device, environmental moni-
enables data analysis and transmission from sensors to toring sensors, and a controller. Additionally, it is
monitoring terminals to smart mobile devices, allowing equipped with air‐assisted nozzles capable of indepen-
remote monitoring of the chicken coop environment. It dently atomizing and dispersing the liquid disinfectant.
holds significant practical significance for precision Actual performance tests conducted in poultry houses
breeding in large‐scale livestock farming. demonstrated that the robot autonomously navigates at
Sun et al. have developed an automated monitoring speeds ranging from 0.1 to 0.5 m/s, with a maximum
and control system for livestock and poultry housing deviation of 50.8 mm from the expected trajectory. The
environments using IoT and robotics technology air‐assisted nozzles enable simultaneous atomization
(Figure 4A) [24]. This system utilizes IoT robots and diffusion of the liquid disinfectant, making it suitable
equipped with sensors to collect environmental data for spraying liquid at flow rates of 200–400 ml/min. The
within the animal housing. It leverages WiFi, LoRa resulting droplet diameter (DV.9) varies from 51.82 to
wireless communication, and IoT technology to control 137.23 μL, increasing with the liquid flow rate. The
102
- ZHANG ET AL.

F I G U R E 4 The role of AI in revolutionizing animal living environments. (A) Intelligent robots for automatic detection and control of livestock
farming environments [24]. (B) Remote‐controllable intelligent robots for automatic disinfection [25]. (C) Disinfection robots guided by
magnetic radio‐frequency identification navigation [26].

deposition density of the spray droplets generated by veterinarians to optimize disease prevention strategies
the nozzles ranges from 116 to 149/cm2, decreasing and improve animal welfare [77]. Overall, AI technology
with the spray distance. These parameters of liquid plays a crucial role in safeguarding animal health,
droplet size and density meet the necessary criteria for enhancing disease management practices, and
effectively eliminating adherent pathogenic microor- ensuring sustainable agriculture.
ganisms in different areas of the cages. In a recent study, researchers proposed an intelli-
gent framework for real‐time monitoring of animal dis-
eases, using swine flu as an example [27]. They
ADVANCEMENTS IN AI‐DRIVEN ANIMAL employed a fog‐centric, IoT‐based intelligent health
DISEASE PREDICTION, MONITORING, care support system to monitor and control the
AND DIAGNOSIS outbreak of the swine flu virus. The researchers used
the iFogSim simulator to evaluate various parameters
Animal disease prediction and monitoring such as accuracy, energy consumption, and latency by
simulating IoT devices and fog nodes. Additionally, they
AI technology has brought about a revolution in the utilized WEKA to develop a hybrid classifier. The results
field of animal disease prediction and monitoring. By indicated that the integration of fog computing and
utilizing advanced machine learning algorithms, AI cloud computing services yielded enhanced network
systems can analyze extensive datasets, including bandwidth reliability, elevated operational efficiency,
animal health records, environmental factors, and ge- and reduced response times in comparison to con-
netic information, to identify patterns and accurately ventional cloud computing models. This integration
forecast disease outbreaks [74]. These AI‐powered facilitated real‐time notifications.
solutions enable the early detection of diseases, facil- In another study, a fog computing‐assisted end‐to‐
itating timely intervention and preventing the spread of end animal behavior analysis and disease monitoring
infections [75, 76]. Moreover, AI algorithms continu- platform called SmartHerd was specifically designed for
ously monitor animal health parameters in real time, dairy farms (Figure 5) [28]. The platform relies on
providing valuable insights for farmers and wearable sensors attached to the cows' bodies for data
CURRENT APPLICATIONS AND RESEARCH PROGRESS
- 103

FIGURE 5 A fog computing assisted for animal behavior analysis and disease monitoring platform [28].

collection. Through data classification and analysis on Diagnosis of animal diseases

the platform, the authors aimed to provide farm owners
with valuable insights. To address the main issue of AI technology has emerged as a powerful tool in diag-
limited Internet connectivity in remote farms, they pro- nosing animal diseases, utilizing comprehensive data
posed adopting a microservice‐oriented design and analysis of clinical symptoms, laboratory results, and
distributed computing paradigm. medical imaging [78]. By identifying patterns, AI algo-
Chatterjee et al. proposed a framework known as rithms can accurately diagnose diseases, leading to a
LiveCare, based on the IoT, for automated monitoring significant reduction in diagnostic errors and improved
of the health status of cows in large‐scale dairy farms efficiency in veterinary practices [79]. Moreover, AI‐
[29]. This framework tracks the behavioral changes of based diagnosis enables early disease detection,
cows on a daily basis and introduces the Cow Disease facilitating prompt treatment and better outcomes for
Prediction (CDP) algorithm as a core component. The animals [32]. Continuous learning and improvement in
CDP algorithm analyzes the behavioral changes of AI technology hold immense promise for revolutionizing
cows to predict various cow diseases. animal disease diagnosis, benefiting both animals and
Feng et al. introduced a framework based on the IoT veterinary professionals [80].
