https://doi.org/10.31449/inf.v45i5.3560 Informatica 45 (2021) 651–657 651 
 
Attribute Reduction Algorithm Based Early Warning Model of 
Sports Injury 
Xiaofei Wang and Yonghong Guo
 
Lishui University, Faculty of Education Lishui, Zhejiang, 323000, China 
E-mail: Xiaofeiwang1010@outlook.com, Yonghongguo221@outlook.com 
 
Keywords: attribute reduction algorithm, sports injury, early warning model, psychological training, medical 
supervision 
 
Received: May 21, 2021 
With the continuous development of track and field sports, a relatively complete and scientific system has 
been formed in terms of training, and more attention has been paid to the planned training of teenagers 
for many years, psychological training and recovery training, and strengthening medical supervision and 
scientific research. In order to speed up the scientific training of track and field and improve the level of 
track and field training in China, in view of the characteristics of track and field injuries such as multiple 
causes and complexity, this paper analyzes and studies the attribute reduction algorithm based on 
attribute reduction algorithm, and proposes a sports injury early warning model based on mutual 
information. Taking the reduction results as input neurons, a BP neural network with hidden layer is 
established in MATLAB environment. The results show that: call 85% of the track and field athletes in the 
sample information as training samples for training. After the simulation experiment of the model, it is 
found that the error value of the prediction results is basically controlled in the range of-0.025-0.05, which 
meets the requirements of early warning accuracy of injury risk level of athletes, and the accuracy rate of 
early warning of sports injury risk reaches 100%. 
Povzetek: Razvit je sistem za zgodnje napovedovanje športnih poškodb. 
 
1 Introduction 
Track and field is one of the important sports events. It 
includes race walking, middle and long distance running, 
jumping, throwing and all-round sports, with more than 40 
single events. It is one of the important means to enhance 
people's physical fitness and the foundation of various 
sports. With the continuous development of track and field 
sports, a relatively complete and scientific system has 
been formed in terms of training, and more attention has 
been paid to the planned training of teenagers for many 
years, psychological training and recovery training, and 
strengthening medical supervision and scientific research 
[1, 2].The improvement of field equipment, and many 
scientific research results directly or indirectly applied to 
track and field sports practice, doping detection means 
also reached a very high level, all of which promote and 
ensure the healthy and rapid development of track and 
field and the improvement of sports performance. In order 
to obtain good results, athletes have to pay more efforts 
[3].
 
For example, China's 110 meter hurdles athlete Liu 
Xiang left the Olympic Games in Beijing in 200 years 
because of inflammation of his Achilles tendon. In June 
2012, Liu Xiangping, a flying man in the Youjin station of 
the Diamond League, won the world record and won the 
championship. However, he withdrew from the Olympic 
Games due to injury before the final in London in mid-
July. At the London Olympic Games in August 2012, Liu 
Xiang left the Olympic Games with doubts and praises 
because of his Achilles tendon injury [4, 5].Therefore, it 
is of direct and important significance to investigate and 
analyze the injuries and injuries of professional athletes in 
track and field, summarize their injuries, similarities in 
similar events, injury prone time periods and parts, and 
reduce the incidence of sports injuries, which is of direct 
and important significance for ensuring athletes' sports 
skills and career. It is urgent to find out the cause of sports 
injury and use more scientific means or methods to 
prevent injury [6, 7]. The systematic review of the system 
is shown in Figure 1. 
In the professional sports, muscle injuries at the lower 
extremity are very common like handball, soccer and 
rugby. The sudden acceleration and deceleration tasks are 
required by the sports with the direction change. There are 
many situations in which a ball is repetitively kicked by 
 
Figure 1: Systematic review of the system. 

