©Journal of Sports Science and Medicine (2018) 17, 402-408

http://www.jssm.org

` Research article

Modelling the HRV Response to Training Loads in Elite Rugby Sevens Players

Sean Williams 1, Stephen West 1, Dan Howells 2, Simon P.T. Kemp 2, Andrew A. Flatt 3
and Keith
Stokes 1
1
Department for Health, University of Bath, Bath, UK; 2 Rugby Football Union, Twickenham, UK; 3 Department of
Health Science & Kinesiology, Georgia Southern University, Savannah, USA

al., 1975). Based on these relatively simple assumptions,

the Banister impulse-response model can explain a sub-
Abstract
A systems modelling approach can be used to describe and opti-
stantial proportion (over 90% in some cases) of the vari-
mise responses to training stimuli within individuals. However, ance in performance data (Busso, 2003; Morton, 1997;
the requirement for regular maximal performance testing has pre- Wood et al., 2005). However, a major limitation of the
cluded the widespread implementation of such modelling ap- Banister model and its extensions is the requirement for
proaches in team-sport settings. Heart rate variability (HRV) can frequent maximal performance tests to accurately deter-
be used to measure an athlete’s adaptation to training load, with- mine model parameters (Jobson et al., 2009). The use of
out disrupting the training process. As such, the aim of the current regular maximal performance testing is especially difficult
study was to assess whether chronic HRV responses, as a repre- in team-sport settings with weekly competitive fixtures,
sentative marker of training adaptation, could be predicted from
given their potential to cause additional fatigue (Nédélec et
the training loads undertaken by elite Rugby Sevens players.
Eight international male players were followed prospectively
al., 2013). To date, the need for regular performance testing
throughout an eight-week pre-season period, with HRV and train- has limited the broader use of the Banister model in team-
ing loads (session-RPE [sRPE] and high-speed distance [HSD]) sports.
recorded daily. The Banister model was used to estimate vagally- Heart rate variability (HRV) is a popular tool for
mediated chronic HRV responses to training loads over the first monitoring wellness and training adaptation in athletes
four weeks (tuning dataset); these estimates were then used to (Bellenger et al., 2016). In particular, the parasympathetic
predict chronic HRV responses in the subsequent four-week pe- activity of the autonomic nervous system, typically repre-
riod (validation dataset). Across the tuning dataset, high correla- sented by the square root of the mean sum of the squared
tions were observed between modelled and recorded HRV for
differences between R–R intervals (rMSSD) component of
both sRPE (r = 0.66 ± 0.32) and HSD measures (r = 0.69 ± 0.12).
Across the sRPE validation dataset, seven of the eight athletes
HRV, has been shown to correlate well with variations in
met the criterion for validity (typical error <3% and Pearson r performance within both cross-sectional (Kenney, 1985),
>0.30), compared to one athlete in the HSD validation dataset. and longitudinal studies (Chalencon et al., 2015) across
The sRPE validation data produced likely lower mean bias values, multiple sports. In addition, HRV measures are associated
and most likely higher Pearson correlations, compared to the HSD with overuse injury risk (Williams et al., 2017) and, more
validation dataset. These data suggest that a systems theory ap- broadly, markers of global health (Adamson et al., 2004;
proach can be used to accurately model chronic HRV responses Kiviniemi et al., 2007). As such, rMSSD may be an appro-
to internal training loads within elite Rugby Sevens players, priate representative parameter to describe an athlete’s
which may be useful for optimising the training process on an
stress-recovery status (Chalencon et al., 2012), especially
individual basis.
given the ease and non-intrusive nature of its collection.
Key words: Cardiac parasympathetic function, monitoring, train- Indeed, the emergence of smartphone applications and
ing load. technologies has dramatically increased the accessibility of
HRV measurement, such that it can now be recorded accu-
rately using only a smartphone device (Plews et al., 2017).
Introduction Recent work in competitive swimmers has demonstrated
that HRV measures may be used as a viable substitute for
Several research groups have applied systems theory ap- performance measurements for the mathematical model-
proaches to quantify and describe responses to physical ling of training effects (Chalencon et al., 2012), and could
training (Busso and Thomas, 2006; Morton, 1997; Mujika therefore be used to optimally plan and monitor training
et al., 1996). The Banister impulse-response model esti- strategies in an individualised manner (Chalencon et al.,
mates performance at a given time to be the difference be- 2015). However, this is yet to be applied and evaluated in
tween the ‘fitness’ and ‘fatigue’ effects of prior training a team sport context where, as stated previously, regularly
loads (Banister et al., 1975). The five adjustable parameters monitoring changes in performance is inherently more
within the model (initial performance level; two time con- complex than in individual, endurance-based sports.
stants that describe fitness and fatigue decay rates; and two Rugby Sevens is a format of Rugby Union that has
gain parameters that describe how daily training impulses grown in popularity in recent years, and is now included in
determine the amplitude in fitness and fatigue effects) are the Summer Olympic Games. The contact and collision
calibrated against measured performance data to provide events that are inherent to Rugby Sevens, alongside the
individualised training response information (Banister et high physiological demands (Higham et al., 2014), means