for sensing animal social behavior [30]. This framework Blömke and their team designed an automated
was applied to simulate the spread of mastitis in dairy system to assess pig welfare indicators and diagnose
cows and infer the risk of cows being infected with ear and tail lesions in pigs at slaughterhouses [31]. The
mastitis. By deploying portable GPS devices on cows system underwent testing on a sample of 5952 pigs and
and tracking their movement patterns and interactions, was compared with assessments made by skilled vet-
a directed and weighted social behavior graph of cows erinarians. The results demonstrated that the system
was established. The researchers proposed a disease exhibited high sensitivity (above 77%), specificity
transmission probability model that considered both (above 96%), and accuracy (above 95%) in detecting
direct contact with infected cows and indirect contact lesions in both the ears and tails. This finding is sig-
through environmental contamination. This model was nificant for diagnosing pig ear and tail diseases and
used to estimate and predict the probability of mastitis ensuring the welfare of these animals.
infection and help identify infected cows in the dairy AI has proven to be an effective solution for
farm, giving them priority for further screening. The addressing challenges faced by remote farms with
effectiveness of this framework was demonstrated limited diagnostic capabilities. In a recent study, re-
through experiments, and the results aligned with so- searchers proposed the use of a convolutional neural
matic cell count (SCC) mastitis tests. network (CNN) to enhance the diagnostic accuracy of
104
- ZHANG ET AL.

an Animal Disease Diagnosis Expert System (ADDES) RGB videos. The overall fitting rate reached 96%,
for swift and precise identification of animal diseases making it an effective tool for heat stress assessment in
[32]. The study utilized a dataset of 1200 images rep- live sheep.
resenting 15 distinct diseases. By extracting features
from these images and employing the CNN for pattern
recognition, the researchers achieved an impressive DISCUSSION AND FUTURE PROSPECTS
accuracy rate of 97.06%. The autonomous disease
diagnosis system developed through this approach re- AI offers significant advantages for enhancing animal
duces reliance on veterinarians, making it especially welfare [61]. Primarily, AI enables meticulous recogni-
valuable for farms lacking veterinary expertise or facing tion of animal behaviors, emotions, as well as the
challenges in maintaining optimal animal husbandry prognosis and diagnosis of diseases. This level of ac-
practices. By harnessing the power of AI and its ability curacy allows for timely interventions, augmenting the
to analyze extensive amounts of data, including clinical overall welfare of animals. Additionally, AI systems
symptoms, laboratory results, and medical imaging, AI automate tasks including environmental management
algorithms can identify patterns and provide accurate and facility sanitation, minimizing human errors, and
diagnoses. These AI‐powered systems have the po- ensuring uniform quality in animal care [22, 27].
tential to significantly reduce diagnostic errors and Furthermore, AI revolutionizes animal nutrition and
enhance the efficiency of veterinary practices. More- health, formulating individualized nutrition strategies
over, AI‐based diagnosis enables early disease detec- and proactively addressing health concerns before they
tion, allowing for prompt treatment and improved escalate.
outcomes for animals. Through continuous learning However, the application of different AI techniques
and improvement, AI technology holds immense also has certain limitations (Figure 6). For example,
promise for revolutionizing the field of animal disease computer vision‐based methods like CNNs can accu-
diagnosis, ultimately benefiting both animals and vet- rately recognize animals and their behaviors, but may
erinary professionals [32]. struggle in low light or cluttered environments [83, 84].
Heat stress has detrimental effects on the health Sensor‐based approaches efficiently collect behavioral
and productivity of cows. Most current intelligent data, yet interpreting complex emotions remains diffi-
models for heat stress assessment rely on pre- cult [85]. ML techniques effectively analyze nutrition
determined relationships between environmental con- and health parameters, but lack human‐like compre-
ditions and cows' physiological responses, established hension of animal needs [86]. Knowledge‐based expert
by researchers [81, 82]. To overcome potential biases systems enable disease diagnosis, but rely on pre-
caused by human judgment, a study employed ma- defined rules and have limited learning capabilities.
chine learning algorithms to evaluate the impact of Overall, a combination of different AI methods can help
environmental heat stress factors (air temperature, overcome individual limitations and provide more
relative humidity, solar radiation, and wind speed) robust solutions.
