Effective Injury Prediction in Professional Soccer

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/317087789
Effective injury prediction in professional soccer with GPS data and machine
learning
Preprint · May 2017
CITATIONS READS
2 2,859
6 authors, including:
Alessio Rossi Luca Pappalardo

Università di Pisa Italian National Research Council
61 PUBLICATIONS 187 CITATIONS 114 PUBLICATIONS 926 CITATIONS
SEE PROFILE SEE PROFILE
Paolo Cintia F. Marcello Iaia

Università di Pisa University of Milan
57 PUBLICATIONS 326 CITATIONS 87 PUBLICATIONS 3,782 CITATIONS
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
thermal imaging in face asymmetry studies View project
Special Issue on EPJ Data Science: "Individual and Collective Human Mobility: Description, Modelling, Prediction" View project
All content following this page was uploaded by Luca Pappalardo on 01 June 2017.
The user has requested enhancement of the downloaded file.

Effective injury prediction in professional soccer
with GPS data and machine learning
Alessio Rossi · Luca Pappalardo · Paolo

Cintia · F. Marcello Iaia · Javier
Fernández · Daniel Medina
arXiv:1705.08079v1 [stat.ML] 23 May 2017
Abstract Injuries have a great impact on professional soccer, due to their

large influence on team performance and the considerable costs of rehabilita-
tion for players. Existing studies in the literature provide just a preliminary
understanding of which factors mostly affect injury risk, while an evaluation
of the potential of statistical models in forecasting injuries is still missing. In
this paper, we propose a multidimensional approach to injury prediction in
professional soccer which is based on GPS measurements and machine learn-
ing. By using GPS tracking technology, we collect data describing the training
workload of players in a professional soccer club during a season. We show
that our injury predictors are both accurate and interpretable by providing a
set of case studies of interest to soccer practitioners. Our approach opens a
novel perspective on injury prevention, providing a set of simple and practical
rules for evaluating and interpreting the complex relations between injury risk
and training performance in professional soccer.
Keywords sports analytics · data science · machine learning · sports science ·
predictive analytics
1 Introduction
Injuries of professional athletes have a great impact on sports industry, due to

their large influence on both the mental state and the performance of a team
L. Pappalardo and P. Cintia
Department of Computer Science, University of Pisa, Italy
E-mail: [email protected], [email protected]
J. Fernandez and D. Medina
Sports Science and Health Department, Football Club Barcelona, Spain
A. Rossi and F.M. Iaia
Department of Biomedical Science for Health, University of Milan, Italy
2 Alessio Rossi et al.
[1, 2]. Furthermore, the costs associated with the complex process of recovery
and rehabilitation for the player is often considerable, both in terms of medical
care and missed earnings deriving from the popularity of the player himself
[3]. Recent research demonstrates that injuries in Spain cause an average of
16% of season absence by players, corresponding to a total cost estimation of
188 million euros just in one season [4]. It is not surprising hence that injury
prediction is attracting a growing interest from researchers, managers, and
coaches, who are interested in intervening with appropriate actions to reduce
the likelihood of injuries of their players.
Historically, academic work on injury prediction has been deterred for
decades by the limited availability of data describing the physical activity of
players during the season. Nowadays, the data revolution and the Internet of
Things have the potential to change rapidly this scenario thanks to Electronic
Performance and Tracking Systems (EPTS) [5, 6], new tracking technologies
that provide high-fidelity data streams, based on video recordings by different
cameras or observations by various kinds of fixed and mobile sensors [5, 7, 8, 9,
10]. Professional soccer clubs are starting to use these new technologies to col-
lect data from official games and training sessions, to ensure they can remain
in control of their players’ performance as much as possible. These soccer data
depict in detail the movements of players on the field [5, 6] and have been used
for many purposes, such as understanding game performance [11], identifying
training patterns [12], or performing automatic tactical analysis [5].
Despite this wealth of data, a little effort has been put on investigating in-
jury prediction in professional soccer so far [13, 14, 15]. Existing studies in the
literature provide just a preliminary understanding of which variables mostly
affect injury risk [13,14,15], while an evaluation of the potential of statistical
models in forecasting injuries is still missing. A major limit of existing studies
is that they follow a monodimensional approach: since they use just one vari-
able at a time to estimate injury risk, they do not fully exploit the complex
patterns underlying the available data. Professional soccer clubs are interested
in practical, usable and interpretable models as support in decision making to
coaches and athletic trainers [16]. The construction of such injury prediction
models poses many challenges. On one hand, an injury prediction model must
be highly accurate. Predictors which rarely forecast injuries are useless for
coaches, as well as predictors which frequently produce “false alarms”, i.e.,
misclassify healthy players as risky ones. On the other hand a “black box”
approach is less desirable for practical use since it does not provide insights
about the reason behind injuries. It is fundamental to understand the complex
relationships between the performance of players and their injury risk through
simple, interpretable, and easy-to-use tools. An interpretable model reveals
the influence of variables to injuries and the reasons behind them, allowing
soccer practitioners to react on time by modifying properly training demands.
Therefore, predictive models for injury prediction must achieve a good tradeoff
between accuracy and interpretability.
In this paper, we propose a data-driven approach to predict the injuries
in a professional soccer club and demonstrate that it is accurate and easy-to-
Title Suppressed Due to Excessive Length 3
interpret by soccer practitioners. We consider injury prediction as the problem

of forecasting that a player will get injured in the next training session or offi-
cial game, given his recent training workload. According to UEFA model [17],
a non-contact injury is defined as any tissue damage sustained by a player
that causes absence in next sports activities for at least the day after the
day of the onset. Our approach is based on automatic data collection through
standard GPS sensing technologies, and it is intended as a supporting tool
to the decision making of soccer managers and coaches. In the first stage of
our study, we collect data about training workload of players through GPS
devices, covering half of a season of a professional soccer club. After a pre-
processing task, we extract from the data a set of features used in sports
science to describe kinematic, metabolic and mechanical aspects of training
workload, and we enrich them with information about all the injuries which
happen during the half season. Based on these features, we implement bench-
mark predictors based on injury risk estimation methods commonly used in
sports science, showing that they are not effective for injury forecasting due
to their low precision. We hence move to a multidimensional approach and
use machine learning to train a repertoire of decision tree classifiers by using
the variables extracted. We find that injuries can be successfully predicted
with a small set of three variables: the presence of recent previous injuries,
high metabolic load distance and sudden decelerations. We then evaluate our
decision tree and observe that it can correctly forecast 76% of the injuries
recorded throughout the half season, achieving a 94% precision after around
16 weeks of training data. We also investigate a real-world scenario where the
classifiers are updated while new training workload and injury data become
available as the season goes by. The machine learning approach is still pre-
cise in this more challenging scenario, stabilizing its predictive performance at
around half of the observed period. Finally, we investigate the structure of the
best decision tree, extract an easy-to-use set of decision rules, highlight critical
values of the considered workload features and discuss related case studies of
interest to soccer practitioners. Our results are remarkable if we consider that
we use data from training sessions only, since the usage of sensing technologies
in games was not allowed by FIFA when our data collection was performed.
Given that FIFA now authorizes the usage of tracking systems in games [18],
our approach can be further improved by including game data, which are the
highest load demand in a player’s week.
Although our results refer to the situation of a specific professional club, it
shows that automatic data collection and machine learning allow for the con-
struction of injury prevention tools which are both accurate and interpretable,
opening an interesting perspective for engineering supporting tools for analy-
sis and prediction in professional soccer. Data-driven models like ours can be
used to describe and “nowcast” the health of players throughout the season in
a fast, cheap and reliable way.
The rest of the paper is organized as follows. Section 2 revises the scientific
literature relevant to our topic, Section 3 describes the data collection and the
feature extraction processes. In Section 4 we implement existing methodologies
for injury risk estimation and construct related predictors. In Section 5 we

describe and evaluate our machine learning approach to injury prediction,
comparing it with existing approaches. We interpret and discuss the results of
our experiments in Section 6 and Section 7. Finally, Section 8 concludes the
paper describing future improvements of our approach.
2 Related Work
The relationship between training workload, sports performance and injury

risk has been recently studied in the sports science literature [19, 20, 21, 22, 23,
24, 25, 26, 27]. These studies observe that injuries related to training workload
are to some extent preventable. For example Gabbett et al. [19, 20, 21, 22, 23, 24,
27] investigate the case of rugby and find that a player has a high risk of injury
when his workloads are increased above certain thresholds. As a consequence,
they conclude that balancing workload is a critical aspect of predicting injuries.
To assess injury risk in cricket, Hulin et al. [28] propose the ratio between
acute workload (i.e., the average workload in the last 7 days) and chronic
workload (i.e., the average workload in the last 28 days) using total week
distance to measure training workload. They observe that acute chronic ratio
values lower the 1 (acute < chronic) refer to cricket players in good physical
fit and correspond to a low injury risk (4%). In contrast, acute chronic ratio
values higher than 2 (acute > chronic) refer to players in a state of fatigue
and correspond to injury risk from 2 to 4 times higher than the other group of
players. Hence, the acute chronic ratio is useful to emphasize both positive and
negative consequences of training. Murray et al. [29] show that in rugby acute
chronic workload ratio is better related to injury risk when it is calculated using
exponential smoothed moving average instead of the simple average. Hulin et
al. [28] and Ehrmann et al. [14] find the same results for elite rugby players
and soccer players respectively. They find that injured players, in both rugby
and soccer, show significantly higher physical activity in the week preceding
the injury with respect to their seasonal averages.
In skating, Foster et al. [30] measure training workload by the “session
load”, i.e., the product of perceived exertion and duration of the training ses-
sion. They find that when the session load outweighs a skater’s ability to fully
recover before the next session, the skater suffers from the so-called “overtrain-
ing syndrome”, a condition that can cause injury [30]. In basketball, Ander-
son et al. [24] find a strong correlation between injury risk and the so-called
“monotony”, i.e., the ratio between the mean and the standard deviation of
the session load recorded in last 7 days. Moreover, Brink et al. [13] observe that
injured young soccer players (age < 18) recorded higher values of monotony
in the week preceding the injury than non-injured players.
Venturelli et al. [15] perform several periodic physical tests on young soc-
cer players (age<18) and measure body size, endurance, flexibility and jump
height. They find that jump height, body size and the presence of previous
injuries are significantly correlated with the probability of thigh strain injury.
Talukder et al. [31] apply a machine learning approach to predict injuries