Received: 16 Nisan 2018 / Accepted: 30 May 2018 / Published (online): 14 August 2018
Williams et al. 403

that the risk of injury associated with the sport is relatively then calculated and used in further analyses, as a repre-
high (Fuller et al., 2010). In particular, the injury incidence sentative parameter of chronic training adaptation
rate associated with elite Rugby Sevens training is substan- (Chalencon et al., 2015). The Ln rMSSD42-exp calculation
tially higher than the 15-a-side game (West et al., 2017), was initiated with the mean Ln rMSSD value observed
which is likely a result of the high training loads that are across the first seven days of the monitoring period.
necessary to meet the physiological demands of competi- The validity of the Elite HRV application for com-
tion. Therefore, the careful monitoring and management of puting Ln rMSSD was established by comparing simulta-
player workloads on an individual basis is of critical im- neous 60 s recordings of the same tools used in this study
portance, in order to protect players from the negative con- (i.e., Polar H7 Bluetooth heart rate strap and application)
sequences of training whilst increasing their performance with an electrocardiograph (Biopac MP100, Goletta, Cali-
capacity and resilience (Gabbett, 2016). Moreover, a con- fornia, USA) among 10 collegiate athletes. Procedures and
sideration of the most appropriate load measures (e.g., in- comparison methods from a previous study were replicated
ternal versus external) for this setting is also required. Ac- (Esco et al., 2017). Measures of Ln rMSSD were acquired
cordingly, the aim of the current study was to assess in the supine, seated and standing position for each indi-
whether chronic HRV responses, as a representative vidual. Differences between supine (Elite HRV = 3.70 ±
marker of training adaptation, could be predicted from the 0.43 ms, ECG = 3.70 ± 0.43 ms) seated (Elite HRV = 3.44
training loads undertaken by elite Rugby Sevens players. ± 0.62 ms, ECG = 3.43 ± 0.59 ms) and standing (Elite HRV
In addition, we sought to compare the effectiveness of in- = 2.84 ± 0.52 ms, ECG = 2.85 ± 0.52 ms) measures were
ternal (session rating of perceived exertion [sRPE]) versus not significant (p = 0.80, 0.52 and 0.49, respectively) and
external (total high speed running distance [HSD]) load the standardized differences were considered trivial (≤ 0.03
measures for this purpose. for each). The correlations between the application and
ECG were near perfect (r = 0.99, p < 0.05 for each posi-
Methods tion). Additionally, upper and lower limits of agreement
were tight (upper and lower limits = 0.03 ms to -0.03 ms
Study design for supine, 0.08 ms to -0.10 ms for seated and 0.13 ms
Eight male international Rugby Sevens players (mean ± to -0.10 ms for standing). These data demonstrate that the
SD; age: 27 ± 4 y, height: 1.86 ± 0.07 m, body mass: validity of the Elite HRV application for computing Ln
93.2 ± 8.6 kg) were followed prospectively throughout an rMSSD is consistent across supine, standing, and seated
eight-week pre-season period that was undertaken in prep- positions. Seated measurements were used within the cur-
aration for the 2016-17 World Rugby Sevens Series. The rent study to ensure a consistent approach throughout the
priority during this phase was to develop central adapta- study period.
tions through the use of extensive intervals, with a linear
increase in intensity. The average weekly sRPE and HSD Training load
loads across this period were 2947 ± 941 AU and 3389 ± Internal training loads were recorded for all sessions using
892 m, respectively. This eight week pre-season period was the sRPE method (Foster, 1998). This approach has been
chosen as each parameter in the model was likely to be em- shown to be a valid method for estimating exercise inten-
phasised across this preparation phase (Clarke and Skiba, sity across multiple training modalities (Herman et al.,
2013), and periods of 60-90 days are recommended for the 2006). Player ratings of perceived exertion were recorded
mathematical modelling of training and performance, after 30 min after completing a given session, and were then
which parameters should be reset (Banister, 1991). The multiplied by the session duration (mins) to provide a sRPE
study was conducted in accordance with the principles of value in arbitrary units.
the Declaration of Helsinki (World Medical Association, External training load was represented by the total
2013) and a local university research ethics committee pro- high-speed distance (distance covered at speeds greater
vided ethical approval. than 5 m/s [HSD]) undertaken during pitch-based sessions,
recorded using global positioning system (GPS) devices
Measures (STATports® Viper Pod, 10 Hz single constellation). In
Heart rate variability Rugby Union, 5 m/s is the most commonly used threshold
Athletes were instructed to perform a 90 second HRV that corresponds to high-speed running (Clarke et al.,
measurement each morning upon waking whilst breathing 2015). The HSD measure was chosen to reflect the high-
spontaneously in a seated position (Esco and Flatt, 2014). intensity nature of the sport’s demands (Suarez-Arrones et
A Polar H7 Bluetooth heart rate strap (Polar Electro, al., 2012), which is considered an important quality for per-
Kempele, Finland) paired with a freely available formance in Rugby Sevens (Higham et al., 2012).
smartphone application (Elite HRV, Ashville, North Caro-
lina, USA) were used for daily HRV acquisition. The Data analysis
rMSSD was the HRV measure used for analysis, as this has The mathematical relationship between training loads (sys-
been demonstrated to have greater reliability than other tem input) and Ln rMSSD42-exp (system output) was mod-
spectral indices (Al Haddad et al., 2011). The rMSSD data elled for each athlete via the two-component impulse-re-
were log-transformed (Ln) to reduce non-uniformity of er- sponse model (Banister et al., 1975). The model is charac-
ror (Plews et al., 2012). The 42-day exponentially- terized by two gain terms (k1 and k2), two time constants
weighted average of this variable (Ln rMSSD42-exp) was (τ1 and τ2), and an initial performance level (p):
404 Modelling HRV response in Rugby Sevens