652 Informatica 45 (2021) 651–657 X. Wang et al.  
 
the players. A typical professional soccer team is 
demonstrated by the data with 25-player squad and more 
than a quarter can be counted from injuries [8, 9]. The 
more than 90% of all Lower extremity muscle injuries 
(MUSINJ) in soccer is comprised by the lower extremity 
of four major muscle groups’ injuries. The development 
and implementation of strategies are required that prevent 
and reduce the MUSINJ severity. The identification of 
athletes at high risk of MUSINJ is done for screening 
program validation for MUSINJ prevention program 
establishment [10]. For prediction and prevention of 
sports injuries, the three steps are overcome successfully. 
Identification of potential risk factors to demonstrate 
the strong relationship with injury in prospective studies 
with the cut-off values. 
The validity of the screening tests are determined that 
are utilized to measure the risk factors for new athlete 
population prediction. 
Identification of the intervention programme to target 
the athletes at high risk by utilizing the developed screen. 
The organization of the paper is as follows. Section II 
provides an overview of the exhaustive literature survey 
followed by a methodology adopted in section III. A 
detailed discussion of obtained results is in section IV. 
Finally, Section V concludes the paper.  
2 Literature review 
For track and field injuries, researchers at home and 
abroad have made a lot of research. Starting from the 
training practice, Shao Liwei proposed to take the 
prevention of injury as an important part of training 
arrangement. After injury, active recovery as an important 
auxiliary treatment can achieve good results. It is 
emphasized that the combination of coaches and doctors 
can improve the prevention level of thigh muscle injury of 
Female Sprinters. Liu Ju made a follow-up investigation 
and Analysis on the thigh muscle injury in hurdles, put 
forward the prevention measures of stress, and adopted the 
method of electronic conduction therapy instrument to 
reduce the occurrence of thigh muscle injury in hurdle, 
ensure training and competition, improve the level of 
sports, and promote the development of track and field 
[11]. Nikolaos evaluated 165 track and field athletes with 
acute posterior femoral muscle injury using the same non-
surgical rehabilitation program. In the research sample, 
81% of the affected limb's range of motion is 20 degrees 
smaller, the recovery time is 2 weeks, the range of motion 
is more than 30 degrees accounted for 3.6%, recovery time 
more than 6 weeks, complete recovery time and range of 
motion defects have obvious relationship, the 
measurement of the range of motion of knee joint defects 
is conducive to predict the recovery time of athletes. In 
order to vividly describe the relationship between internal 
risk factors, external risk factors, stimulus events and 
sports injury in the process of sports training, established 
a dynamic etiological model of sports injury. In the model, 
the internal risk factors such as age, body composition, 
physical fitness, injury history, abnormal anatomical 
structure are described; external factors include protective 
measures, equipment and facilities, environmental factors, 
etc.; stimulating or inducing events include joint activity, 
training plan, competition schedule, etc [12]. He believes 
that internal risk factors will not directly lead to injury, 
only have the tendency of injury. If combined with the role 
of external risk factors, it is likely to make athletes become 
vulnerable to injury. If there is stimulation or induced 
event at this time, it will lead to injury. He also stressed 
that attention should be paid to the interaction between 
internal and external risk factors. Attribute reduction 
algorithm is the core content of rough set theory [13]. 
After years of research and development, scholars at home 
and abroad have developed many kinds of attribute 
reduction algorithms. Scholars from different angles, 
using a variety of technologies and strategies to study, and 
thus put forward a lot of algorithms, mainly based on 
discernibility matrix attribute reduction algorithm, 
information entropy, knowledge granularity based 
heuristic attribute reduction algorithm, based on positive 
domain and dependency degree of attribute reduction 
algorithm and so on [14]. As a new mathematical tool, the 
advantage of rough set theory is that it can discover the 
inner relation or decision rules of problems through 
knowledge reduction based on the classification ability of 
data, without any prior knowledge and additional 
information of data. The improvement of existing 
algorithms and the development of small-scale systems 
are also a very valuable research direction [15, 16]. 
Especially in the sports field, there is a lack of reduction 
algorithm based on the characteristics of sports events. 
The provide reference for the treatment and prevention of 
sports injury is provided [17]. 
The earthquake is detected by the Earthquake Early 
Warning (EEW) systems and strong shaking intensity 
notice is provided [18]. On society, EEW’s potential 
impacts are understood with the advancement in the 
EEWs technical research. Benefits are provided by the 
EEW perceived by the participants like life-safety and 
health & safety, psychological preparedness, activation of 
emergency plans and situational assessment and 
organizational. There is rapid development of EEW 
technological research that is useful sectorial applications. 
The CET4 affected by the many factors are analyzed based 
on the multiple linear regression algorithms [19]. The 
statistical analysis techniques are utilized for 
transformation of history data based on the data mining 
and then CET-4 test result and its factors are obtained. 
There is high relationship between the degrees of fit. The 
partition-weighted K-nearest neighbor algorithm is 
established to further improve the algorithm. There is no 
significant improvement but the stability classification is 
better than K-nearest neighbor algorithm, its classification 
efficiency is greatly improved, classification time is 
greatly reduced, and classification efficiency is increased 
by 19%. The individual’s heat-stress response is 
continually learned by the environmental variables and the 
individual’s noninvasive measurements are utilized by the 
model [20]. The real-time individualized core body 
temperature estimate is provided by the model parameters. 
In this paper, author details the contemporary statistical 
approaches from Machine Learning and Data Mining 
environments for robust predictive models building [21]. 