which affect cows' physiological responses (respiratory To develop and train AI systems, a substantial
rate, skin temperature, and vaginal temperature) [33]. amount of data are essential [87]. However, in certain
Machine learning algorithms automatically consider the cases, data availability may be limited or difficult to
nonlinearity of the data, eliminating subjectivity in the obtain, posing challenges to the effectiveness of AI
selection of predictor–response relationships. The applications [88, 89]. While AI can effectively monitor
study developed computational models using penalized the physiological indicators of animals, understanding
linear regression, random forest, gradient boosting their emotions and behaviors remains beyond its ca-
machine, and neural networks, which can effectively pabilities. Consequently, there is a risk of failing to
help farmers take evidence‐based interventions before identify crucial welfare issues. Additionally, the suc-
expected stressful environmental conditions occur [33]. cessful implementation of AI technology requires
Early interventions can improve animal health and ongoing training and maintenance, necessitating
provide effective support for ensuring animal welfare. personnel with the relevant knowledge and skills to
Visual algorithms not only enable individual recog- manage and utilize these systems. Unfortunately, the
nition but also find relevant applications in heat stress absence of such knowledge and training can potentially
assessment. S. Fuentes et al. developed an AI‐based impede the adoption of this technology. Implementation
noncontact biometric system for automated animal of AI systems, especially in expansive facilities, can
welfare assessment in farm and transportation pro- incur significant initial expenses, potentially hindering
cesses [3]. The system utilizes an automated tracking adoption by smaller entities or individual proprietors
algorithm to extract skin temperature as a head feature [90]. Moreover, persistent surveillance of animals via AI
from IRTVs. Two parameter engineering procedures ignites ethical debates concerning their privacy and
were employed to extract the heart rate (HR) in beats autonomy [91]. A judicious equilibrium between welfare
per minute (BPM) and the respiratory rate (RR) from monitoring and preserving their independence is vital.
CURRENT APPLICATIONS AND RESEARCH PROGRESS
- 105

F I G U R E 6 The Summary of the Advantages and Disadvantages of Several Technologies. Icons were from www.freecodecamp.org/news/
introduction‐to‐iot‐internet‐of‐things/, www.futura‐sciences.com/tech/definitions/intelligence‐artificielle‐deep‐learning‐17262/, datascience.eu/
machine‐learning/interpreting‐your‐deep‐learning‐model‐by‐shap/, www.mime.asia/the‐difference‐between‐artificial‐intelligence‐deep‐
learning‐and‐machine‐learning/.

To propel animal care advancements through AI, By leveraging the precision, automation, and holistic
various avenues warrant exploration. Integration of AI management capabilities of AI, addressing the limita-
with the IoT can foster more intelligent animal care tions and concerns, and exploring future directions, the
facilities, facilitating perpetual monitoring and potential for enhancing animal welfare through AI
enhancement of all facets, from diet to habitat [4, 19]. technology is promising. With continued research and
Real‐time data from IoT devices can be analyzed by AI responsible implementation, the integration of AI in
algorithms to provide optimal care for animals [9]. animal care will likely contribute to the well‐being of
Secondly, the future may witness the development of animals, sustainable agriculture, and veterinary
AI systems capable of personalizing care plans for in- practices.
dividual animals based on their unique behaviors,
emotions, and health needs [16, 46]. Similar to how AI
customizes ads for individuals, personalized animal CONCLUSION
care can lead to better outcomes. Finally, collaborative
research involving technologists, veterinarians, and In conclusion, the utilization of AI offers substantial
animal behaviorists is essential. This interdisciplinary potential for enhancing animal welfare. By capitalizing
approach ensures that AI technology in animal care is on its advantages, addressing limitations and concerns,
developed and implemented in a way that truly benefits and exploring future directions, we can make significant
the animals. By combining expertise from different progress in this field. To achieve this, it is crucial to
fields, advancements can be made in AI algorithms, continue conducting research, implement AI technology
data collection, and ethical considerations, ultimately responsibly, and foster interdisciplinary collaboration.
improving animal welfare. These efforts will contribute to improved animal well‐
106
- ZHANG ET AL.

being, sustainable agriculture, and advancements in 7. Mehrish, A., Majumder, N., Bharadwaj, R., Mihalcea, R., &
veterinary practices. However, it is vital that we remain Poria, S. (2023). A review of deep learning techniques for
mindful of ethical considerations and ensure that AI is speech processing. Information Fusion.101869.
8. Bao, J., & Xie, Q. (2022). Artificial intelligence in animal farming:
used in a responsible and ethical manner, thereby A systematic literature review. Journal of Cleaner Production,
creating a brighter future for animals. 331, 129956. https://doi.org/10.1016/j.jclepro.2021.129956
9. Gehlot, A., Malik, P. K., Singh, R., Akram, S. V., & Alsuwian, T.
AUT HO R CO NT RI BUT I ON S (2022). Dairy 4.0: Intelligent communication ecosystem for the
Conceptualization, methodology and writing—original cattle animal welfare with blockchain and IoT enabled technol-
ogies. Applied Sciences, 12(14), 7316. https://doi.org/10.3390/
draft preparation, Li Zhang; writing—review and edit- app12147316
ing, Li Zhang, Wenqiang Guo, Chenrui Lv, Meng Guo., 10. Hou, J., He, Y., Yang, H., Connor, T., Gao, J., Wang, Y., Zeng,
Mei Yang, Qiuyue Fu and Xiaomeng Liu; supervision, Li Y., Zhang, J., Huang, J., Zheng, B., & Zhou, S. (2020). Identi-
Zhang; funding acquisition, Li Zhang All authors have fication of animal individuals using deep learning: A case study
of giant panda. Biological Conservation, 242, 108414. https://
read and agreed to the published version of the
doi.org/10.1016/j.biocon.2020.108414
manuscript. 11. Ji, B., Banhazi, T., Phillips, C. J., Wang, C., & Li, B. (2022). A
machine learning framework to predict the next month's daily
AC KNOWL E DG ME NT S milk yield, milk composition and milking frequency for cows in a
We would like to thank all those who have advised and robotic dairy farm. Biosystems Engineering, 216, 186–197.