in basketball, based on performance features extracted from NBA games. In
particular, they create a classifier able to predict 19% of injuries occurred in
NBA. They also show that the most important features for predicting injuries
are the average speed, the number of past competitions played, the average
distance covered, the number of minutes played to date and the average field
goals attempted. However, Talukder et al. [31] do not take into account the
players’ workload in training sessions.
Position of our work. In literature, the only studies addressing the problem of
injury prediction in soccer are the ones by Brink et al. [13], Ehrmann et al. [14]
and Venturelli et al. [15]. However, these studies suffer a major limitation: while
they observe the existence of a correlation between specific aspects of training
workload and the chance of injury, they do not construct any predictor as a
practical tool for coaches and athletic trainers to prevent injuries. Therefore,
to the best of our knowledge, there is no quantification of the potential of
predictive analytics in preventing non-traumatic injuries in professional soccer.
In this paper, we fill this gap by proposing a machine learning approach to
injury prevention and show that outperforms existing injury risk estimation
methods for professional soccer players.
3 Data collection and feature extraction
We set up a study on twenty-six professional players of a professional soccer

club in Italian Serie B (age = 26±4 years; height = 179±5 cm; body mass =
78±8 kg) during season 2013/2014. Six central backs, three fullbacks, seven
midfielders, eight wingers and two forwards were recruited. Participants gave
their written informed consent to participate in the study. The study was also
approved by the Ethical Committee of the University of Milan.
We monitored the physical activity of players during 23 weeks of training
sessions – from January 1st to May 31st – using portable 10 Hz GPS devices,
integrated with 100 Hz 3-D accelerometer, a 3-D gyroscope, a 3-D digital com-
pass (STATSports Viper). Despite some concerns over the reliability of GPS
measurement of accelerations [32], especially at low sampling rates, it has been
demonstrated that GPS traces are an useful tool for analyzing the activity pro-
file in team sports, including soccer [33]. The devices were placed between the
players’ scapulae through a tight vest and were activated 15 minutes before the
data collection, in accordance with the manufacturer’s instructions to optimize
the acquisition of satellite signals. We recorded a total of 954 individual train-
ing sessions corresponding to 80 collective training sessions performed during
the 23 weeks. From the data collected by the devices, we extracted a set of
training workload indicators through the software package Viper Version 2.1
provided by STATSports 2014. Figure 1a visualizes the GPS trace of a player
during a training session in our dataset.
The club’s medical staff recorded all the non-contact injuries occurred dur-
ing 23 weeks. According to UEFA regulations [17], a non-contact injury is de-
fined as any tissue damage sustained by a player that causes absence in next
football activities for at least the day after the day of the onset. We observed
21 non-contact injuries during the period of observation, 19 of them associ-
ated with players who got injured at least once in the past. The distribution of
injuries per player is provided in Figure 1. We observe that half of the players
never get injured during the period of observation, while the others get injured
once (seven players), twice (five players) or four times (one player). For every
player, we also collected information about his age, weight, height and role on
the field. Moreover, for every single training session of a player, we collected
information about the play time in the official game before the training session
and the number of official games played before the training session.
From the players’ GPS data of every training session we select 12 features
describing different aspects of workload [34]. Two features – Total Distance
(dTOT ) and High Speed Running Distance (dHSR ) – are kinematic, i.e., they
quantify the overall movement of a player during a training session. Three
features – Metabolic Distance dMET , High Metabolic Load Distance dHML and
High Metabolic Load Distance per minute dHML /m – are metabolic, i.e., they
quantify the energy expenditure of the overall movement of a player during
a training session. The remaining seven features – Explosive Distance (dEXP ),
number of accelerations above 2m/s2 (Acc2 ), number of accelerations above
3m/s2 (Acc3 ), number of decelerations above 2m/s2 (Dec2 ), number of decel-
erations above 3m/s2 (Dec3 ), Dynamic Stress Load (DSL) and Fatigue Index
(FI) – are mechanical features describing the overall muscular-scheletrical load
of a player during a training session. Table 1 provides a description of the
workload features, Appendix A provides descriptive statistics of the variables.
4 Classic approaches to injury risk estimation
In this section, we replicate the two most used state-of-the-art approaches to

injury risk estimation that can be implemented with the workload features
available in our study. In particular we consider the acute chronic workload
ratio (ACWR) used in cricket, rugby and soccer [14, 28, 29] and the mean
standard deviation workload ratio (MSWR) used in non-professional soccer
[13, 24]. First, since these approaches have been initially developed for other
sports or for non-professional players, we test their effectiveness on injury risk
estimation for professional soccer players. Secondly, since these approaches
aim just at observing the relations between training workload and injury risk
without providing any predictive model, we transform them in predictors and
assess their practical usability for soccer practitioners.
(a)

IUHTXHQF\
IUHTXHQF\

LQMXULHVSHUSOD\HU WUDLQLQJVSHUSOD\HU
(b) (c)
Fig. 1 (a) A visualization of the GPS trace of a player during a training session in our
dataset. The color of the lines indicates, in a gradient from white to red, the speed of the
player. (b) Distribution of the number of injuries per player. (c) Distribution of the number
of training sessions per player.
4.1 Acute Chronic Workload Ratio (ACWR)
Many existing works on injury risk estimation rely on the so-called Acute
Chronic Workload Ratio (ACWR) [14, 20, 22, 28, 29, 35], i.e., the ratio between
the acute workload and the chronic workload of a player. Although its validity
has been questioned by some authors [36, 37], ACWR is still one of the most
used techniques in professional soccer clubs [38]. As proposed by Murray et
al. [29], the acute workload of a player can be estimated by the exponential
weighted moving average of the workload in the previous 7 days; the chronic
workload of a player can be estimated as the exponential weighted moving av-
erage of the workload in the previous 28 days. ACWR can be calculated on any
Training workload features [34, 39, 40]

dTOT Distance in meters covered during the training session
dHSR Distance in meters covered above 5.5m/s
dMET Distance in meters covered at metabolic power
dHML Distance in meters covered by a player with a Metabolic Power is
above 25.5W/Kg
dHML /m Average dHML per minute
dEXP Distance in meters covered above 25.5W/Kg and below 19.8Km/h
Acc2 Number of accelerations above 2m/s2
Acc3 Number of accelerations above 3m/s2
Dec2 Number of decelerations above 2m/s2
Dec3 Number of decelerations above 3m/s2
DSL Total of the weighted impacts of magnitude above 2g. Impacts are
collisions and step impacts during running
FI Ratio between DSL and speed intensity
Age age of players
BMI Body Mass Index: ratio between weight (in kg) and the square of
height (in meters)
Role Role of the player
PI Number of injuries of the players before each training session
Play time Minutes of play in previous games
Games Number of games played before each training session
Table 1 Training workload features used in our study. Description of the training
workload features extracted from GPS data and the players’ personal features collected
during the study. We defined four categories of features: kinematic features (blue), metabolic
features (red), mechanical features (green) and personal features (white).
variable describing workload with a monodimensional approach: just one vari-

able is considered at a time to estimate a player’s injury risk [28, 29]. ACWR
values higher than 1 indicate that acute workload exceeds chronic workload,
i.e., the average training workload in the last 7 days is much higher than the
average workload in the previous 28 days, suggesting a training overload by
the player. In contrast, ACWR values lower than 1 indicate an acute workload
lower than the chronic workload.
Murray et al. [29] compute the ACWR for a set of workload features and
categorize rugby players’ training sessions in five groups: (1) ACWR < 0.49
(very low); (2) ACWR ∈ [0.50, 0.99] (low); (3) ACWR ∈ [1.00, 1.49] (moder-
ate); (4) ACWR ∈ [1.50, 1.99] (high); (5) ACWR > 2.00 (very high). Then
injury likelihood (IL) is estimated in every ACWR group as the ratio between
the number of players who get injured after the training session assigned to
that ACWR group and the number of players who do not. Murray et al. [29]
observe that players whose training sessions result in ACWR > 2 have a high
injury risk (i.e., a high IL). We reproduce the ACWR methodology on our
dataset for each of the 12 workload features described in Section 3, using the
ACWR groups suggested by Murray et al. [29]. In contrast with literature,
$&:5
,QMXU\OLNHOLKRRG ,QMXU\OLNHOLKRRG ,QMXU\OLNHOLKRRG ,QMXU\OLNHOLKRRG