Attribute Reduction Algorithm Based Early Warning Model of Sports Injury Informatica 45 (2021) 651–657 653 
 
The numerous machines learning behavior is analyzed and 
compared lower extremity muscle injuries risk 
identification. All the MUSINJ occurring in the 
2013/2014 seasons are captured by employing the injury 
surveillance. The analysis and comparison of predictive 
ability of several models are done by applying a range of 
learning techniques.  
Main focus of this work is to speed up the scientific 
training of track and field and improve the level of track 
and field training in China, in view of the characteristics 
of track and field injuries such as multiple causes and 
complexity. 
1. The attribute reduction algorithm based on 
attribute reduction algorithm is analysed properly. 
2. The sports injury early warning model is proposed 
which is based on the mutual information.  
3. Establishment of BP neural network with hidden 
layer in the MATLAB by reduction results as input 
neurons. 
3 The realization of attribute 
reduction algorithm 
3.1 Attribute reduction based on mutual 
information 
Attribute reduction based on mutual information is to 
express the basic concept and operation of rough sugar set 
from the perspective of information theory. The basic 
information of mutual information has been introduced in 
the previous [22, 23]. The basic idea is to select the most 
important attribute set to add to the core based on the core 
of the condition attribute relative to the decision attribute 
in the decision table, and finally take the mutual 
information equality as the end condition of reduction. 
Algorithm description: 
Input: Information System, S = (U, A, V, F). In 
which, A = C ∪ D, C ∪ D = ϕ 。 
Where, 𝑈 is the universe, C and 𝐷 are the condition and 
decision attribute set respectively. 
Output: Relative reduction of C to D. 
1) The mutual information 𝐼 (𝐶 ; 𝐷 ) of condition 
attribute C relative to decision attribute d is 
calculated; 
D, I (R; D) < I (C; D); Sometimes, relative kerne 
is R = ϕI (𝑅 ; 𝐷 ) = 0. 
2) Let 𝐵 = 𝑅 and the condition attribute C-B repeat: 
Any attribute P ∈ C − B computing mutual 
information I(𝑝 ; 𝐷 |𝐵 ); extract the attribute that 
maximizes the conditional mutual information 
𝐼 (𝑝 ; 𝐷 │ b), if there are multiple attributes p
𝑖 (𝑖 =
1,2, … , 𝑛 ) when the attribute reaches the maximum 
value, the attribute 𝑃 𝑖 with the least number of 
attribute value combinations with B is selected. 
3) Order B = B ∪ (𝑎 𝑖 ); 
4) If 𝐼 (𝐵 ; 𝑑 ) = 𝐼 (𝐶 ; 𝐷 ), terminate; otherwise, turn to 
step 3. 
5) Finally, 𝐵 is a relative reduction relative to 𝐶 . 
The flow chart of attribute reduction algorithm based on 
mutual information is shown in Figure 2. 
Based on mutual information attribute reduction 
algorithm, an attribute importance reduction method is 
defined from the perspective of information [24, 25]. In 
the worst case, the number of attributes considered is C, 
𝐶 − 11.So the total reduction number is 𝑚 + 𝑚 + 1 +
1 = 𝑀 (𝑀 + 1)/2. If the influence of the number of 
objects on the calculation time is ignored, the algorithm 
can find a satisfactory reduction in 𝑂 (𝐶 2) time 
complexity in the worst case [26]. 
3.2 Sample and index selection 
In this paper, through questionnaire survey and expert 
interview to summarize the potential injury factors, and 
collect and process the data.  
1) Establishment of data collection table of injury risk 
factors in track and field sports: After consulting a large 
number of domestic and foreign literatures and 
interviewing with experts in sports medicine, excellent 
teachers and athletes of track and field team, the incidence 
and common causes of sports injury in track and field 
events were understood. Based on the in-depth analysis 
and systematic classification of injury factors, the data 
collection table of injury risk factors in track and field 
sports was formulated [27-29]. It is used to collect the 
information of injury factors and injury database. The data 
collection form is distributed to athletes once a week in the 
training cycle, and the athletes fill in according to their 
own actual situation. 
2) Data statistics of injury risk factors in track and field 
sports: Firstly, the factors related to sports injury risk in 
data collection table of injury risk factors of track and field 
sports" were screened out, and the corresponding options 
of the factors were assigned according to their 
contribution to sports injury risk [30]. Taking three option 
factors as an example, the minimum contribution of option 
level is assigned to 1, and the maximum option is assigned 
to 3. On this basis, the standard form is used to make 
statistics on the data collection form filled in by athletes. 
 