n 1 n 1 ranging from -0.08 to 0.96. The HSD tuning dataset pro-

pˆ n  p  k1  w ie ( n  i)/1  k 2  w ie (n  i)/ 2 duced a similar model fit (r = 0.69 ± 0.12), with r values
i1 i1 ranging from 0.57 to 0.92. Figure 1 shows the best-fitting
The model parameters were determined by mini- individual model (r = 0.96 for sRPE tuning dataset) for one
mizing the Residual Sum of Squares (RSS) between esti- athlete.
mated and measured Ln rMSSD42-exp within a customised
spreadsheet (Microsoft Excel, Microsoft, Redmond, USA) Accuracy of model predictions using validation dataset
(Clarke and Skiba, 2013). Model parameters were estab- The accuracy of the model parameters (estimated from the
lished using data collected across the first four-week train- tuning dataset) in predicting Ln rMSSD42-exp responses
ing block (tuning dataset). As per Chalencon et al. (2015), across the validation period are displayed in Table 1. The
the term ‘tuning dataset’ was used in this study over the CV was less than 3% for all eight athletes across the sRPE
more customary expression ‘training dataset’ to avoid am- validation dataset, and seven of the eight athletes across the
biguity when discussing physical training. The estimates HSD validation dataset. Seven of the eight athletes had at
obtained from the tuning dataset were then used to predict least moderate positive relationships between their pre-
Ln rMSSD42-exp responses across the subsequent four-week dicted and recorded chronic HRV responses across the
training block (validation dataset). The accuracy of models sRPE validation dataset (range: -0.24 to 0.78), compared to
for predicting chronic HRV responses was assessed using one athlete within the HSD validation dataset (range: -0.87
an Excel spreadsheet (Hopkins, 2015) designed to calcu- to 0.33). Overall, seven of the eight athletes met the crite-
late the mean bias, typical error of the estimate (in raw rion for validity (CV <3% and Pearson r >0.30) for the
units, and expressed as a coefficient of variation [CV, %]), sRPE validation data, compared to one athlete in the HSD
and Pearson correlation coefficient. The magnitude of cor- validation dataset.
relation was defined as trivial (<.10), low (.10-.29), mod-
erate (.30-.49), high (.50-.69), very high (.70-.89), or Comparison of prediction accuracy between load
nearly perfect (.90-.99) (Hopkins, 2015). A CV of less than measures
3% and Pearson correlation >0.30 (moderate) was set as The sRPE validation data produced likely lower mean bias
the criterion for validity, based on the established smallest values, and most likely higher Pearson correlations, com-
worthwhile change for Ln rMSSD (Buchheit, 2014). Dif- pared to the HSD validation dataset (Table 1). The sRPE
ferences in bias and precision between the internal training models also produced lower typical errors compared to the
load model (sRPE) and external training load model (HSD) HSD models, but these differences were unclear.
were compared using standardised differences, and were