dTOT dHSR dMET

dHML dHML /m dEXP

Acc2 Acc3 Dec2

Dec3 DSL FI

@
@
@
@
@
@
!
!
!
@
@
@

>
>
>
>
>
>
>
>
>
Fig. 2 Injury risk in ACWR groups. The plots show Injury Likelihood (IL) for pre-
defined ACWR groups [29], for every of the 12 training workload features considered in our
study. Bars are colored according to feature categorization defined in Table 3.
we do not find any individual training session resulting in ACWR > 2 (Fig-
ure 2), while we observe that players whose individual training sessions result
in ACWR < 1 have the highest injury risk, i.e., the highest IL values (Fig-
ure 2). We replicate the experiment by using ACWR groups defined by the
quintiles of the ACWR distribution instead of the pre-defined groups proposed
by Murray et al. [29], finding similar results (see Appendix B).
ACWR predictor. The literature about the ACWR approach provides just a
quantification of the relation between training workload and injury risk. We
explore the usability in practice of this approach as a tool for injury predic-
tion by constructing 12 predictive models, one for each of the 12 workload
features introduced in Section 3 (Table 1). Given a player’s training session,
every predictive model CACWR
h predicts whether or not the player will get in-
jured during next game or training session based on the value of workload
feature h recorded in the current training session. If considering feature h the
individual training session results in ACWR < 1, model ChACWR predicts an
injury (class 1) otherwise it predicts a non-injury (class 0). We validate the
accuracy of predictive models against real data and observe that, in average,
they have high recall (0.80 ± 0.08) and low precision (0.03 ± 0.003) on the
injury class (1): while most of the injuries (80%) are detected, in average the
models wrongly predict an injury in 97% of the cases (see Table 7). We also
explore the predictive power of the combination of the single predictors by
constructing three combined predictive models. Predictive model CACWR vote pre-
dicts that a player will get injured if his training session results in ACWR<1
for the majority of the workload features. Predictive model CACWR all predicts
that a player will get injured if his training session results in ACWR < 1 for
all the workload features. Model CACWR one predicts an injury if ACWR < 1 for
at least one workload feature. Table 7 reports the accuracy of the three pre-
dictive models. Only model CACWR vote achieves a slightly better performance, in
terms of precision on the injury class, w.r.t. models based on single features.
We compare these predictive models with four baselines. Baseline B1 randomly
assigns a class to an example by respecting the distribution of classes. Baseline
B2 always assigns the majority class (i.e., class 0, a non-injury), while baseline
B3 always assigns the minority class (i.e., class 1, injury). Baseline B4 is a
classifier which assigns class 1 (injury) if the exponentially weighted average
of variable PI > 0 (see Appendix D), and 0 (no injury) otherwise. Although
the 15 predictive models constructed are significantly better than the baseline
classifiers in terms of recall on the injury class, our results suggest that a pre-
dictor based on ACWR is not usable in practice due to its low precision. In a
scenario where an athletic trainer bases his decisions on the suggestions of the
predictors, in the vast majority of the cases he would generate “false alarms”
by stopping a player with no risk of injury, which is not a practical solution
to injury prevention in professional soccer.
4.2 Mean Standard deviation Workload Ratio (MSWR)
Another widely used approach to injury risk estimation is MSWR, defined in

the literature as the ratio between the mean and the standard deviation of a
player’s workload obtained in one week [24, 13, 30]. The higher the MSWR of
a player, the lower is the variability of his workloads during the training week.
In literature, it has been demonstrated that low MSWR values are associated
with positive game performance while high MSWR values are associated with
negative game performance and high injury risk [30]. Foster et al. [30] use this
approach on a workload feature defined as the product of a player’s training
duration and training workload as subjectively perceived by the players. In our
study, we use objective workload as measured by the 12 features introduced in
Section 3 and we compute MSWR for each of them. In detail, the MSWR value
of a player at the end of a training session with respect to a workload feature
h is the ratio between the mean and the standard deviation of the player’s
distribution of feature h in the week preceding the individual training session.
We investigate the relation between MSWR and injury risk by grouping the
individual training sessions into quintiles according to the distribution of the
workload features. For every quintile, we compute the corresponding injury
likelihood (IL). We observe that high MSWR values are related to high injury
ACWR MSWR
class prec rec F1 AUC prec rec F1 AUC
0 0.99 0.44 0.61 0.98 0.80 0.88
CdTOT 0.65 0.57
1 0.03 0.86 0.06 0.04 0.33 0.07
0 0.99 0.37 0.54 0.98 1.00 0.99
CdHSR 0.57 0.50
1 0.03 0.76 0.05 0.00 0.00 0.00
0 0.99 0.43 0.60 0.98 0.95 0.96
CdMET 0.59 0.55
1 0.03 0.76 0.06 0.06 0.14 0.09
0 0.99 0.43 0.60 0.98 0.96 0.97
CdHML 0.60 0.53
1 0.03 0.76 0.06 0.06 0.10 0.07
0 0.99 0.39 0.56 0.98 0.96 0.97
CdHML/m 0.60 0.55
1 0.03 0.81 0.06 0.08 0.14 0.10
0 1.00 0.43 0.60 0.98 0.94 0.96
CdEXP 0.67 0.49
1 0.04 0.91 0.07 0.02 0.05 0.03
0 0.99 0.47 0.64 0.98 0.93 0.95
CAcc2 0.64 0.46
1 0.03 0.80 0.06 0.00 0.00 0.00
0 0.99 0.45 0.64 0.98 0.98 0.98
CAcc3 0.58 0.49
1 0.03 0.71 0.06 0.00 0.00 0.00
0 0.99 0.46 0.63 0.98 0.94 0.96
CDec2 0.66 0.52
1 0.04 0.86 0.07 0.04 0.10 0.05
0 0.99 0.46 0.63 0.98 0.99 0.98
CDec3 0.66 0.49
1 0.04 0.86 0.07 0.00 0.00 0.00
0 1.00 0.42 0.60 0.98 0.97 0.97
CDSL 0.66 0.48
1 0.03 0.90 0.07 0.00 0.00 0.00
0 0.98 0.47 0.64 0.98 0.72 0.83
CFI 0.55 0.50
1 0.03 0.62 0.05 0.03 0.29 0.04
0 1.00 0.09 0.17 0.98 0.56 0.71
Cone 0.54 0.54
1 0.02 1.00 0.05 0.03 0.52 0.05
0 0.99 0.83 0.90 0.97 0.99 0.98
Cvote 0.65 0.49
1 0.06 0.48 0.11 0.00 0.00 0.00
0 0.98 0.82 0.90 0.97 1.00 0.99
Call 0.58 0.50
1 0.04 0.33 0.07 0.00 0.00 0.00
0 0.98 0.98 0.98 0.98 0.98 0.98
B1 0.51 0.51
1 0.06 0.05 0.05 0.06 0.05 0.05
0 0.98 1.00 0.99 0.98 1.00 0.99
B2 0.50 0.50
1 0.00 0.00 0.00 0.00 0.00 0.00
0 0.00 0.00 0.00 0.00 0.00 0.00
B3 0.50 0.50
1 0.02 1.00 0.04 0.02 1.00 0.04
0 0.98 0.77 0.86 0.98 0.77 0.86
B4 0.60 0.60
1 0.04 0.43 0.07 0.04 0.43 0.07
Table 2 Performance of ACWR and MSWR predictors. We report precision (prec),

recall (rec), F1-score (F1) and Area Under the Curve (AUC) for the injury class and the non-
injury class for all the predictors based on ACWR and MSWR. We also provide predictive
performance of four baseline predictors B1 , B2 , B3 and B4 .
risk for the majority of workload features, substantially confirming results

observed in the literature by Foster et al. [30] (see Figure 3).
MSWR predictor. As done for ACWR, we explore the usability in practice of

MSWR by constructing 12 predictive models based on the 12 training work-
load features introduced in Section 3. Given a player’s training session, every
predictive model ChMSWR predicts whether or not the player will get injured
during next game or training session based on the value of workload feature h.
If considering feature h the individual training session is associated with the
MSWR group with the highest injury risk, model ChMSWR predicts an injury
(class 1), otherwise it predicts a non-injury (class 0). We validate the accuracy
of these predictive models against real data and observe that in average they
have both low recall (mean is 0.10 ± 0.10) and low precision (mean is 0.03 ±
0.03) on the injury class: just 10% of the injuries are detected and in average
the models wrongly predict an injury in 97% of the cases. As done for ACWR,
MSWR MSWR MSWR
we also construct three combined models – Cvote , Call and Cone – and
observe that they have poor accuracy in detecting the injury class. In partic-
ular, the MSWR predictors have predictive power comparable to the ACWR
predictors. Both methodologies, in conclusion, are not usable in practice due
to their low precision.

06:5
,QMXU\OLNHOLKRRG
dTOT dHSR dMET

,QMXU\OLNHOLKRRG
dHML dHML /m dEXP

,QMXU\OLNHOLKRRG

Acc2 Acc3 Dec2

,QMXU\OLNHOLKRRG
Dec3 DSL FI

Fig. 3 Injury risk in MSWR groups. The plots show the Injury Likelihood (IL) for the
MSWR groups for every of the 12 training workload features considered in our study. Bars
are colored according to feature categorization defined in Table 1.
5 A machine learning approach to injury prediction
Given the inadequacy of the ACWR and MSWR predictors, we construct a

repertoire of classifiers to predict whether or not a player will get injured based
on his recent training workload. The construction of each classifier is made by
using a training dataset where each example refers to a single player’s training
session and consists of: (i) a vector of the workload features describing the
player’s recent workload including the current training session; and (ii) the
injury label, indicating whether or not the player got injured in next game or
training session. We use a decision tree as predictive algorithm because it is
easier to translate into practice and easy to understand even for people with
a non-analytical background.
5.1 Construction of training dataset
Given a feature set S, the training dataset TS for the learning task is con-
structed by a two-step procedure:
(1) For every individual training session i we construct a feature vector mi =
(h1 , . . . , hk ) where hj ∈ S, (j=1, . . . , k), is a training workload feature
and k = |S| is the number of features considered. All the feature vectors
compose matrix FS = (m1 , . . . , mn ), where n is the number of individual
training sessions in our dataset (n = 954);
(2) Every feature vector mi is associated to a label ci ∈ {0, 1} which is 1 if the
player gets injured in the next game or training session and 0 otherwise.
Matrix FS is hence associated to a vector of labels c = (c1 , . . . , cn ) (one for
each training session). The training dataset for the learning task is finally
TS = (FS , c). Table 3 shows an example of training dataset.
d TOT d EXP ... ACC 3 label
s1 4,018.19 426.42 ... 16.99 0
s2 3,465.81 326.41 ... 16.91 0
s3 3,227.15 256.85 ... 18.25 1