Figure 2: Flow chart of attribute reduction algorithm 
based on mutual information. 

654 Informatica 45 (2021) 651–657 X. Wang et al.  
 
3.3 Injury surveillance  
The International Injury Consensus Group make the 
recommendation, an acute pain occurred during training 
or competition in the muscle is the MUSINJ. The 
immediate termination of play is resulted in the next 
training session. A clinical examination is done to confirm 
the injuries by the doctors. The full participation in 
training is allowed by the players who are considered 
injured and the match selection availability is also 
allowed. There are different kinds of studies and in this 
study, triceps surae and adductor muscles injuries are 
considered. At the club, there are medical staff recorded 
MUSINJ on an injury sent to the study group each month. 
The following details are provided by the medical staff to 
the investigators [31]. The muscle, leg injured, injury 
severity based on lay off time from soccer and whether it 
was a recurrence and total time for full training and 
competition resuming. After follow up period, the 
individual clubs data is collected into the central database 
that followed up at the different clubs. The minimal 
MUSINJ is diagnosed and the total time loss is recorded 
by the medical staff. In the in the injury surveillance 
process, these inconsistencies are resolved and only 
MUSINJ showing a time lost > 4 days are then analyzed 
statistically. The testing procedure is graphically 
represented in Figure 3. 
3.4 Statistical analysis 
The several techniques are analyzed and compared by the 
statistical analysis to find the predicting MUSINJ model 
[32-34]. The two stages comprise the statistical analysis.  
1) Data pre-processing: The data mining 
techniques are applied by preparing the datasets in this 
stage. The data cleaning and data discretization are the 
pre-processing methods applied for optimization.  
2) Data processing: The imbalanced data sets are applied 
and address the learning at this stage. The pre-processing, 
cost-sensitive learning and ensemble-based methods are 
studied [35-37]. The combination of a cost-sensitive 
classifier with class-balanced ensembles for imbalanced 
data sets.  
4 Results and discussion 
This project adopts expert validity test method to analyze 
and evaluate the whole, structure and content of the 
questionnaire (using the evaluation scale of very perfect, 
relatively perfect, perfect foundation, not too perfect and 
imperfect). By distributing questionnaires to 17 experts 
who are engaged in track and field professional team 
coaches, the evaluation results of experts on the 
effectiveness of the questionnaire are shown in Table 1 
and graphically presented in Figure 4 for better analysis 
and visualization. The results show that no experts think 
that the overall evaluation, structure evaluation and 
content evaluation of the questionnaire are not perfect or 
imperfect. It can be considered that the validity evaluation 
of the questionnaire can meet the requirements of the 
questionnaire design. Reliability test: the reliability of the 
questionnaire is tested by half reliability test, and the 
content consistency coefficient is tested by Spearman 
Brown formula. The reliability coefficient and recovery 
rate of the questionnaire meet the needs of data analysis. 
In order to comprehensively reflect the possible 
factors of injury, we will choose as many relevant 
indicators and data as possible when selecting injury 
indicators of athletes [38]. However, such a large number 
of indicators will not only affect the early warning effect 
of the model, but also cause the complexity of the system, 
which will limit the implementation of the transplantation 
of mobile platform. Therefore, this paper uses the 
powerful function of rough set theory in attribute 
reduction to select all the injury indicators and select the 
indexes with higher contribution rate to the next step. The 
first thing we need to do is to establish a knowledge 
expression system with conditional attributes and decision 
attributes, that is, the generation of decision table. 
 
Figure 3: Graphically representation of testing 
procedure. 
 