interpreted as: <0.20, trivial; 0.20-0.59, small; 0.60-1.19, Discussion
moderate; 1.20-1.99, large; >2.00, very large (Hopkins,
2010). Magnitude-based inferences were used to provide The primary purpose of this study was to assess whether
an interpretation of the real-world relevance of the out- chronic Ln rMSSD responses, as a representative marker of
comes. A value equivalent to a standardised difference in training adaptation, could be predicted from the training
means of 0.20 was set as the smallest worthwhile effect loads undertaken by elite Rugby Sevens players. Across
threshold (Hopkins, 2010). Effects were classified as un- the tuning dataset, high correlations were observed be-
clear if the percentage likelihood that the true effect tween modelled and measured HRV data. When model pa-
crossed both positive and negative smallest worthwhile ef- rameters estimated from the tuning dataset were used to
fect thresholds were both greater than 5%. Otherwise, the predict future responses to training loads, seven of the eight
effect was deemed clear, and was qualified with a proba- athletes met the criterion for validity (CV <3% and Pearson
bilistic term using the following scale: <0.5%, most un- r >0.30) for the sRPE data, compared to one athlete in the
likely; 0.5-5%, very unlikely; 5-25%, unlikely; 25-75%, HSD validation dataset. The sRPE validation data pro-
possible; 75-95%, likely; 95-99.5%, very likely; >99.5%, duced likely lower mean bias values, and most likely higher
most likely (Hopkins, 2010). Pearson correlations, compared to the HSD validation da-
taset.
Results In the present study, the goodness-of-fit between
modelled and measured chronic HRV responses across the
Estimates of model parameters tuning period (r = 0.66 and 0.69 for sRPE and HSD model,
The mean ± SD values of the gain (k1 and k2) and time de- respectively) was lower than that obtained by Chalencon et
cay constants (τ1 and τ2) for the sRPE tuning dataset were al. (2015) in competitive swimmers (r = 0.93). Chalencon
0.0000767 ± 0.0000767 AU, 0.0000893 ± 0.0000716 AU, et al.’s (2015) use of the high-frequency component of
20 ± 14 d, and 11 ± 7 d, respectively. For the HSD tuning HRV, as opposed to Ln rMSSD used in the current study,
set, the corresponding values were 0.000278 ± 0.000271 may account for the improved relationships, although Ln
AU, 0.000276 ± 0.000263 AU, 32 ± 38 d, and 35 ± 43 d, rMSSD has been proposed as the “most reliable and prac-
respectively. tically applicable measure for day-to-day monitoring”
(Plews et al., 2013). In addition, Chalencon et al. (2015)
Model fit using tuning datasets used weekly measures of nocturnal HRV, as opposed to the
For the sRPE tuning dataset, the Banister model produced daily measurements used in the present study. Rolling av-
high correlations between modelled and actual Ln erages of Ln rMSSD data may represent a more meaningful
rMSSD42-exp (r = 0.66 ± 0.32), with individual r values assessment of any change in cardiac autonomic nervous
Williams et al. 405