.. .. .. .. ..
. . . . .
sn 3,199.58 273.69 ... 19.64 1
Table 3 Structure of the training dataset used for injury prediction. Every ex-
ample describes a player’s training session and consists of k features and a label c ∈ {0, 1}.
We construct four different training datasets based on four different feature

sets built starting from the workload features in Table 1. Feature set wf con-
sists of the exponential weighted moving average (ewma) of the 12 workload
features in Table 1. We use ewma since, as suggested in literature [29, 31],

a player’s injury risk depends on both the workload in the current training
session and the workload in the recent training sessions. In detail, we consider
a span equal to six in the ewma and adjust every workload feature using
ewma. This choice is data-driven and justified by our experiments where we
find that the best classification results are achieved when using a span equal
to six (see Appendix C). Feature set acwr uses the ACWR version of the 12
workload features in Table 1: given a player’s training session, for every feature
we compute its acute chronic workload ratio as described in Section 4.1. Sim-
ilarly, feature set mswr uses the MSWR version of the 12 workload features
in Table 1: given a player’s training session, for every feature we compute the
exponential weighted moving standard deviation of the six most recent train-
ing sessions of the player. To take into account both the number of a player’s
previous injuries and their distance to the current training sessions we com-
pute PI(WF) , the ewma of feature PI with a span equal to 6. PI(WF) reflects the
distance between the current training session and the training session when
the player returned to regular training after an injury. PI(WF) = 0 indicates
that the player never got injured in the past; PI(WF) > 0 indicates that the
player got injured at least once in the past; PI(WF) > 1 indicate that the player
got injured more than once in the past (see Appendix D). Finally, feature set
all contains 42 features and it is the union of the three feature sets described
above, plus the personal variables in Table 1 and PI(WF) . Every training set
consists of 954 examples (i.e., individual training sessions) corresponding to
80 collective training sessions.
Example 1 (Construction of training dataset) Let us consider a toy dataset of
a portion of the training sessions of a player D = {s6 , s7 , s8 , s9 }, where the
last one is associated with an injury, i.e., the player will get injured during
training session s10 . We construct the training dataset TALL based on feature
set all as follows:
(1) We create a new example in dataset TALL for every of the four train-
ing session in D, by computing the ewma of the 42 player’s workload
features in feature set all. Every example is described by a vector of
length 42, mi = (h1 , . . . , h42 ). All the four vectors compose matrix FALL =
(m1 , m2 , m3 , m4 );
(2) Since the first three training sessions are not associated with injuries, the
first three examples have label 0. In contrast, the last example has label 1
since it is associated with an injury. Therefore, the labels vector is hence c =
(0, 0, 0, 1), indicating that the first three examples are not associated with
an injury while the last training session produces an injury. The training
dataset based on D and feature set all is finally TALL = (FALL , c).
5.2 Experiments
For every feature set – wf, acwr, mswr, all – we construct the corresponding
training datasets as shown in Section 5.1 and we perform two tasks. First, we
perform a feature selection process to determine the most relevant features for
classification. The feature selection process has two motivations: (i) it aims
at reducing the dimensionality of the feature space and the risk of overfitting;
(ii) it allows for an easier interpretation of the machine learning models, due
to the lower number of features [41]. The feature selection task is especially
important for feature set all because it has 42 variables, a large number
considering the small size of the dataset. We use recursive feature elimination
with cross-validation (RFECV) [42] to select the best set of features.1 RFECV
is a wrapper method for feature selection [43] which initially starts by training
a predictive model (a decision tree in our experiments) with all the features
in the feature set. Then at every step, RFECV eliminates one feature, trains
the decision tree on the reduced feature set and calculates the score on the
validation data. The subset of features producing the maximum score on the
validation data is considered to be the best feature subset [42]. On the new
training dataset derived from the feature selection, we train a Decision Tree
classifier (DT).2 We choose DT because it provides a set of decision rules that
are easy to understand. As already stated we are not interested in a “black
box” approach since soccer practitioners are also interested in understanding
why injury happened and what predictors are associated with it.
We hence construct four classification models CWF , CACWR , CMSWR and
CALL . We validate the classifiers with a 3-fold stratified cross-validation strat-
egy [44]: the real dataset is divided into 3 parts or folds and, for each fold,
we use the 10% of the target values as test set, and the remaining 90% as
training set. Each fold is made by preserving the percentage of samples for
each class. Thus, each sample in the dataset was tested once, using a model
that was not fitted with that sample. We also try cross-validation with 5 and
10 folds without finding significant differences in the results presented in this
paper. We measure the goodness of the classifiers by four metrics:
(1) precision: prec = TP/(TP+FP), where TP and FP are true positives and
false positives resulting from classification, where we consider the injury
class (1) as the positive class. Given a class, precision indicates the fraction
of examples that the classifier correctly classifies over the number of all
examples the classifier assigns to that class;
(2) recall: rec = TP/(TP + FN), where FN are false negatives resulting from
classification. Given a class, the recall indicates the ratio of examples of a
given class correctly classified by the classifier;
(3) F-measure: F = 2(prec × rec)/(prec + rec). This measure is the harmonic
mean of precision and recall, which in our case coincides with the square
of the geometric mean divided by the arithmetic mean;
(4) Area Under the Curve (AUC): the probability that a classifier will
rank a randomly chosen positive instance higher than a randomly chosen
negative one (assuming “positive” ranks higher than “negative”). An AUC
1 We use the RFECV function provided by the publicly available Python package
scikit-learn (http://scikit-learn.org/).
2 We use the Python package scikit-learn to train and validate the decision tree.
close to 1 represents an accurate classification, while an AUC close to 0.5

represents a classification close to randomness.
For injury prevention purposes, we are interested in achieving high val-

ues of precision and recall on class 1 (injury). Let us assume that a coach or
an athletic trainer makes a decision about whether or not to “stop” a player
based on the suggestion of the classifier, i.e., to prevent an injury, the play-
ers will skip next training session or game every time the classifier prediction
associated with the player’s current training session is 1 (injury). In this sce-
nario, precision indicates how much we can trust the classifier’s predictions:
the higher the precision, the more a classifier’s predictions are reliable, i.e.,
the probability that the player will actually get injured is high. Trusting a
classifier with low precision means producing many false positives (i.e., “false
alarms”) and frequently stop players unnecessarily, a condition clubs want to
avoid especially for the team’s key players.
Recall indicates the fraction of injuries the classifier detects over the total
number of injuries: the higher the recall the more injuries the classifier can de-
tect. A classifier with low recall detects just a small fraction of all the injuries,
meaning that many players will attend next training session or game and ac-
tually get injured. In other words, a classifier with low recall would misclassify
many actual injuries as non-injuries, i.e., in many cases, the classifier would
classify a risky player as a healthy player.
5.3 Results
The feature selection task selects a small subset of the available features (see
Figure 4a). Although just a small subset of features is selected, we observe
that they cover all the four categories introduced in Table 1 – kinematic,
metabolic, mechanical and personal – indicating that all these aspects are
related to injury risk. In particular, two features are selected in feature sets
wf, acwr and mswr, while three features are selected in feature set all (see
Table 9). Figure 4a also shows the importance of the features in the classifiers,
computed as the mean decrease in Gini coefficient, a measure of how each
variable contributes to the homogeneity of nodes and leaves in the resulting
decision tree [45]. Feature PI is the only one that is selected in every feature
set, accounting for a relative importance ≥ 0.75. Such a high importance of
PI suggests that in our dataset a player’s injury risk increases if he already
experienced injuries in the past. All the other personal features (i.e., age, BMI,
role, play time and games) are discarded by the feature selection task.3
Table 9 shows the performance of classification resulting from our experi-
ments: CALL is the best classifier in terms of both precision and recall, being
3 We observe that the feature selection task improves our predictive performance, since
the decision trees trained on the the entire sets of features (AUC=0.76) produces worse
results than the decision trees trained on the reduced sets of features (AUC=0.88). In this
paper we hence present results with feature selection only for the sake of brevity.
able to detect 76% of the injuries and achieving a 94% precision on the in-
jury class (Figure 4b).4 We observe that all the decision trees constructed
on the four feature sets provide a significant improvement in the prediction
with respect to both the baselines B1 , . . . , B4 and the ACWR and MSWR
predictors, which have a precision close to zero (Table 7).5 The strikingly
good performance of CALL indicates that: (i) the classifier is reliable due to
its high precision, reducing false alarms and hence scenarios where players are
“stopped” unnecessarily before games or training sessions; (ii) the classifier
can detect more than 2/3 of the injuries occurred in the season, which can
lead to a significant saving of economic resources by the club. Decision tree
CWF interestingly achieves the best precision (prec = 1.0), paying the cost
for a low recall rec = 0.33. This classifier could be preferred to CALL if the
club aims at having the absolute certainty that player stoppings are actually
necessary to avoid their future injuries.
We investigate how long it takes for the classifiers – CALL , CWF , CMSWR and
CACWR – to stabilize their predictive performance by training and evaluating
them after every training week wi . In other words, at week wi we use all
the training sessions up to week wi , validating every classifier on the same
data through 3-fold stratified cross-validation. Figure 4c shows the evolution
of the cross-validated F1-score of the five classifiers compared to baseline B4 .
We observe that the classifiers’ performance significantly improve as new data
become available during the season, stabilizing after 16 weeks (Figure 4c).
In contrast, the performance of baseline B4 is constant over time (F1-score
≈ 0.1).
5.4 Evolutive scenario of injury prevention
Here we investigate a scenario where we assume that, at the beginning of a sea-

son, no data are available to a soccer club who aims to use our approach. The
soccer club equips with appropriate sensor technologies and starts recording
training workload data since the first training session of the season. Assuming
that the club train the classifier with new data after every training week, how
many injuries the club can actually forecast? To answer this question we in-
vestigate the evolution of the classifier’s performance during the season, and
how many of the injuries the classifier can actually forecast.
In the evolutive scenario, we proceed from the least recent training week,
w1 , to the most recent one, wn−1 . At training week wi we train the five clas-
sifiers – CALL , CWF , CMSWR , CACWR , and B4 – on data extracted from sessions
w1 , . . . , wi and evaluate their accuracy as the ability in predicting injuries in
week wi+1 . Figure 5 shows the evolution of the F1-score of the five classifiers
as the season goes by, where sessions are grouped by week for visualization
4 the standard deviations in the 3 folds of precision and recall on the injury class are 0.14
and 0.06, respectively.