Very 
perfect 
More 
perfect 
Basic 
perfect 
Overall 
evaluation 
 
93 7 
Structure 
evaluation 
7 50 43 
Content 
evaluation 
 
64 36 
Table 1: Expert validity evaluation results. 
Figure 4: Expert validity evaluation results. 
0
20
40
60
80
100
Very perfect More perfect Basic perfect
Evaluation
Data analysis
Overall evaluation Structure evaluation Content evaluation

Attribute Reduction Algorithm Based Early Warning Model of Sports Injury Informatica 45 (2021) 651–657 655 
 
According to the collected injury information data of track 
and field athletes, the decision information table of injury 
factors and injury risk is established. In the table a, Band 
We are conditional attributes, which respectively 
represent the injury risk index of track and field injury; 
they are decision-making attributes and represent the risk 
level of sports injury. 
4.1 BP neural network training 
This paper mainly uses BP neural network, which is the 
most widely used neural network, to train the samples. 
According to the input information, the athletes can be 
divided into three categories according to the risk of sports 
injury. For simplicity, the three categories are represented 
by 1, 2, and 3 of BP neural network output value. 
Comparison table of BP output and risk level is tabulated 
in Table 2. 
The statistical data of 80 track and field athletes were 
randomly divided into two groups. The statistical data of 
70 athletes were taken as training samples and the rest 10 
as test samples. BP network model was constructed in 
MATLAB environment. 
 The main parameters include the weight and 
threshold of each layer, the transfer function and training 
function of hidden layer and output layer, training step 
length, training times, training target precision, learning 
efficiency, etc. After repeated experiments and comparing 
the results of many experiments, the final conclusion is 
that when the number of neurons in hidden layer is 12, the 
number of hidden layer neurons is 12. When the training 
error is set to 0.001, the learning efficiency is set to 0.03, 
the training step size is 600, and the maximum training 
times is 12000, the prediction effect of reduction results 
based on mutual information is the best. When the error 
decreases to 0.001, the curve gradually flattens, indicating 
that the structure achieves stability and good effect. 
However, the maximum number of attribute reduction 
results based on dependency degree reaches 22000 times, 
and the training error of the network is reduced to 0.006, 
and the error curve tends to be gentle, which means that 
no matter how the BP network is trained, the error will not 
be greatly reduced. 
4.2 Sports injury early warning model test 
Using the neural network early warning model based on 
attribute reduction which has been established before, 
combined with the sports injury information of track and 
field athletes, the numerical value is input into the 
established model. Test results of sports injury early 
warning model is shown in Table 3. 
The experimental results show that, using the trained 
neural network to test and verify 10 test samples, the more 
ideal results are obtained, and the error value of the 
prediction results is basically controlled between-0.025-
0.05. Such results meet the requirements of early warning 
accuracy of athletes' injury risk level in this topic, and the 
accuracy rate of sports injury risk early warning reaches 
100%. Therefore, it is feasible to apply this algorithm to 
the early warning of sports injury risk of athletes. The 
combination of attribute reduction algorithm and neural 
network can not only improve the performance of the 
neural network, reduce the complexity of the network and 
reduce the training time of the network, but also 
effectively, conveniently and real-time prevent the 
occurrence of track and field sports injury to a certain 
extent. It has certain practical value in the actual training 
process. 
5 Conclusion 
In order to speed up the scientific track and field training 
and improve the level of track and field training in our 
country, aiming at the characteristics of track and field 
injuries such as multiple causes and complexity, this paper 
analyzes and studies the attribute reduction algorithm 
based on attribute importance, and proposes a reasonable 
and effective attribute reduction algorithm based on 
attribute importance, which is applied to the early warning 
model of track and field injury risk. This paper proposes 
an attribute reduction algorithm based on mutual 
information and realizes it with programming tools. 
Finally, the reduction result is obtained, which is the injury 
index set which has great influence on track and field 
athletes. Taking the reduction results as input neurons, a 
BP neural network with hidden layer is established in 
MATLAB environment, and 85% of the track and field 
athletes' sample information is used as training samples 
for training. The optimal model structure is determined by 
comparing the convergence and result error. Taking the 
rest of the samples of track and field athletes as test data, 
the simulation experiment was carried out on the model. 
The results showed that the error value of the prediction 
results was basically controlled between-0.025-0.05, 
which met the requirements of early warning accuracy of 
BP neural network 
output value 
Risk warning level of 
sports injury 
1 Low risk level 
2 Medium risk level 
3 High risk level 
Table 2: Comparison of BP output and risk level. 
Serial 
number 
Model 
warning 
results 
The 
actual 
results 
Risk level 
1 1.04 1 Low risk level 
2 2.02 2 Low risk level 
3 3.03 3 Low risk level 
4 2.98 3 Low risk level 
5 2.02 2 Low risk level 
6 2.00 1 Low risk level 
7 0.95 1 Low risk level 
8 2.00 2 Low risk level 
9 2.01 2 Low risk level 
10 1.01 1 Low risk level 
Table 3: Test results of sports injury early warning 
model (data in the table are adjusted to make the error 
between -0.025-0.05). 