helped us. This research did not receive any specific https://doi.org/10.1016/j.biosystemseng.2022.02.013
12. Manteuffel, C. (2019). Parturition detection in sows as test case
grant from funding agencies in the public, commercial, for measuring activity behaviour in farm animals by means of
or not‐for‐profit sectors. radar sensors. Biosystems Engineering, 184, 200–206. https://
doi.org/10.1016/j.biosystemseng.2019.06.018
CONFL ICT OF I NT ERE ST ST ATE ME NT 13. Park, M., Britton, D., Daley, W., McMurray, G., Navaei, M.,
Samoylov, A., Usher, C., & Xu, J. (2022). Artificial intelligence,
The authors declare no conflict of interest.
sensors, robots, and transportation systems drive an innovative
future for poultry broiler and breeder management. Animal
DA TA AV AI LA BI L I TY S TAT EM EN T Frontiers, 12(2), 40–48. https://doi.org/10.1093/af/vfac001
No new data were created or analyzed in this study. 14. Jiang, K., Xie, T., Yan, R., Wen, X., Li, D., Jiang, H., Jiang, N.,
Feng, L., Duan, X., & Wang, J. (2022). An attention mechanism‐
ETHI CS S TAT EM ENT improved YOLOv7 object detection algorithm for hemp duck
count estimation. Agriculture, 12(10), 1659. https://doi.org/10.
This research involved no experimental animal or hu- 3390/agriculture12101659
man participant and did not contain offensive language. 15. Feighelstein, M., Shimshoni, I., Finka, L. R., Luna, S. P., Mills,
D. S., & Zamansky, A. (2022). Automated recognition of pain in
O RCI D cats. Scientific Reports, 12(1), 9575. https://doi.org/10.1038/
Li Zhang https://orcid.org/0000-0001-8718-0216 s41598‐022‐13348‐1
16. Lencioni, G. C., de Sousa, R. V., de Souza Sardinha, E. J.,
Corrêa, R. R., & Zanella, A. J. (2021). Pain assessment in
R EFE RENCE S horses using automatic facial expression recognition through
1. Mellor, D. J. (2016). Updating animal welfare thinking: Moving deep learning‐based modeling. PLoS One, 16(10), e0258672.
beyond the “five freedoms” towards. “A Life Worth Living”, 6(3), https://doi.org/10.1371/journal.pone.0258672
21. https://doi.org/10.3390/ani6030021 17. Saar, M., Edan, Y., Godo, A., Lepar, J., Parmet, Y., & Halachmi,
2. Alonso, M. E., González‐Montaña, J. R., & Lomillos, J. M. I. (2022). A machine vision system to predict individual cow feed
(2020). Consumers’ concerns and perceptions of farm animal intake of different feeds in a cowshed. Animal, 16(1), 100432.
welfare. Animals, 10(3), 385. https://doi.org/10.3390/ https://doi.org/10.1016/j.animal.2021.100432
ani10030385 18. Ravi, G., & Choi, J. W. Data‐driven intelligent feeding system for
3. Fuentes, S., Gonzalez Viejo, C., Chauhan, S. S., Joy, A., Tong- pet care (pp. 2013–2018).
son, E., & Dunshea, F. R. (2020). Non‐invasive sheep biometrics 19. Fuentes, S., Gonzalez Viejo, C., Tongson, E., Lipovetzky, N., &
obtained by computer vision algorithms and machine learning Dunshea, F. R. (2021). Biometric physiological responses from
modeling using integrated visible/infrared thermal cameras. dairy cows measured by visible remote sensing are good pre-
Sensors, 20(21), 6334. https://doi.org/10.3390/s20216334 dictors of milk productivity and quality through artificial intelli-
4. Schindler, F., & Steinhage, V. (2021). Identification of animals gence. Sensors, 21(20), 6844. https://doi.org/10.3390/
and recognition of their actions in wildlife videos using deep s21206844
learning techniques. Ecological Informatics, 61, 101215. https:// 20. Feuser, A.‐K., Gesell‐May, S., Müller, T., & May, A. (2022).
doi.org/10.1016/j.ecoinf.2021.101215 Artificial intelligence for lameness detection in horses—A pre-
5. Jirkof, P., Rudeck, J., & Lewejohann, L. (2019). Assessing af- liminary study. Animals, 12(20), 2804. https://doi.org/10.3390/
fective state in laboratory rodents to promote animal welfare— ani12202804
What is the progress in applied refinement research? Animals, 21. Fuentes, S., Viejo, C. G., Tongson, E., Dunshea, F. R., Dac,
9(12), 1026. https://doi.org/10.3390/ani9121026 H. H., & Lipovetzky, N. (2022). Animal biometric assessment
6. Castiglioni, I., Rundo, L., Codari, M., Di Leo, G., Salvatore, C., using non‐invasive computer vision and machine learning are
Interlenghi, M., Gallivanone, F., Cozzi, A., D'Amico, N. C., & good predictors of dairy cows age and welfare: The future of
Sardanelli, F. (2021). AI applications to medical images: From automated veterinary support systems. Journal of Agriculture
machine learning to deep learning. Physica Medica, 83, 9–24. and Food Research, 10, 100388. https://doi.org/10.1016/j.jafr.