5 We also implement a Random Forest Classifier and observe that it achieves slightly
worse performance than DT (see Appendix section E).

)HDWXUHVHOHFWLRQ $8&FODVV

7UXH3RVLWLYH5DWH

PI
LPSRUWDQFH

dTOT
dHML
ACC3 $//
DEC2 06:5
$&:5
:)
% 4

:) $&:506:5 $// )DOVH3RVLWLYH5DWH
(a) (b)

$//
06:5
$&:5
:)
%

)VFRUH

ZHHN
(c)
Fig. 4 (a) Relative importance of the features selected by the feature selection task, for
every decision tree. (b) AUC curves of decision trees and baselines for class 1 (injury). (c)
F1-score of classifiers as more data become available during the half season.
purposes. Due to the scarcity of data, the classifiers have a poor predictive
performance at the beginning of the season and miss many injuries (the black
crosses in Figure 5). However, predictive ability improves significantly by time
and classifiers predict most of the injuries in the second half of the season
(the red crosses in Figure 5). The best classifier, CALL , detects in this scenario
11 injuries out of 21, resulting in a cumulative F-score= 0.45 (Figure 5). The
other classifiers – CWF , CMSWR and CACWR – follow a similar pattern as CALL ,
model class prec rec F1 AUC selected features

0 0.98 1.00 0.99
CALL 0.88 PI (WF)
, d(MSWR) , Dec(WF)
decision trees
1 0.94 0.76 0.84 HML 2

0 0.99 0.99 0.99
CMSWR 0.78 PI, dTOT
1 0.80 0.57 0.66
0 0.99 1.00 0.99
CACWR 0.71 PI, dTOT , Acc3
1 0.69 0.43 0.53
0 0.98 1.00 0.99
CWF 0.64 PI, Dec2
1 1.00 0.33 0.47
0 0.99 0.83 0.90
ACWR CACWR
vote 0.65 12 workload features in Table 1
1 0.06 0.48 0.11
0 0.98 0.77 0.86
B4 0.60
1 0.04 0.43 0.07
baselines
0 0.98 0.98 0.98

B1 0.51
1 0.06 0.05 0.05
0 0.98 1.00 0.99
B2 0.50
1 0.00 0.00 0.00
0 0.00 0.00 0.00
B3 0.50
1 0.02 1.00 0.04
Table 4 Performance of classifiers compared to baselines. We report the perfor-

mance of classifiers CALL , CWF , CMSWR , CACWR in terms of precision, recall, F1 and AUC.
We also report for every classifier the features selected by the feature selection task. We
compare the classifier with four baseline B1 , . . . , B4 and predictor CACWR
vote .
in contrast with B4 which does not improve predictive performance as the

season goes by.
6 Interpretation of injury prediction model
Figure 6 is a schematic visualization of the best decision tree resulting from

our experiments, CALL . In the tree there are two types of node: decision nodes
(black boxes) and leaf nodes (green and red boxes). Every decision node has
two branches each indicating the next node to select in the tree depending
on the range of values of the feature associated with the decision node. The
next node to select can be another decision node or directly a leaf node. A leaf
node represents the final prediction of the classifier based on the feature vector
describing a player’s individual training session. There are two possible final
decisions: Injury (red boxes) indicates that the player will get injured in next
game or training session; No-Injury (green boxes) indicates that the player has
no risk to get injury in next game or training session. Given a feature vector
si describing a player’s training session, the prediction associated with si can
be obtained by following the path from the root of the tree down to a leaf
node, through decision nodes. For example, given a player’s training session
with PI(WF) = 0.15, d(MSWR)
HML > 1.80 and DEC2(WF) = 50.20, the decision path

$//

73
)3PLVVHG
FXPXODWLYH)VFRUH
GHWHFWHG
$//
:)
06:5
$&:5
%

ZHHN
Fig. 5 Performance of classifiers in the evolutive scenario. As the season goes by,
we plot week by week the F-score of the classifiers CALL , CWF , CMSWR , CACWR and B4
trained on the data collected up to that week. Black crosses indicate injuries not detect by
CALL , red crosses indicate injures correctly predicted by CALL . For every week we highlight
in red the number of injuries detected by CALL up to that week.
associated with this example is:
PI(WF) >0.14 → PI(WF) >0.40 → d(MSWR)

HML ≤1.36 → d(MSWR)
HML >1.35 → Injury
One of the most useful characteristics of decision trees is that they allow for
the extraction of decision rules which summarize the reasoning of the classifier
when taking specific decisions. In particular, CALL has 6 decision paths asso-
ciated with injuries. We highlight in red these paths in Figure 6a and report
them separately in Figure 6c. The structure of these rules provides us with
useful insights about the reason behind injuries, such as the physical state of
players and their training patterns before the injury. We extract three main
injury scenarios which can be helpful to soccer practitioners to understand the
reasons behind injuries in the considered club:
Previous injuries can lead to new ones.

The six rules derived from CALL highlight that the presence of previous
injuries is crucial to determine whether or not a player will get injured
in the near future. Indeed, feature PI(WF) is present in all the six rules of
Figure 6c, depicting three different scenarios:
1. a player can get injured immediately after its return to regular training
after an injury (Rule 1: PI(WF) ∈(0.14, 0.40]);
2. a player can get injured three days after its return to regular training
after an injury (Rules 2, 4, 5 and 6: PI(WF) >0.40);
Fig. 6 Interpretation of injury prediction model. (a) A schematic visualization of

decision tree CALL . Black boxes are decision nodes, green boxes are leaf nodes for class
No-Injury, red boxes are leaf nodes for class Injury. (b) Mean (AVG) and standard deviation
(STD) of the tree features used in the DT. (c) The six injury rules extracted from CALL .
For every rule we show the range of values of every feature, the frequency and the accuracy
of the rule.
3. a player can get injured the day after its return to regular training after
two previous injuries (Rule 3: PI(WF) >1.15).
From this scenario, it clearly emerges that, in order to prevent future in-
juries, players who return from a recent injury must be observed with
particular attention by the club’s athletic trainers. In particular, the first
three days of training after an injury are the most critical, since 58% of in-
juries occur immediately after a player’s return to regular training (Rule 1).
Special attention to high metabolic workload.

Among the players who got injured once in the past, metabolic workload
d(MSWR)
HML represents a critical value for subsequent injuries. When it is higher
than the seasonal average (d(MSWR)
HML > 1.73, Rule 2), or close to a value in
the 50th percentile range (d(MSWR)
HML ∈ (1.35, 1.55+), Rules 4-6), metabolic
workload is predictive of 31% of the injuries (Rules 2, 4, 5 and 6). A high
metabolic workload can be risky for players that got injured in the past.
This is true for players who have incurred in at least two injuries during
the season (PI(WF) > 1.15).
High mechanical workload is related to injury risk.

In some cases, mechanical workload (DEC2(WF) ) is also predictive of injuries.
In particular, DEC2(WF) has impact on predicting the injuries of players
with a previous injury (DEC2(WF) >52.65 or DEC2(WF) ∈(62.54, 67.88), Rules
2-5) or two previous injuries (DEC2(WF) >58.60, Rule 3). A high mechani-
cal workload higher or close to the seasonal average of DEC2(WF) , in con-
junction with a high metabolic workload (Rules 2-5), identify 37% of the
injuries. Such a result highlights the impact that high stress has, in terms
of mechanical workload, on players who got injured in the past.
7 Discussion of results
The experiments on our dataset produce three interesting results. First, the
best decision tree, CALL , can detect 76% of the injuries with 94% precision,
far better than ACWR, MSWR and the baselines (Table 9). The decision
tree’s false positive rate is indeed small (FPR=0.01, Table 5), indicating that
it infrequently produces “false alarms”, i.e., situations where the classifier pre-
dicts that an injury will happen when it will not. In professional soccer, false
alarms are despicable because the scarcity of players can negatively affect the
performance of a team, both in single games and during the season [2]. CALL
produces false alarms in less than 1% of the cases, while ACWR, MSWR, and
the baselines produce false alarms in >20% of the cases. CALL also produces
a moderate false negative rate (FNR=0.24, Table 5), meaning that situations
where a player that will get injured is classified as out of risk are infrequent
(Table 5). Although in general ACWR, MSWR and the baselines have a false
negative rate lower or comparable to the decision trees (Table 5), they pay the
price in precision producing a high number of false alarms. CALL achieves the
best trade-off between precision and recall indicating that, in contrast with ex-
isting approaches in sports science, a machine learning approach can forecast
most of the injuries while still producing few false alarms.
prediction
1 0
1 16 5 TPR=0.76 FNR=0.24
true
CALL
0 1 902 FPR=0.01 TNR=0.99
1 19 2 TPR=0.90 FNR=0.10
true
CMSWR
dEXP
0 178 725 FPR=0.20 TNR=0.80
1 19 2 TPR=0.90 FNR=0.10
true
CACWR
dTOT
0 514 389 FPR=0.57 TNR=0.43
1 9 12 TPR=0.43 FNR=0.67
true
B4
0 217 686 FPR=0.24 TNR=0.76
Table 5 Confusion matrix of classification. Confusion matrix for CALL , the best pre-
dictors for both MSWR and ACWR approaches (CdEXP and CdTOT , respectively) and
baseline B4 . TNR, TPR, FNR and FPR stays for True Negative Rate, True Positive Rate,
False Negative Rate and False Positive Rate, respectively.
Second, in an evolutive scenario where a soccer club starts collecting the

data for the first time and updates the classifiers after every session as the
season goes by, at the end of the season CALL results in a cumulative F1-
score=0.45 on the injury class (Figure 4):6 while it predicts more than half
of the injuries throughout the season, it also produces many false alarms.
The cumulative performance of the classifiers is highly affected by the initial
period, where both training data and injury data are scarce. This suggests that
trying to prevent injuries since the beginning could not be a good strategy
since classification performance can be initially poor due to data scarcity.
An initial period of data collection is needed in order to collect the adequate
amount of data, and only then reliable classifiers can be trained on the collected
data. The length of the data collection period clearly depends on the needs
and the strategy of the club, including the frequency of training and games,
the frequency of injuries, the number of available players and the tolerated
level of false alarms. Regarding this aspect, in our dataset, we observe that
the performance of the classifiers stabilizes after 16 weeks of data collection
(Figure 4c). Moreover, in Figure 7 we investigate how many weeks of training
we need to achieve a certain level of precision and recall on the injury class. We
6 cumulative precision and cumulative recall on the injury class at the end of the season
are prec = 0.40 and rec = 0.50, respectively.