system balance, compared with a single day value (Plews their training time at speeds below the blood lactate accu-
et al., 2012). A minimum of three valid data points per mulation threshold (Mujika et al., 1995). Conversely,
week are required when calculating rolling averages Rugby Sevens training is more varied with respect to both
(Plews et al., 2014), which supports the use of daily meas- the intensity and modalities of training used (Higham et al.,
urements in the current study. Perhaps more importantly, 2016), and so chronic HRV responses are likely to be less
there were considerable differences between the nature of predictable in comparison to sports with less varied train-
the athletes used in these studies; swimming is an endur- ing stimuli.
ance-based sport in which athletes spend the majority of

Figure 1. Application of the Banister impulse-response model to athlete #4. The left column pertains to sRPE training load
data, the right column represents HSD training load data. Charts (A) and (B) display the daily training loads undertaken across
the study period. Charts (C) and (D) display the fit between modelled and measured chronic HRV responses. Charts (E) and
(F) display the fitness and fatigue influences on HRV. Charts (G) and (F) display the residual differences between measured
and modelled HRV across the study period. Data to the left of the dashed vertical line relate to the tuning dataset, whilst data to the right of this
line relate to the (unseen) validation dataset.

Table 1. Comparison of mean ± SD typical error (raw and %), Pearson r, and mean bias between the predicted and recorded
chronic HRV responses across the sRPE and HSD validation datasets.
Measure sRPE dataset HSD dataset Effect size (90% CIs) Inference
Typical error (ms) 0.03 ± 0.01 0.08 ± 0.15 0.50 (-0.39 to 1.39) Unclear
Typical error (%) 0.60 ± 0.27 1.84 ± 3.56 0.50 (-0.38 to 1.37) Unclear
Pearson r 0.45 ± 0.34 -0.13 ± 0.40 -1.63 (-2.56 to -0.69) Most likely ↑
Mean bias (%) 0.09 ± 0.72 -1.22 ± 1.46 -1.14 (-2.13 to -0.15) Likely ↓

The model predictions produced using the valida- ferences in the utility of the parameter estimates obtained
tion dataset resulted in typical errors that were <3%, and from the tuning dataset for predicting subsequent chronic
Pearson correlations that were moderate and small for the HRV responses. For those athletes with at least moderate
sRPE and HSD validation datasets, respectively. The range positive relationships and acceptably small (<3%) typical
of r values across both validation datasets (sRPE: -0.24 to errors (n = 7 in the sRPE validation, and n = 1 in the HSD
0.78; HSD: -0.87 to 0.33) implies large inter-individual dif- validation), the tuning dataset could be used to predict
406 Modelling HRV response in Rugby Sevens