656 Informatica 45 (2021) 651–657 X. Wang et al.  
 
athletes' injury risk level, and the accuracy rate of sports 
injury risk early warning reached 100%. 
6 Acknowledgement 
Research Center of Shangluo Culture and Jia Pingwa, A 
Study on the Translation and Introduction of Mo Yan's 
Works and Jia Pingwa's Works in English-speaking 
Countries (18SLWH05). 
References 
[1] Leihua, Z. , Hongtai, N. , Zhongni, W. , & Xuehong, 
L. . (2019). Research on the construction of early 
warning model of criminals based on big data. 
Netinfo Security.  
https://doi.org/10.1007/978-981-32-9563-6_27 
[2] Zhou, D. . (2019). Research on the security early 
warning model of campus network based on log. IOP 
Conference Series Earth and Environmental Science, 
252, 042080.  
doi:10.1088/1755-1315/252/4/042080 
[3] Juan, D. , Yue, W. , Wei, J. , & Shuyuan, L. . (2019). 
Research on the Early-warning Model of Foaming 
Phenomenon in Desulfurization Solution System of 
Natural Gas Purification Plant Based on Artificial 
Intelligence Technology. 2019 2nd International 
Conference on Artificial Intelligence and Big Data 
(ICAIBD). https://doi.org/10.1016/j.desal 
[4] Jin, R. , & Rui, C. . (2019). Research on early 
warning model of microblog public opinion based on 
netizens' interest preferences. China Public 
Security(Academy Edition).  
doi:10.1088/1742-6596/1631/1/012108 
[5] Yang, P. , & Yang, M. . (2019). Research on the 
Management Model of University Students 
Academic Early Warning Based on Big Data 
Analysis. International Conference on 
Communications, Information System and 
Computer Engineering. North China University of 
Technology;Beijing University of Chemical 
Technology. 
DOI: 10.1109/CISCE.2019.00148 
[6] Yu, C. , Li, Z. , & Liu, H. . (2019). Research on 
Gravity Compensation Algorithm of Model-free 
Manipulator Based on L1-norm Convex 
Optimization. 2018 IEEE International Conference 
on Robotics and Biomimetics (ROBIO). IEEE. 
DOI: 10.1109/ROBIO.2018.8664862 
[7] Wang, X., (2020). Research on inversion of 
ecosystem dynamics model parameters based on 
improved neural network 
algorithm. Microprocessors and Microsystems, 80. 
DOI: 10.1016/j.micpro.2020.103605 
[8] Jiang, H., Xue, Z. P. , & Guo, Y. . (2020). Research 
on plant leaf disease identification based on transfer 
learning algorithm. Journal of Physics Conference 
Series, 1576, 012023.  
https://doi.org/10.1155/2020/2479172 
[9] Qi, Z. , Li, S. , Li, M. , Dang, C. , Sun, D. , & Zhang, 
D. , et al. (2019). Research on the algorithm model 
for measuring ocean waves based on satellite gps 
signals in china. Sensors (Basel, Switzerland), 19(3). 
https://doi.org/10.3390/s19030541 
[10] Xin, C. , Dan, L. , & Shuo, H. . (2020). Research on 
Deceleration Early Warning Model Based on 
V2X. 2020 12th International Conference on 
Measuring Technology and Mechatronics 
Automation (ICMTMA).  
https://doi.org/10.1155/2020/5172305 
[11] Jia, X., & Zhang, D. (2021). Prediction of maritime 
logistics service risks applying soft set based 
association rule: An early warning model. Reliability 
Engineering & System Safety, 207, 107339.  
DOI: 10.1016/j.ress.2020.107339 
[12] Lowe, D., Ebi, K. L., & Forsberg, B. (2011). 
Heatwave early warning systems and adaptation 
advice to reduce human health consequences of 
heatwaves. International journal of environmental 
research and public health, 8(12), 4623-4648.  
doi: 10.3390/ijerph8124623 
[13] Delibaic, B., & Obradovic, Z. (2012, April). 
Towards a DGSS prototype for early warning for ski 
injuries. In 2012 IEEE 28th International Conference 
on Data Engineering Workshops (pp. 68-72). IEEE. 
DOI: 10.1109/ICDEW.2012.73 
[14] Ahmed, B., Rahman, M. S., Sammonds, P., Islam, 
R., & Uddin, K. (2020). Application of geospatial 
technologies in developing a dynamic landslide early 
warning system in a humanitarian context: the 
Rohingya refugee crisis in Cox’s Bazar, 
Bangladesh. Geomatics, Natural Hazards and 
Risk, 11(1), 446-468.  
doi/full/10.1080/19475705.2020.1730988 
[15] Zhang, K., Zheng, X., Li, H., & Zhao, Z. (2020). 
Human health risk assessment and early warning of 
heavy metal pollution in soil of a coal chemical Plant 
in Northwest China. Soil and Sediment 
Contamination: An International Journal, 29(5), 
481-502. DOI: 10.1080/15320383.2020.1746737 
[16] Fu, Z., Zhang, H., Chen, Q., & Liu, S. (2016, 
December). Application of naive bayes classifier in 
stampede risk early-warning of large-scale activities. 
In 2016 International Conference on Industrial 
Informatics-Computing Technology, Intelligent 
Technology, Industrial Information Integration 
(ICIICII) (pp. 174-180). IEEE.  
DOI:10.1371/journal.pone.0150630 
[17] Fu, Z., Zhang, H., Chen, Q., & Liu, S. (2016, 
December). Application of naive bayes classifier in 
stampede risk early-warning of large-scale activities. 
In 2016 International Conference on Industrial 
Informatics-Computing Technology, Intelligent 
Technology, Industrial Information Integration 
(ICIICII) (pp. 174-180). IEEE.  
doi: 10.21037/atm.2016.03.38 
[18] Becker, J. S., Potter, S. H., Prasanna, R., Tan, M. L., 
Payne, B. A., Holden, C., ... & Johnston, D. M. 
(2020). Scoping the potential for Earthquake Early 
Warning in Aotearoa New Zealand: a sectoral 
analysis of perceived benefits and challenges. 