https://doi.org/10.1016/j.ejmp.2021.02.006 2022.100388
CURRENT APPLICATIONS AND RESEARCH PROGRESS
- 107

22. Almalawi, A., Alsolami, F., Khan, A. I., Alkhathlan, A., Fahad, 38. W. Andrew, C. Greatwood, and T. Burghardt. Aerial animal bio-
A., Irshad, K., Qaiyum, S., & Alfakeeh, A. S. J. E. R. (2022). An metrics: Individual friesian cattle recovery and visual identification
IoT based system for magnify air pollution monitoring and via an autonomous uav with onboard deep inference, 237‐243
prognosis using hybrid artificial intelligence technique (Vol. 39. Carslake, C., Vázquez‐Diosdado, J. A., & Kaler, J. (2020).
206).112576. Machine learning algorithms to classify and quantify multiple
23. Li, H.‐L., Yang, X.‐J., Li, J.‐Y., Zhan, K., Li, M., Gao, H.‐Y., & Hu, behaviours in dairy calves using a sensor: Moving beyond
Z.‐L. Design and application of intelligent monitoring system for classification in precision livestock. Sensors, 21(1), 88. https://
laminated henhouse with six overlap tiers cages (pp. 345–348). doi.org/10.3390/s21010088
24. Sun, H., Palaoag, T. D., & Quan, Q. Design of automatic 40. D. Gutierrez‐Galan, J. P. Dominguez‐Morales, E. Cerezuela‐
monitoring and control system for livestock and poultry house Escudero, A. Rios‐Navarro, R. Tapiador‐Morales, M. Rivas‐
environment based on Internet of Things robot (pp. 224–230). Perez, M. Dominguez‐Morales, A. Jimenez‐Fernandez, and
25. Feng, Q. C., & Wang, X. J. P. C. S. (2020). Design of disin- A. Linares‐Barranco (2018). Embedded neural network for real‐
fection robot for livestock breeding. Design of disinfection robot time animal behavior classification. nNeurocomputing, 272,
for livestock breeding, 166, 310–314. https://doi.org/10.1016/j. 17‐26. https://doi.org/10.1016/j.neucom.2017.03.090
procs.2020.02.093 41. Tedeschi, L. O., Greenwood, P. L., & Halachmi, I. (2021). Ad-
26. Qingchun, F., Xiu, W., Quan, Q., Chunfeng, Z., Bin, L., Ruifeng, vancements in sensor technology and decision support intelli-
X., and Liping, C. J. S. A. (2020). Design and test of disinfection gent tools to assist smart livestock farming. Journal of Animal
robot for livestock and poultry house, 2, 79 Science, 99(2), skab038. https://doi.org/10.1093/jas/skab038
27. P. D. Singh, R. Kaur, K. D. Singh, G. Dhiman, and M. Soni 42. D. S. Beggs, A. D. Fisher, E. C. Jongman, and P. H.
(2021). Fog‐centric IoT based smart healthcare support service Hemsworth (2015). A survey of Australian dairy farmers to
for monitoring and controlling an epidemic of Swine Flu virus. investigate animal welfare risks associated with increasing
Informatics in Medicine Unlocked, 26, 100636 scale of production. Journal of Dairy Science, 98(8), 5330‐5338.
28. Taneja, M., Jalodia, N., Byabazaire, J., Davy, A., & Olariu, C. https://doi.org/10.3168/jds.2014‐9239
(2019). SmartHerd management: A microservices‐based fog 43. Neethirajan, S. (2022). Affective state recognition in livestock—
computing–assisted IoT platform towards data‐driven smart Artificial intelligence approaches. Animals, 12(6), 759. https://
dairy farming. Software: Practice and Experience, 49(7), doi.org/10.3390/ani12060759
1055–1078. https://doi.org/10.1002/spe.2704 44. Neethirajan, S., Reimert, I., & Kemp, B. (2021). Measuring farm
29. Chatterjee, P. S., Ray, N. K., & Mohanty, S. P. (2021). LiveCare: animal emotions—Sensor‐based approaches. Sensors, 21(2),
An IoT‐based healthcare framework for livestock in smart agri- 553. https://doi.org/10.3390/s21020553
culture. IEEE Transactions on Consumer Electronics, 67(4), 45. Buller, H., Blokhuis, H., Lokhorst, K., Silberberg, M., & Veissier,
257–265. https://doi.org/10.1109/tce.2021.3128236 I. (2020). Animal welfare management in a digital world. Animal
30. Feng, Y., Niu, H., Wang, F., Ivey, S. J., Wu, J. J., Qi, H., Welfare Management in a Digital World, 10(10), 1779. https://
Almeida, R. A., Eda, S., & Cao, Q. (2021). SocialCattle: IoT‐ doi.org/10.3390/ani10101779
based mastitis detection and control through social cattle 46. Aich, S., Chakraborty, S., Sim, J.‐S., Jang, D.‐J., & Kim, H.‐C.
behavior sensing in smart farms. IEEE Internet of Things (2019). The design of an automated system for the analysis of
Journal, 9(12), 10130–10138. https://doi.org/10.1109/jiot.2021. the activity and emotional patterns of dogs with wearable sen-
3122341 sors using machine learning. Applied Sciences, 9(22), 4938.