observe that at least 7 weeks of training are needed to achieve a precision on

the injury class higher than 0.5, while at least 16 weeks of training are needed
for a precision higher than 0.8. In our dataset a reasonable strategy could be
to use the classifiers for injury prevention starting from the 16th week. This
suggests that the considered club can effectively use the classifiers trained on
data from a season to perform injury prediction since the first session of the
next season.
$// $//

06:5 06:5
$&:5 $&:5
:) :)

ZHHN
ZHHN

SUHFLVLRQ UHFDOO
(a) (b)
Fig. 7 Weeks of training vs precision and recall of the classifiers. (a) Number
of weeks of training needed to achieve a given precision. (b) Number of weeks of training
needed to achieve a given recall.
Third remarkable result is that, of 42 features, just 3 features are selected

by the feature selection task for CALL : PI(WF) , d(MSWR)
HML and DEC2(WF) . Feature
PI(WF) , the most important among the three, reflects the temporal distance
between a player’s current training session and the coming back to regular
training of a player who got injured in the past. PI(WF) >0 indicates that the
player has incurred in at least one injury in the past, while PI(WF) > 1 indi-
cates that the player injured at least twice in the past (see Appendix D). We
remind that our dataset cover the second half of a season and just two injuries
refer to players who never got injured before. Presumably, this is the reason
why CALL can detect only injuries happened after a previous one (PI(WF) >0).
More than half of the injuries detected by the classifier (58%) happened im-
mediately after the coming back to regular training of players who got injured
(Rule 1, Figure 6c). Furthermore, 42% of the injuries detected by the classifier
happened long after a previous one and are characterized by specific values
of metabolic and mechanical features, d(MSWR)
HML and DEC2(WF) , which indicate
the metabolic workload variability and the weighted mean of DEC2 in the
previous 6 days (Rules 2-6, Figure 6c). Rule 2, 4 and 5 describe injuries when
players record low levels of workload variability (i.e., high value of d(MSWR)
HML )
and values higher or equal than the seasonal average of DEC2(WF) . Rule 3 de-
scribes injuries when a player has a high value of DEC2(WF) in the last week
and high variability in metabolic features (i.e., low value of d(MSWR)
HML ), more
than two day after the second injury. Finally, Rules 6 detects injuries in day
long after previous injuries when the variability is on a specific values (i.e,
d(MSWR)
HML ∈ (1.35,1.36]).
Our results suggest that the professional soccer club should take care of
the first training sessions of players which come back to regular training after
an injury since in this conditions they are more likely to get injured again.
In days long after the player’s coming back to regular physical activities after
an injury (i.e., more than 3 days), the club should control metabolic workload
(i.e., d(MSWR)
HML ) and mechanical workload (i.e., DEC2(WF) ), which can lead to
injuries at specific values.
8 Conclusion
In this paper, we propose a machine learning approach to injury prediction

for a professional soccer club. To the best of our knowledge, this is the first
time machine learning is successfully applied to this challenging problem in
professional soccer. Due to its high precision and interpretability, the proposed
approach can be a valid support to coaches and athletic trainers, which can use
it to prevent injuries and at the same time to understand the reasons behind
the occurrences of these events. Our approach can be easily implemented for
every club that performs automatic data collection with the standard sensing
technologies available nowadays.
Our work can be extended in many interesting directions. First, in our
experiments we extract a limited set of workload features widely accepted in
the sports science literature. A way to enlarge our description of a player’s
health is to include performance features extracted from official games, where
the player is exposed to high physical and psychological stress. When infor-
mation about match workload is available, we expect physical effort features
to be more relevant in the classification since games are the highest intensity
demand in a week and considerably impact the general fitness of a player.
Second, we plan to investigate the “transferability” of our approach from a
club to another, i.e., if a classifier trained on a set of players can be successfully
applied to a distinct set of players, not used during the training process. If so,
it would be possible to exploit collective information to train a more powerful
classifier which includes training examples from different players, clubs, and
leagues. Finally, if data covering several seasons of a player’s activity are avail-
able, a specific classifier can be trained for every player to monitor his physical
behavior. The quality of our approach could greatly benefit from the availabil-
ity of other types of data measuring different aspects of health, such as heart
rate, ventilation, and lactate, as new sensor technologies are available today
to measure these specific quantities. When these data will become available,
they will allow us to refine our study on the relation between performance and
injury risk.
In the meanwhile, experiences like ours may contribute to shaping the
discussion on how to forecast injuries by a multidimensional approach based
on automatically collected features of training workload and performance, that
are starting to be massively available everywhere in the soccer industry. As
we show in this paper, we have the potential of creating easy-to-use tools in
support of soccer practitioners. This is crucial since the decisions of managers
and coaches, and hence the success of clubs, also depend on what they measure,
how good their measurements are, the quality of predictions and how well
predictions are understood.
A Descriptive statistics of workload features
Table A shows the average (AVG) and the standard deviation (SD) of the distributions of
the 12 training workload features considered in our study. We assess the normality of the
distributions by using the Shapiro-Wilks’ Normality test (SW) and observe that none of them
is normally distributed (see Table A). Indeed, by a visual inspection of the distributions, we
observe that they tend to be bimodal and right skewed (Figure 8).
AVG SD SW
dT OT 3882.94 1633.21 <0.01
dHSR 133.22 66.41 <0.01
dM ET 1151.99 694.25 <0.01
dHM L 543.89 339.64 <0.01
dHM L /min 8.70 6.09 <0.01
dEXP 410.67 221.29 <0.01
Acc2 64.26 31.72 <0.01
Acc3 16.16 10.97 <0.01
Dec2 62.44 33.09 <0.01
Dec3 19.14 12.78 <0.01
DSL 117.98 78.52 <0.01
FI 0.63 0.31 <0.01
SD = Standard Deviation; SW = Shapiro-Wilks’ Normality test.
Table 6 Descriptive statistics of the 12 training workload features. We provide three cate-
gories of training workload features: kinematic features (blue), metabolic features (red) and
mechanical features (green).

dTOT dHSR
dMET
)UHTXHQF\

dHML dHML /m dEXP
)UHTXHQF\Q

Acc2 Acc3 Dec2
)UHTXHQF\

Dec3 DSL FI
)UHTXHQF\

Fig. 8 Distribution of workload features. We provide three categories of training work-

load features: kinematic features (blue), metabolic features (red) and mechanical features
(green).
B ACWR with data-driven groups
We repeat the ACWR approach by using the quintiles of the ACWR distribution (ACWRq )
instead of predefined ACWR groups suggested by Murray et al. [29]. Figure 9 shows the
injury likelihood (IL) for every ACWR group in all the 12 workload features. We observe that
groups with low ACWRq are associated to the highest injury risk, substantially confirming
the experiments made using predefined ACWR groups. We construct ACWRq predictors
following the same strategy described in the manuscript but using quantiles instead of pre-
defined groups. Table 7 visualizes the results of classification. We observe results similar to
those presented in the manuscript for the pre-defined ACWR groups: the predictors are a
little usable in practice due to their too low precision.

$&:5 q
dTOT dHSR dMET
,QMXU\OLNHOLKRRG

dHML dHML /m dEXP
,QMXU\OLNHOLKRRG

Acc2 Acc3 Dec2
,QMXU\OLNHOLKRRG

Dec3 DSL FI
,QMXU\OLNHOLKRRG

Fig. 9 Injury likelihood in ACWRq groups. The plots show IL for the ACWR groups
defined the quantiles of the distribution, for every of the 12 training workload features
considered in our study. We provide three categories of training workload features: kinematic
features (blue), metabolic features (red) and mechanical features (green).
C Exponential Weighted Moving Average (EWMA)
To consider the recent training workload of a player, we compute the exponential weighted
moving average (ewma) of his most recent training sessions. The ewma decreases exponen-
tially the weights of the values according to their recency [46, 47], i.e., the more recent a value
is the more it is weighted in an exponential function according to a decay α = 2/(span − 1).
In accordance with the exponential function the moving average is computed as:
EWMAt = α[xt − (xt−1 + (1 − α)2 xt−2 + . . . + (1 − α)n−1 xt−n )] + xt .