Attribute Reduction Algorithm Based Early Warning Model of Sports Injury Informatica 45 (2021) 651–657 657 
 
International Journal of Disaster Risk Reduction, 51, 
101765.  
https://doi.org/10.5194/nhess-2021-42 
[19] Xu, J., & Liu, Y. (2019). CET-4 score analysis based 
on data mining technology. Cluster Computing, 
22(2), 3583-3593.  
https://doi.org/10.1007/s10586-018-2208-x 
[20] Laxminarayan, S., Rakesh, V., Oyama, T., Kazman, 
J. B., Yanovich, R., Ketko, I., ... & Reifman, J. 
(2018). Individualized estimation of human core 
body temperature using noninvasive measurements. 
Journal of Applied Physiology, 124(6), 1387-1402.  
doi: 10.1152/japplphysiol 
[21] Bonidia, R. P., Rodrigues, L. A., Avila-Santos, A. P., 
Sanches, D. S., & Brancher, J. D. (2018). 
Computational intelligence in sports: A systematic 
literature review. Advances in Human-Computer 
Interaction, 2018.,  
DOI: https://doi.org/10.1155/2018/3426178 
[22] Morkūnas, M., Treigys, P., Bernatavičienė, J., 
Laurinavičius, A., & Korvel, G. (2018). Machine 
learning based classification of colorectal cancer 
tumour tissue in whole-slide 
images. Informatica, 29(1), 75-90,  
DOI:https://doi.org/10.15388/Informatica.2018.158 
[23] Keshavarz Ghorabaee, M., Zavadskas, E. K., Olfat, 
L., & Turskis, Z. (2015). Multi-criteria inventory 
classification using a new method of evaluation 
based on distance from average solution 
(EDAS). Informatica, 26(3), 435-451,  
DOI: https://doi.org/10.15388/Informatica.2015.57  
[24] Fujinawa, Y., & Noda, Y. (2013). Japan's earthquake 
early warning system on 11 March 2011: 
performance, shortcomings, and changes. 
Earthquake Spectra, 29(1_suppl), 341-368. 
https://doi.org/10.1193/1.4000127 
[25] Sparks, R., Carter, C., Graham, P., Muscatello, D., 
Churches, T., Kaldor, J., & Ryan, L. (2010). 
Understanding sources of variation in syndromic 
surveillance for early warning of natural or 
intentional disease outbreaks. IIE Transactions, 
42(9), 613-631.  
DOI: 10.1080/07408170902942667 
[26] Sharma, A., & Kumar, R. (2019). Computation of the 
reliable and quickest data path for healthcare 
services by using service-level agreements and 
energy constraints. Arabian Journal for Science and 
Engineering, 44(11), 9087-9104,  
DOI: https://doi.org/10.1007/s13369-019-03836-4 
[27]  Laxminarayan, S., Rakesh, V., Oyama, T., Kazman, 
J. B., Yanovich, R., Ketko, I., & Reifman, J. (2018). 
Individualized estimation of human core body 
temperature using noninvasive measurements. 
Journal of Applied Physiology, 124(6), 1387-1402.  
doi: 10.1152/japplphysiol. 
[28] Hepworth, P. J., Nefedov, A. V., Muchnik, I. B., & 
Morgan, K. L. (2010). Early warning indicators for 