future chronic HRV responses to training loads with satis- That being said, workloads are more easily prescribed via
factory accuracy, and thus be used to optimise their train- external load measures (e.g., by setting a target HSD for a
ing on an individualised basis. For instance, the parameters given day), and so external load measures remain im-
obtained from the Banister model could be used to simulate portant for planning training programmes in this setting
the effects of different periodization schemes (Clarke and (Gabbett et al., 2017).
Skiba, 2013), to objectively plan training progressions for A limitation of the current study is the lack of a true
athletes rehabilitating from injury (Clarke and Skiba, ‘performance’ measure, against which changes in HRV
2013), or to individualise the ‘fitness’ and ‘fatigue’ time- could be validated as a marker of training adaptation.
decay constants within acute:chronic workload calcula- Whilst there is a wealth of evidence to support the fact that
tions (Carey et al., 2017). In addition, the Banister model the parasympathetic activity of the autonomic nervous sys-
could be used to produce individual influence curves that tem is a good indicator of an athlete’s adaptation to training
may inform the optimal taper strategy for each athlete lead- loads (Adamson et al., 2004; Chalencon et al., 2015;
ing into a Sevens tournament (Fitz-Clarke et al., 1991). Gisselman et al., 2016; Williams et al., 2017), this remains
Such influence curves can be created using freely available to be shown in Rugby Sevens athletes. Although it is diffi-
spreadsheets (Clarke and Skiba, 2013). However, the opti- cult to define a single performance indicator for Rugby
mal HRV response to training overload and pre-competi- Sevens, as performance is dependent on numerous physi-
tion tapers in elite athletes is yet to be fully understood cal, tactical, psychological, and environmental factors
(Plews et al., 2013). To date, the need for regular maximal (Higham et al., 2012), the ‘critical velocity’ model may
performance testing has limited the many potential uses of provide a useful framework against which the HRV-
the Banister model in team sports, but the use of HRV data performance relationship could be validated in future stud-
as a surrogate measure of training adaptation may facilitate ies (Jones and Vanhatalo, 2017). The ‘critical velocity’
the practical application of the Banister model in these set- threshold represents a running velocity that can (theoreti-
tings. cally) be maintained indefinitely, whilst a W' constant rep-
Athletes for whom the predictive capacity across resents the finite work capacity available to an athlete at
the validation period was low, may have experienced velocities greater than their critical velocity threshold
changes to their life stressors since the tuning data period (Jones and Vanhatalo, 2017). These parameters can be es-
that influenced their subsequent chronic HRV responses to timated from a single three minute all-out exercise test
training. HRV is known to be influenced by a wide range (Burnley et al., 2006). Given the significance of high-speed
of factors, including physiological/pathological, neuropsy- running ability to performance in Rugby Sevens (Higham
chological, non-modifiable, lifestyle and environmental et al., 2012), these constants are likely to be of substantial
factors (Fatisson et al., 2016). Thus, chronic HRV re- importance to overall performance in this setting. Moreo-
sponses that diverge from the predicted response to train- ver, the widespread use of GPS units in elite Rugby Sevens
ing stimuli compared to a baseline period may serve as a could theoretically enable the dynamic modelling of W' uti-
useful (and objective) ‘flag’ for the investigation of life lisation during training and matches (Jones and Vanhatalo,
stressors and lifestyle factors in that athlete (Gabbett et al., 2017).
2017). However, this concept requires further evaluation
via the inclusion of ‘life stress/wellbeing’ measures in fu- Conclusion
ture studies.
In the present study, sRPE data produced more ac- These data demonstrate that a systems theory approach can
curate predictions of future HRV responses when com- be used to describe the variation in chronic HRV responses
pared to HSD data obtained from GPS devices, with likely to training within elite Rugby Sevens players, and thus may
lower levels of bias and most likely higher Pearson corre- be used to optimise training responses in this setting. For
lations observed between predicted and measured re- the majority of athletes in the sRPE validation dataset, the
sponses. As stated above, HRV responses may be influ- modelling of training effects also allowed for the accurate
enced by a range of lifestyle and/or environmental factors prediction of future responses to training stimuli. Re-
(e.g., sleep quality) (Burton et al., 2010), which can also sponses that diverged from the ‘tuning dataset’ predictions
influence the sRPE internal load measure produced in re- may serve as a useful flag for the investigation of life
sponse to a given external load (Impellizzeri et al., 2004). stressors. The sRPE training load measure provided more
In contrast, external load measures will not be influenced accurate predictions of future HRV responses compared to
by such factors, and are instead primarily determined by an external load measure (HSD). The mathematical mod-
the workloads prescribed by coaching staff (Impellizzeri et elling of HRV responses to training loads may enable prac-
al., 2005). In addition, the sRPE method enables the cap- titioners to more accurately assess and optimise the train-
ture of loads undertaken across all training modalities (e.g., ing process.
gym or pool-based sessions), whereas HSD could only be
recorded for pitch-based sessions. As such, it is perhaps Acknowledgements
The authors would like to acknowledge with considerable gratitude all
unsurprising that the sRPE measure outperformed the ex- those who volunteered to take part in this study. All authors contributed
ternal load measure in predicting HRV responses to train- to data collection and manuscript preparation. No funding to declare. The
ing impulses. These findings add to existing literature re- authors have no conflicts of interest to declare. All experiments comply
garding the importance of monitoring and controlling ath- with the current laws of the country.
letes’ internal training loads, to ensure they are receiving
an appropriate training stimulus (Impellizzeri et al., 2004).
Williams et al. 407