31. L. Blömke, N. Volkmann, and N. Kemper (2020). Evaluation of https://doi.org/10.3390/app9224938
an automated assessment system for ear and tail lesions as 47. Weary, D. M., & Robbins, J. A. (2019). Understanding the
animal welfare indicators in pigs at slaughter. Meat Science, multiple conceptions of animal welfare. Animal Welfare, 28(1),
159, 107934. https://doi.org/10.1016/j.meatsci.2019.107934 33–40. https://doi.org/10.7120/09627286.28.1.033
32. Mohan, A., Raju, R. D., & Janarthanan, P. Animal disease 48. Hall, N. J., Johnston, A. M., Bray, E. E., Otto, C. M., MacLean,
diagnosis expert system using convolutional neural networks E. L., & Udell, M. A. R. (2021). Working dog training for the
(pp. 441–446). twenty‐first century. Frontiers in Veterinary Science, 8. https://
33. Gorczyca, M. T., & Gebremedhin, K. G. (2020). Ranking of doi.org/10.3389/fvets.2021.646022
environmental heat stressors for dairy cows using machine 49. Bray, E. E., Otto, C. M., Udell, M. A. R., Hall, N. J., Johnston,
learning algorithms. Computers and Electronics in Agriculture, A. M., & MacLean, E. L. (2021). Enhancing the selection and
168, 105124. https://doi.org/10.1016/j.compag.2019.105124 performance of working dogs. Frontiers in Veterinary Science,
34. Fuentes, S., Viejo, C. G., Tongson, E., & Dunshea, F. R. (2022). 8. https://doi.org/10.3389/fvets.2021.644431
The livestock farming digital transformation: Implementation of 50. Giraudet, C. S. E., Liu, K., McElligott, A. G., & Cobb, M. (2022).
new and emerging technologies using artificial intelligence. Are children and dogs best friends? A scoping review to explore
Animal Health Research Reviews, 1–13. the positive and negative effects of child‐dog interactions.
35. Li, N., Ren, Z., Li, D., & Zeng, L. (2020). Automated techniques PeerJ, 10, e14532. https://doi.org/10.7717/peerj.14532
for monitoring the behaviour and welfare of broilers and laying 51. Fennell, D. A. (2022). Animal‐informed consent: Sled dog tours
hens: Towards the goal of precision livestock farming. Animal, as asymmetric agential events. Tourism Management, 93,
14(3), 617–625. https://doi.org/10.1017/s1751731119002155 104584. https://doi.org/10.1016/j.tourman.2022.104584
36. Congdon, J. V., Hosseini, M., Gading, E. F., Masousi, M., 52. Vandergraff, S. (2017). The effects of reward‐based training on
Franke, M., & MacDonald, S. E. (2022). The future of artificial the behavioral assessment of the domestic dog. M.S., Murray
intelligence in monitoring animal identification. Health, and State University.
Behaviour, 12(13), 1711. https://doi.org/10.3390/ani12131711 53. Ferres, K., Schloesser, T., & Gloor, P. A. (2022). Predicting dog
37. M. Marsot, J. Mei, X. Shan, L. Ye, P. Feng, X. Yan, C. Li, emotions based on posture analysis using DeepLabCut. Future
and Y. Zhao (2020). An adaptive pig face recognition approach Internet, 14(4), 97. https://doi.org/10.3390/fi14040097
using Convolutional Neural Networks. Computers and Elec- 54. Wang, H., Liu, J., Dong, Z., Song, J., & Zhu, Z. (2023). Artificial
tronics in Agriculture, 173, 105386, https://doi.org/10.1016/j. intelligence‐based metabolic energy prediction model for animal
compag.2020.105386 feed proportioning optimization. Italian Journal of Animal
108
- ZHANG ET AL.

Science, 22(1), 942–952. https://doi.org/10.1080/1828051x. 72. Lewis, B. M., Battye, W. H., Aneja, V. P., Kim, H., & Bell, M. L. J.
2023.2236132 E. H. P. (2023). Modeling and analysis of air pollution and
55. Y. Qiao, H. Kong, C. Clark, S. Lomax, D. Su, S. Eiffert, and environmental justice: The case for North Carolina’s hog
S. Sukkarieh (2021). Intelligent perception for cattle monitoring: concentrated animal. Feeding Operations, 131(8), 087018.