We vary span = 1, . . . , 10 to detect the value leading to the best classification perfor-
mance. We hence train a decision tree on the feature set all by using every of the ten span
values. Figure 10 shows the cross-validated AUC and F1-score of the decision tree CALL
varying the value of span. We observe that a 6 training span is the best predictive window
to injury prediction in our dataset (Figure 10).
D Computation of PI
To take into account the injury history of a player, we compute the EWMA of the number of
injuries in previous weeks. PI(WF) reflects the temporal distance between a player’s training
session and the return of the player to regular trianing after an injury. PI(WF) = 0 represents
ACWRq class prec rec F1 AUC

0 0.98 0.80 0.88
CdT OT 0.59
1 0.04 0.38 0.08
0 0.99 0.37 0.54
CdHSR 0.57
1 0.03 0.76 0.05
0 0.98 0.80 0.88
CdM ET 0.59
1 0.04 0.38 0.08
0 0.99 0.81 0.89
CdHM L 0.67
1 0.06 0.52 0.10
0 0.98 0.81 0.89
CdHM L /m
/min 0.62
1 0.05 0.43 0.09
0 0.98 0.80 0.88
CdEXP 0.57
1 0.04 0.33 0.07
0 0.98 0.80 0.88
CAcc2 0.59
1 0.04 0.38 0.08
0 0.98 0.80 0.64
CAcc3 0.54
1 0.03 0.29 0.06
0 0.98 0.80 0.88
CDec2 0.57
1 0.04 0.33 0.07
0 0.99 0.46 0.63
CDec3 0.66
1 0.04 0.86 0.07
0 0.98 0.80 0.88
CDSL 0.59
1 0.04 0.38 0.08
0 0.98 0.80 0.88
CF I 0.57
1 0.04 0.33 0.07
0 0.99 0.21 0.34
CACWRq
one 0.56
1 0.02 0.91 0.05
0 0.99 0.83 0.90
CACWRq
vote 0.64
1 0.06 0.46 0.10
0 0.98 1.00 0.99
CACWRq
all
0.50
1 0.00 0.00 0.00
0 0.98 0.98 0.98
B1 0.51
1 0.06 0.05 0.05
0 0.98 1.00 0.99
B2 0.50
1 0.00 0.00 0.00
0 0.00 0.00 0.00
B3 0.50
1 0.02 1.00 0.04
0 0.98 0.77 0.86
B4 0.60
1 0.04 0.43 0.07
Table 7 Injury prediction report of ACWRq . We report precision (prec), recall (rec),
F1-score (F1) and Area Under the Curve (AUC) for the injury class and the non-injury
class for all the predictors defined on ACWR and monotony methodologies. We also provide
predictive performance of four baseline predictors B1 , B2 , B3 and B4 .
players who never got injured in the past. PI(WF) > 0 represents players who get injured et
least once in the past. Table 8 provides specific PI(WF) thresholds in players incurred from 1
to 4 previous injuries. For example, PI(WF) = 0.50 reflects a training performed by a player
3 days since his return to regular training after an injury.
$//

VFRUH

$8&
)VFRUH

VSDQRI(:0$
Fig. 10 We plot the AUC and F1-score of EWMA with span = 1, . . . , 10 in CALL . The
red line reflects the best span to injury prediction.
PIi
injuries 1 2 3 4 5 6+
1 0.29 0.49 0.64 0.74 0.81 > 0.86
2 1.27 1.48 1.63 1.74 1.81 > 1.86
3 2.27 2.46 2.62 2.72 2.80 > 2.85
4 3.25 3.46 3.53 3.66 3.76 > 3.83
Table 8 PI(WF) values after n training days (i.e., n = 1...6) since the return of a player to
regular training. We report the values for different n of previous injuries (i.e., n = 1...4).
PIi is the number of training days long after players return to regular physical activity. 6+
indicates values for 6 and more than 6 days.
E Predictions using Random Forest
In addition to the Decision Tree (DT) provided in the manuscript, we also train an Extra
Tree Random Forest Classifier (ETRFC) to assess the predictive power of ensemble meth-
ods.7 ETRFC is a robust model that fits a number of randomized decision trees on various
sub-samples of the dataset, controlling the possible over-fitting. However, a ETRFC is more
complex to interpret due to its complexity. Table 9 shows the performance of ETRFC classi-
fier trained on wf, acwr, mon, all. We use the same 3-fold stratified cross-validation used
for DT. We observe that the performance of ETRFC is slightly worse than DT’s perfor-
mance. Again, CALL is the best classifier in terms of both precision and recall, being able to
detect 78% of the injuries and achieving 88% precision. Moreover, the classifiers outperform
both the baselines and predictors based on ACWR and MSWR.
Acknowledgements This work has been partially funded by the EU project SoBigData
grant n. 654024.
7 We use the Python package scikit-learn to train and validate the decision tree.
model class prec rec F1 AUC selected features

0 0.99 0.99 0.99
CALL 0.86 PI (WF)
, d(MON) , Dec(WF)
random forest
1 0.88 0.78 0.82 HML 3

0 0.99 1.00 0.99
CMON 0.76 PI, dTOT
1 1.00 0.52 0.67
0 0.99 1.00 0.99
CACWR 0.73 PI, dTOT , Acc3
1 0.90 0.43 0.58
0 0.98 1.00 0.99
CWF 0.63 PI, Dec3
1 1.00 0.29 0.44
0 0.99 0.83 0.90
ACWR CACWR
vote 0.65 12 workload features in Table 1
1 0.06 0.48 0.11
0 0.98 0.77 0.86
B4 0.60
1 0.04 0.43 0.07
baselines
0 0.98 0.98 0.98

B1 0.51
1 0.06 0.05 0.05
0 0.98 1.00 0.99
B2 0.50
1 0.00 0.00 0.00
0 0.00 0.00 0.00
B3 0.50
1 0.02 1.00 0.04
Table 9 Performance of ETRFC and baseline classifiers. The goodness of classifiers

and baselines in terms of recall, precision, F1 and AUC. For every classifier we also show
the features selected by the feature selection task.
Authors contribution A.R. and L.P. designed and performed all the experiments, A.R.
made the plots and the figures. M.I. collected the GPS data and the injury data. A.R., L.P.,
J.F., D.M. and P.C. contributed to writing the manuscript.
References
1. Hägglund M, Waldén M, Magnusson H, Kristenson H, Bengtsson H and Exstrand J.