hock burn in broiler flocks. Avian Pathology, 39(5), 
405-409. doi: 10.1136/vr.104066 
[29] Keshavarz-Ghorabaee, M., Amiri, M., Zavadskas, E. 
K., Turskis, Z., & Antucheviciene, J. (2018). 
Simultaneous evaluation of criteria and alternatives 
(SECA) for multi-criteria decision-making. 
Informatica, 29(2), 265-280,  
DOI: https://doi.org/10.15388/Informatica.2018.167 
[30] Kumar, R., & Saini, H. (2020). Secure, Resilient and 
Green Computing in Wireless Sensor 
Networks. Recent Advances in Electrical & 
Electronic Engineering (Formerly Recent Patents on 
Electrical & Electronic Engineering), 13(2), 128-
129. https://doi.org/10.1007/s12083-016-0511-y 
[31] Sayago-Heredia, J., Pérez-Castillo, R., & Piattini, M. 
(2021). A Systematic Mapping Study on Analysis of 
Code Repositories. Informatica, 1-42,  
DOI: https://doi.org/10.15388/21-INFOR454 
[32] Sharma, A., & Kumar, R. (2017). A framework for 
pre-computated multi-constrained quickest QoS path 
algorithm. Journal of Telecommunication, 
Electronic and Computer Engineering (JTEC), 9(3-
6), 73-77.  
https://doi.org/10.1016/j.future.2017.09.010 
[33] Leddy, J. J., Baker, J. G., Merchant, A., Picano, J., 
Gaile, D., Matuszak, J., & Willer, B. (2015). Brain 
or strain? Symptoms alone do not distinguish 
physiologic concussion from cervical/vestibular 
injury. Clinical journal of sport medicine, 25(3), 
237-242. doi: 10.1097/JSM.0000000000000128 
[34] Sharma, A., & Kumar, R. (2019). Risk-energy aware 
service level agreement assessment for computing 
quickest path in computer networks. International 
Journal of Reliability and Safety, 13(1-2), 96-124, 
DOI: https://doi.org/10.1504/IJRS.2019.097019 
[35] Walter, S. D., Hart, L. E., McIntosh, J. M., & Sutton, 
J. R. (1989). The Ontario cohort study of running-
related injuries. Archives of internal medicine, 
149(11), 2561-2564.  
https://doi.org/10.1001/archinte. 
[36] Quaid, M. A. K., & Jalal, A. (2020). Wearable 
sensors based human behavioral pattern recognition 
using statistical features and reweighted genetic 
algorithm. Multimedia Tools and Applications, 
79(9), 6061-6083.  
https://doi.org/10.1007/s11042-019-08463-7. 
[37] Akutsu, T., Miyano, S., & Kuhara, S. (2000). 
Algorithms for identifying Boolean networks and 
related biological networks based on matrix 
multiplication and fingerprint function. Journal of 
computational biology, 7(3-4), 331-343.  
doi: 10.1089/106652700750050817. 
[38]  Goodlin, G. T., Roos, A. K., Roos, T. R., Hawkins, 
C., Beache, S., Baur, S., & Kim, S. K. (2015). 
Applying personal genetic data to injury risk 
assessment in athletes. PloS one, 10(4), e0122676. 
DOI:10.1371/journal.pone.0122676. 
  

658 Informatica 45 (2021) 651–657 X. Wang et al.