A review for cattle identification, body condition score evalua- https://doi.org/10.1289/ehp11344
tion, and weight estimation. Computers and Electronics in 73. Zhao, Y.‐L., Huang, H.‐P., Chen, T.‐L., Chiang, P.‐C., Chen, Y.‐
Agriculture, 185, 106143, https://doi.org/10.1016/j.compag. H., Yeh, J.‐H., Huang, C.‐H., Lin, J.‐F., & Weng, W.‐T. J. I. S. J.
2021.106143 (2021). A smart sterilization robot system with chlorine dioxide
56. Maharjan, P., & Liang, Y. (2020). Precision livestock farming: for spray disinfection. A smart sterilization robot system with
The opportunities in poultry sector. J. Agric. Sci. Technol. A, chlorine dioxide for spray disinfection, 21(19), 22047–22057.
10(2), 45–53. https://doi.org/10.17265/2161‐6256/2020.02.001 https://doi.org/10.1109/jsen.2021.3101593
57. Muir, S. K., Linden, N. P., Kennedy, A., Calder, G., Kearney, G., 74. Ezanno, P., Picault, S., Beaunée, G., Bailly, X., Muñoz, F.,
Roberts, R., Knight, M. I., & Behrendt, R. (2020). Technical Duboz, R., Monod, H., & Guégan, J.‐F. J. V. R. (2021).
note: Validation of an automated feeding system for measuring Research perspectives on animal health in the era of artificial
individual animal feed intake in sheep housed in groups. intelligence (Vol. 52, pp. 1–15).
Translational Animal Science, 4(2), 1006–1016. https://doi.org/ 75. Ezanno, P., Picault, S., Beaunée, G., Bailly, X., Muñoz, F.,
10.1093/tas/txaa007 Duboz, R., Monod, H., & Guégan, J.‐F. (2021). Research per-
58. Bezen, R., Edan, Y., & Halachmi, I. (2020). Computer vision spectives on animal health in the era of artificial intelligence.
system for measuring individual cow feed intake using RGB‐D Veterinary Research, 52(1), 40. https://doi.org/10.1186/s13567‐
camera and deep learning algorithms. Computers and Elec- 021‐00902‐4
tronics in Agriculture, 172, 105345. https://doi.org/10.1016/j. 76. Patel, M., Surti, M., & Adnan, M. (2023). Artificial intelligence (AI)
compag.2020.105345 in Monkeypox infection prevention. J Biomol Struct Dyn, 41(17),
59. W. C. Wu, K. C. Cheng, and P. Y. Lin. A remote pet feeder 8629–8633. https://doi.org/10.1080/07391102.2022.2134214
control system via MQTT protocol, 487‐489 77. Ma, S., Yao, Q., Masuda, T., Higaki, S., Yoshioka, K., Arai, S.,
60. Y. Chen, and M. Elshakankiri. Implementation of an IoT based Takamatsu, S., & Itoh, T. (2021). Development of noncontact
pet care system, 256‐262. body temperature monitoring and prediction system for livestock
61. Knaebe, B., Weiss, C. C., Zimmermann, J., & Hayden, B. Y. cattle. IEEE Sensors Journal, 21(7), 9367–9376. https://doi.org/
(2022). The promise of behavioral tracking systems for 10.1109/jsen.2021.3056112
advancing primate animal welfare. Animals, 12(13), 1648. 78. Niu, B., Liang, R., Zhou, G., Zhang, Q., Su, Q., Qu, X., & Chen,
https://doi.org/10.3390/ani12131648 Q. (2020). Prediction for Global Peste des Petits Ruminants
62. Zemanova, M. A. (2020). Towards more compassionate wildlife Outbreaks Based on a Combination of Random Forest Algo-
research through the 3Rs principles: Moving from invasive to rithms and Meteorological Data. Frontiers in Veterinary Science,
non‐invasive methods. Wildlife Biology, 2020(1). https://doi.org/ 7, 570829. https://doi.org/10.3389/fvets.2020.570829
10.2981/wlb.00607 79. Kour, S., Agrawal, R., Sharma, N., Tikoo, A., pandey, N., &
63. Xiao, Y., Wang, H. H., & Li, J. (2021). A new market for pet food Sawhney, A. (2022). Artificial intelligence and its application in
in China: Online consumer preferences and consumption. The animal disease diagnosis. Journal of Animal Research, 12,
Chinese Economy, 54(6), 430–440. https://doi.org/10.1080/ 1–10. https://doi.org/10.30954/2277‐940x.01.2022.1
10971475.2021.1890360 80. Ghaffar Nia, N., Kaplanoglu, E., & Nasab, A. (2023). Evaluation
64. Li, Y. (2016). China perspective: Emerging interest in animal of artificial intelligence techniques in disease diagnosis and
behaviour and welfare science. Animals and Us, 241–252. Wage- prediction. Discover Artificial Intelligence, 3(1), 5. https://doi.org/
ningen Academic. https://doi.org/10.3920/978‐90‐8686‐828‐5_12 10.1007/s44163‐023‐00049‐5
65. Y. Xu, R. Wei, R. Mao, Z. Zheng, D. Nie, Z. Xu, and C. Tian. 81. Rashamol, V. P., Sejian, V., Pragna, P., Lees, A. M., Bagath,
Intelligent pet protection system based on IoT devices, 629‐634. M., Krishnan, G., & Gaughan, J. B. (2019). Prediction models,
66. M. O. Alrukban, Y. A. Alekrish, M. H. Alshehri, Y. A. Bajeaifer, assessment methodologies and biotechnological tools to
M. H. Alhamad, F. A. Sambas, and A. A. Alsouan (2022). quantify heat stress response in ruminant livestock. Interna-
Awareness of pet owners in riyadh regarding pet‐related health tional Journal of Biometeorology, 63(9), 1265–1281. https://doi.