References Gabbett, T.J. (2016) The training-injury prevention paradox: Should

athletes be training smarter and harder? British Journal of Sports
Adamson, P.B., Smith, A.L., Abraham, W.T., Kleckner, K.J., Stadler, Medicine 50, 273-280.
R.W., Shih, A. and Rhodes, M.M. (2004) Continuous autonomic Gabbett, T.J., Nassis, G.P., Oetter, E., Pretorius, J., Johnston, N., Medina,
assessment in patients with symptomatic heart failure: D., Rodas, G., Myslinski, T., Howells, D. and Beard, A. (2017)
prognostic value of heart rate variability measured by an The athlete monitoring cycle: a practical guide to interpreting
implanted cardiac resynchronization device. Circulation 110, and applying training monitoring data. British Journal of Sports
2389-2394. Medicine 15, 1451-1452.
Al Haddad, H., Laursen, P., Chollet, D., Ahmaidi, S. and Buchheit, M. Gisselman, A.S., Baxter, G.D., Wright, A., Hegedus, E. and Tumilty, S.
(2011) Reliability of resting and postexercise heart rate (2016) Musculoskeletal overuse injuries and heart rate
measures. International Journal of Sports Medicine 32, 598-605. variability: Is there a link? Medical Hypotheses 87, 1-7.
Banister, E. (1991) Modeling elite athletic performance. Physiological Herman, L., Foster, C., Maher, M., Mikat, R. and Porcari, J. (2006)
testing of elite athletes 403-424. Validity and reliability of the session RPE method for monitoring
Banister, E., Calvert, T., Savage, M. and Bach, T. (1975) A systems model exercise training intensity: original research article. South
of training for athletic performance. Australian Journal of Sports African Journal of Sports Medicine 18, 14-17.
Medicine 7, 57-61. Higham, D.G., Hopkins, W.G., Pyne, D.B. and Anson, J.M. (2014)
Bellenger, C.R., Fuller, J.T., Thomson, R.L., Davison, K., Robertson, Relationships between rugby sevens performance indicators and
E.Y. and Buckley, J.D. (2016) Monitoring athletic training status international tournament outcomes. Journal of Quantitative
through autonomic heart rate regulation: a systematic review and Analysis in Sports 10, 81-87.
meta-analysis. Sports Medicine 46, 1461-1486. Higham, D.G., Pyne, D.B., Anson, J.M. and Eddy, A. (2012) Movement
Buchheit, M. (2014) Monitoring training status with HR measures: do all patterns in rugby sevens: effects of tournament level, fatigue and
roads lead to Rome? Frontiers in physiology 5, 1-19. substitute players. Journal of Science and Medicine in Sport 15,
Burnley, M., Doust, J.H. and Vanhatalo, A. (2006) A 3-min all-out test to 277-282.
determine peak oxygen uptake and the maximal steady state. Higham, D.G., Pyne, D.B., Anson, J.M., Hopkins, W.G. and Eddy, A.
Medicine and Science in Sports and Exercise 38, 1995-2003. (2016) Comparison of activity profiles and physiological
Burton, A., Rahman, K., Kadota, Y., Lloyd, A. and Vollmer-Conna, U. demands between international rugby sevens matches and
(2010) Reduced heart rate variability predicts poor sleep quality training. The Journal of Strength & Conditioning Research 30,
in a case–control study of chronic fatigue syndrome. 1287-1294.
Experimental Brain Research 204, 71-78. Hopkins, W.G. (2010) Linear models and effect magnitudes for research,
Busso, T. (2003) Variable dose-response relationship between exercise clinical and practical applications. Sportscience 14, 49-57.
training and performance. Medicine and Science in Sports and Hopkins, W.G. (2015) Spreadsheets for analysis of validity and
Exercise 35, 1188-1195. reliability. Sportsscience 19, 36-42.
Busso, T. and Thomas, L. (2006) Using mathematical modeling in Impellizzeri, F.M., Rampinini, E., Coutts, A.J., Sassi, A. and Marcora,
training planning. International Journal of Sports Physiology S.M. (2004) Use of RPE-based training load in soccer. Medicine
and Performance 1, 400-405. and Science in Sports and Exercise 36, 1042-1047.