Injuries affect team performance negatively in professional football: an 11-year follow-up
of the UEFA Champions League injury study. British Journal of Sports Medicine, doi:
10.1136/bjsports-2013-092215, 2013.
2. Hurley OA. Impact of Player Injuries on Teams’ Mental States, and Subsequent Perfor-
mances, at the Rugby World Cup 2015. Frontiers in Psychology 7:807, doi: 10.3389/fp-
syg.2016.00807, 2016.
3. Lehmann EE, Schulze GG. What Does it Take to be a Star? – The Role of Performance
and the Media for German Soccer Players. Applied Economics Quarterly 54:1, pp.
59-70, doi: 10.3790/aeq.54.1.59, 2008.
4. Fernández-Cuevas I., Gomez-Carmona P, Sillero-Quintana M, Noya-Salces J, Arnaiz-
Lastras J, Pastor-Barrón A. Economic costs estimation of soccer injuries in first and
second Spanish division professional teams. 15th Annual Congress of the European
College of Sport Sciences ECSS, 23th 26th june. 2010.
5. Gudmundsoon H, Horton M. Spatio-Temporal Analysis of Team Sports - A Survey.
CoRR: abs/1602.06994, 2016.
6. Stein M., Janetzko H., Seebacher D., Jäger A., Nagel M., Hölsch, J., Grossniklaus M.
How to Make Sense of Team Sport Data: From Acquisition to Data Modeling and
Research Aspects. Data, 2:1, 2, doi:10.3390/data2010002, 2017.
7. Cintia P., Pappalardo L., Pedreschi D. Engine Matters: A First Large Scale Data Driven
Study on Cyclists’ Performance. 13th IEEE International Conference on Data Mining
Workshops, pp. 147–153, doi:10.1109/ICDMW.2013.41, 2013.
8. Cintia P., Coscia M., Pappalardo L. The Haka network: Evaluating rugby team per-
formance with dynamic graph analysis. 2016 IEEE/ACM International Conference on
Advances in Social Networks Analysis and Mining (ASONAM), pp. 1095–1102, doi:
10.1109/ASONAM.2016.7752377, 2016
9. Cintia P., Pappalardo L., Pedreschi D., Giannotti F., Malvaldi M. The harsh rule
of the goals: Data-driven performance indicators for football teams. 2015 IEEE In-
ternational Conference on Data Science and Advanced Analytics (DSAA), pp. 1–10,
doi:10.1109/DSAA.2015.7344823, 2015
10. Pappalardo L., Cintia P. Quantifying the relation between performance and success in
soccer. CoRR: abs/1705.00885, 2017.
11. Fernández J, Medina D, Gómez A, Arias M, Gavaldá R. From Training to Match
Performance: A Predictive and Explanatory Study on Novel Tracking Data. 16th IEEE
International Conference on Data Mining Workshops, doi:10.1109/ICDMW.2016.0027,
2016
12. Rossi A., Savino M., Perri E., Aliberti G., Trecroci A., Iaia M. Characterization of
in-season elite football trainings by GPS features: The Identity Card of a Short-Term
Football Training Cycle. 16th IEEE International Conference on Data Mining Work-
shops, pp. 160-166, doi: 10.1109/ICDMW.2016.0030, 2016.
13. Brink MS1, Visscher C, Arends S, Zwerver J, Post WJ, Lemmink KA. Monitoring
stress and recovery: new insights for the prevention of injuries and illnesses in elite
youth soccer players. Br J Sports Med. 2010;44: 809-15.
14. Ehrmann FE, Duncan CS, Sindhusake D, Franzsen WN, Greene DA. GPS and injury
prevention in professional soccer. J Strength Cond Res. 2015;30:306-307.
15. Venturelli M, Schena F, Zanolla L, Bishop D. Injury risk factors in young soccer players
detected by a multivariate survival model. Journal of Science and Medicine in Sport.
2011;14:293298.
16. Kirkendall D.T., Dvorak J. Effective Injury Prevention in Soccer. The physician and
sportsmedicine, 38:1, doi: http://dx.doi.org/10.3810/psm.2010.04.1772, 2010.
17. Hagglund M, Walden M, Bahr R, Ekstrand J. Methods for epidemiological study of
injuries to professional football players: developing the UEFA model. British Journal of
Sports Medicine, 39:6, pp. 340–346, doi:10.1136/bjsm.2005.018267, 2005.
18. FIFA circular no. 1494 of 8 July 2015, http://bit.ly/2pBmhK7.
19. Gabbett TJ. Reductions in pre-season training loads reduce training injury rates in
rugby league players. British Journal of Sports Medicine. 2004;38: 743749.
20. Gabbett TJ, Jenkins DG. Relationship between training load and injury in professional
rugby league players. Journal of Science and Medicine in Sport. 2011;14: 204209.
21. Gabbett TJ. Influence of training and match intensity on injuries in rugby league.
Journal of Sports Sciences. 2004;22(5):409-417.
22. Gabbett TJ. The development and application of an injury prediction model for non-
contact, soft-tissue injuries in elite collision sport athletes. The Journal of Strength &
Conditioning Research. 2010;24(10):2593-2603.
23. Gabbett TJ, Domrow N. Relationships between training load, injury, and fitness in
sub-elite collision sport athletes. Journal of Sports Sciences. 2007;25(13):1507-1519.
24. Anderson L, Triplett-McBride T, Foster C, Doberstein S, Brice G. Impact of training
patterns on incidence of illness and injury during a women’s collegiate basketball season.
The Journal of Strength & Conditioning Research. 2003; 17: 734738.
25. Gabbett TJ, Ullah S. Relationship between running loads and soft-tissue injury in elite
team sport athletes. J Strength Cond Res. 2012;26: 953960.
26. Rogalski B, Dawson B, Heasman J, Gabbett TJ. Training and game loads and injury
risk in elite Australian footballers. J Sci Med Sport. 2013;16: 499503.
27. Gabbett TJ. The training-injury prevention paradox: should athletes be training
smarter and harder? Br J Sports Med. 2016; bjsports-2015-095788.
28. Hulin BT, Gabbett TJ, Blanch P, Chapman P, Bailey D, Orchard JV. Spikes in acute
workload are associated with increased injury risk in elite cricket fast bowlers. Br J
Sports Med. 2014;48:708-712.
29. Murray NB, Gabbett TJ, Townshend AD, Blanch P. Calculation acute:chronic workload
ratios using exponential weighted moving averages provides a more sensitive indicator
of injury likelihood than rolling averages. Br J Sports Med. 2016; bjsports-2016-097152.
30. Foster C. Monitoring training in athletes with reference to overtraining syndrome. Med
Sci Sports Exerc. 1998;30:11641168.
31. Talukder H, Vincent T, Foster G, Hu C, Huerta J, Kumar A, et al. Preventing in-game
injuries for NBA players. MIT Sloan Analytics Conference. Boston; 2016.
32. Buchheit M, Al Haddad H, Simpson BM, Palazzi D, Bourdon PC, Di Salvo V, Mendez-
Villanueva A. Monitoring accelerations with GPS in football: time to slow down?
International Journal of sports physiology and performance, 9 (3), pp. 442-445, doi:
10.1123/ijspp.2013-0187, 2014.
33. Medina D, Pons E, Gomez A, Guitart M, Martin A, Vazquez-Guerrero J, Camen-
forte I, Carles B, Font R. Are There Potential Safety Issues Concerning the Safe Us-
age of Electronic Personal Tracking Devices? The Experience of a Multi-sport Elite
Club. International Journal of sports physiology and performance, 4, pp. 1-12, doi:
10.1123/ijspp.2016-0368, 2017.
34. Duncan MJ, Badland HM, Mummery WK. Applying GPS to enhance understanding of
transport-related physical activity. Journal of Science and Medicine in Sport. 2009;12:
549556.
35. Hulin BT, Gabbett TJ, Lawson DW, Caputi P, Sampson JA. The acute:chronic work-
load ratio predicts injury: high chronic workload may decrease injury risk in elite rugby
league players Br J Sports Med. 2016;50:231-236.
36. Bowen L., Gross AS, Gimpel M, Li FX. Accumulated workloads and the acute:chronic
workload ratio relate to injury risk in elite youth football players. British Journal of
Sports Medicine, 51, 452-459, doi:10.1136/bjsports-2015-095820, 2017.
37. Menaspá P. Are rolling averages a good way to assess training load for injury prevention?
British Journal of Sports Medicine, 51, doi:http://dx.doi.org/10.1136/bjsports-2016-
096131, 618-619, 2017.
38. Fuller CW, Ekstrand J, Junge A, Andersen TE, Bahr R, Dvorak J, Hägglund M, Mc-
Crory P, Meeuwisse WH. Consensus statement on injury definitions and data collection
procedures in studies of football (soccer) injuries. British Journal of Sports Medicine,
40, pp. 193-201, doi:http://dx.doi.org/10.1136/bjsm.2005.025270, 2006.
39. Di Prampero PE, Fusi S, Sepulcri L, Morin JB, Belli A, Antonutto G. Sprint running:
a new energetic approach. J Exp Biol. 2005;208: 28092816.
40. Gaudino P, Iaia FM, Alberti G, Strudwick AJ, Atkinson G, Gregson W. Monitoring
training in elite soccer players: systematic bias between running speed and metabolic
power data. Int J Sports Med. 2013;34: 963968.
41. James G, Witten D, Hastie T, Tibshirani R. An Introduction to Statistical Learning.
New York, NY: Springer New York; 2013.
42. Guyon I, Weston J, Barnhill S, Vapnik V. Gene Selection for Cancer Classification Using
Support Vector Machines. Machine Learning 46, 2002, 10.1023/A:1012487302797.
43. Li J., Cheng K., Wang S., Morstatter F., Trevino R., Jiliang T., Huan L. Feature
Selection: A Data Perspective. arXiv:1601.07996, 2016
44. Friedman J, Hastie T, Tibshirani R. The elements of statistical learning. Springer series
in Statistics, volume 1, Springer, Berlin, 2001.
45. Hastie TJ and Tibshirani RJ and Friedman JH The elements of statistical learning: data
mining, inference, and prediction. Springer series in statistics, isbn: 978-0-387-84857-0,
2009
46. Lowry CA, Woodall WH, Champ CW, Rigdon SE. A Multivariate Exponentially
Weighted Moving Average Control Chart. Technometrics. Technometrics. 1992;34: 4653.
47. Lucas JM, Saccucci MS. Exponentially Weighted Moving Average Control Schemes:
Properties and Enhancements. Technometrics. 1990;32: 112.
View publication stats

Effective Injury Prediction in Professional Soccer

Uploaded by

Copyright:

Available Formats

Effective Injury Prediction in Professional Soccer

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Effective Injury Prediction in Professional Soccer

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Preprint · May 2017

Alessio Rossi Luca Pappalardo

SEE PROFILE SEE PROFILE

Paolo Cintia F. Marcello Iaia

SEE PROFILE SEE PROFILE

thermal imaging in face asymmetry studies View project

The user has requested enhancement of the downloaded file.

Alessio Rossi · Luca Pappalardo · Paolo

Abstract Injuries have a great impact on professional soccer, due to their

Injuries of professional athletes have a great impact on sports industry, due to

interpret by soccer practitioners. We consider injury prediction as the problem

for injury risk estimation and construct related predictors. In Section 5 we

The relationship between training workload, sports performance and injury

Talukder et al. [31] apply a machine learning approach to predict injuries

3 Data collection and feature extraction

We set up a study on twenty-six professional players of a professional soccer

4 Classic approaches to injury risk estimation

In this section, we replicate the two most used state-of-the-art approaches to

4.1 Acute Chronic Workload Ratio (ACWR)

Training workload features [34, 39, 40]

variable describing workload with a monodimensional approach: just one vari-

4.2 Mean Standard deviation Workload Ratio (MSWR)

Another widely used approach to injury risk estimation is MSWR, defined in

Table 2 Performance of ACWR and MSWR predictors. We report precision (prec),

risk for the majority of workload features, substantially confirming results

MSWR predictor. As done for ACWR, we explore the usability in practice of

 dTOT dHSR dMET

 dHML dHML /m dEXP

 Dec3 DSL FI

5 A machine learning approach to injury prediction

Given the inadequacy of the ACWR and MSWR predictors, we construct a

5.1 Construction of training dataset

d TOT d EXP ... ACC 3 label

s1 4,018.19 426.42 ... 16.99 0

s2 3,465.81 326.41 ... 16.91 0

s3 3,227.15 256.85 ... 18.25 1

sn 3,199.58 273.69 ... 19.64 1

We construct four different training datasets based on four different feature

features in Table 1. We use ewma since, as suggested in literature [29, 31],

close to 1 represents an accurate classification, while an AUC close to 0.5

For injury prevention purposes, we are interested in achieving high val-

5.4 Evolutive scenario of injury prevention

Here we investigate a scenario where we assume that, at the beginning of a sea-

and 0.06, respectively.

worse performance than DT (see Appendix section E).

)HDWXUHVHOHFWLRQ  $8&FODVV

model class prec rec F1 AUC selected features

1 0.94 0.76 0.84 HML 2

0 0.98 0.98 0.98

Table 4 Performance of classifiers compared to baselines. We report the perfor-

in contrast with B4 which does not improve predictive performance as the

6 Interpretation of injury prediction model

Figure 6 is a schematic visualization of the best decision tree resulting from

associated with this example is:

PI(WF) >0.14 → PI(WF) >0.40 → d(MSWR)

Previous injuries can lead to new ones.

Fig. 6 Interpretation of injury prediction model. (a) A schematic visualization of

dTOT dHSR dMET

dHML dHML /m dEXP

Dec3 DSL FI

)HDWXUHVHOHFWLRQ $8&FODVV

$// $//