risks and their associated preventative measures. Vector Borne org/10.1007/s00484‐019‐01735‐9
and Zoonotic Diseases, 22(8), 419‐424, https://doi.org/10.1089/ 82. Mbuthia, J. M., Mayer, M., & Reinsch, N. (2021). Modeling heat
vbz.2022.0017 stress effects on dairy cattle milk production in a tropical environ-
67. Yantong, L., Jiashuo, W., & Binrui, C. Smart pet collars possess ment using test‐day records and random regression models. Ani-
potential in the Chinese market (pp. 393–397). mal, 15(8), 100222. https://doi.org/10.1016/j.animal.2021.100222
68. Bhatia, M. (2020). Fog computing‐inspired smart home frame- 83. Parmiggiani, A., Liu, D., Psota, E., Fitzgerald, R., & Norton, T.
work for predictive veterinary healthcare. Microprocessors and (2023). Don’t get lost in the crowd: Graph convolutional network
Microsystems, 78, 103227. https://doi.org/10.1016/j.micpro. for online animal tracking in dense groups. Computers and
2020.103227 Electronics in Agriculture, 212, 108038. https://doi.org/10.1016/
69. Staley, M., Conners, M. G., Hall, K., & Miller, L. J. (2018). j.compag.2023.108038
Linking stress and immunity: Immunoglobulin A as a non‐ 84. Siriani, A. L. R., Kodaira, V., Mehdizadeh, S. A., de Alencar
invasive physiological biomarker in animal welfare studies. Nääs, I., de Moura, D. J., & Pereira, D. F. (2022). Detection and
Hormones and Behavior, 102, 55–68. https://doi.org/10.1016/j. tracking of chickens in low‐light images using YOLO network
yhbeh.2018.04.011 and Kalman filter. Neural Computing & Applications, 34(24),
70. Patel, H., Samad, A., Hamza, M., Muazzam, A., & Harahap, M. K. 21987–21997. https://doi.org/10.1007/s00521‐022‐07664‐w
(2022). Role of artificial intelligence in livestock and poultry 85. Neethirajan, S., Reimert, I., & Kemp, B. (2021). Measuring farm
farming. Sinkron: Jurnal Dan Penelitian Teknik Informatika, 7(4), animal emotions—Sensor‐based approaches. Sensors, 21(2),
2425–2429. https://doi.org/10.33395/sinkron.v7i4.11837 553. https://doi.org/10.3390/s21020553
71. Cheng, Y. H. A development architecture for the intelligent an- 86. Seibel, H., Weirup, L., & Schulz, C. (2020). Fish welfare – Be-
imal care and management system based on the internet of tween regulations, scientific facts and human perception. Food
Things and artificial intelligence (pp. 078–081). Ethics, 5(1), 4. https://doi.org/10.1007/s41055‐019‐00063‐3
CURRENT APPLICATIONS AND RESEARCH PROGRESS
- 109

87. Upadhyay, A. K., & Khandelwal, K. (2019). Artificial intelligence‐ Agricultural and Environmental Ethics, 34(2), 12. https://doi.org/
based training learning from application. Development and 10.1007/s10806‐021‐09852‐5
Learning in Organizations: An International Journal, 33(2),
20–23. https://doi.org/10.1108/dlo‐05‐2018‐0058
88. Davis, S. E., Lasko, T. A., Chen, G., Siew, E. D., & Matheny,
M. E. (2017). Calibration drift in regression and machine
How to cite this article: Zhang, L., Guo, W., Lv,
learning models for acute kidney injury. J Am Med Inform Assoc,
24(6), 1052–1061. https://doi.org/10.1093/jamia/ocx030 C., Guo, M., Yang, M., Fu, Q., & Liu, X. (2024).
89. Jarrahi, M. H., Memariani, A., & Guha, S. J. (2022). The prin- Advancements in artificial intelligence technology
ciples of data‐centric ai (dcai). for improving animal welfare: Current applications
90. Parikh, R. B., & Helmchen, L. A. (2022). Paying for artificial and research progress. Animal Research and
intelligence in medicine. Npj Digital Medicine, 5(1), 63. https://
One Health, 2(1), 93–109. https://doi.org/10.
doi.org/10.1038/s41746‐022‐00609‐6
91. Cooke, S. (2021). The ethics of touch and the importance of 1002/aro2.44
nonhuman relationships in animal agriculture. Journal of