Carey, D.L., Blanch, P., Ong, K.-L., Crossley, K.M., Crow, J. and Morris, Impellizzeri, F.M., Rampinini, E. and Marcora, S.M. (2005)
M.E. (2017) Training loads and injury risk in Australian Physiological assessment of aerobic training in soccer. Journal
football—differing acute: chronic workload ratios influence of Sports Sciences 23, 583-592.
match injury risk. British Journal of Sports Medicine 51(16), Jobson, S.A., Passfield, L., Atkinson, G., Barton, G. and Scarf, P. (2009)
1215-1220. The analysis and utilization of cycling training data. Sports
Chalencon, S., Busso, T., Lacour, J.-R., Garet, M., Pichot, V., Connes, P., Medicine 39, 833-844.
Gabel, C.P., Roche, F. and Barthélémy, J.C. (2012) A model for Jones, A.M. and Vanhatalo, A. (2017) The ‘Critical Power’concept:
the training effects in swimming demonstrates a strong Applications to sports performance with a focus on intermittent
relationship between parasympathetic activity, performance and high-intensity exercise. Sports Medicine 47, 65-78.
index of fatigue. PLoS ONE 7, e52636. Kenney, W.L. (1985) Parasympathetic control of resting heart rate:
Chalencon, S., Pichot, V., Roche, F., Lacour, J.-R., Garet, M., Connes, P., relationship to aerobic power. Medicine and Science in Sports
Barthélémy, J.C. and Busso, T. (2015) Modeling of performance and Exercise 17, 451-455.
and ANS activity for predicting future responses to training. Kiviniemi, A.M., Tulppo, M.P., Wichterle, D., Hautala, A.J., Tiinanen,
European Journal of Applied Physiology 115, 589-596. S., Seppänen, T., Mäkikallio, T.H. and Huikuri, H.V. (2007)
Clarke, A.C., Anson, J. and Pyne, D. (2015) Physiologically based GPS Novel spectral indexes of heart rate variability as predictors of
speed zones for evaluating running demands in Women’s Rugby sudden and non-sudden cardiac death after an acute myocardial
Sevens. Journal of Sports Sciences 33, 1101-1108. infarction. Annals of Medicine 39, 54-62.
Clarke, D.C. and Skiba, P.F. (2013) Rationale and resources for teaching Morton, R. (1997) Modelling training and overtraining. Journal of Sports
the mathematical modeling of athletic training and performance. Sciences 15, 335-340.
Advances in physiology education 37, 134-152. Mujika, I., Busso, T., Lacoste, L., Barale, F., Geyssant, A. and Chatard,
Esco, M.R. and Flatt, A.A. (2014) Ultra-short-term heart rate variability J.-C. (1996) Modeled responses to training and taper in
indexes at rest and post-exercise in athletes: evaluating the competitive swimmers. Medicine and Science in Sports and
agreement with accepted recommendations. Journal of Sports Exercise 28, 251-258.
Science & Medicine 13, 535-541. Mujika, I., Chatard, J.-C., Busso, T., Geyssant, A., Barale, F. and Lacoste,
Esco, M.R., Flatt, A.A. and Nakamura, F.Y. (2017) Agreement between L. (1995) Effects of training on performance in competitive
a smartphone pulse sensor application and electrocardiography swimming. Canadian Journal of Applied Physiology 20, 395-
for determining lnRMSSD. The Journal of Strength & 406.
Conditioning Research 31, 380-385. Nédélec, M., McCall, A., Carling, C., Legall, F., Berthoin, S. and Dupont,
Fatisson, J., Oswald, V. and Lalonde, F. (2016) Influence diagram of G. (2013) Recovery in soccer. Sports Medicine 43, 9-22.
physiological and environmental factors affecting heart rate Plews, D.J., Laursen, P.B., Kilding, A.E. and Buchheit, M. (2012) Heart
variability: an extended literature overview. Heart International rate variability in elite triathletes, is variation in variability the
11, e32. key to effective training? A case comparison. European Journal
Fitz-Clarke, J.R., Morton, R. and Banister, E. (1991) Optimizing athletic of Applied Physiology 112, 3729-3741.
performance by influence curves. Journal of Applied Physiology Plews, D.J., Laursen, P.B., Le Meur, Y., Hausswirth, C., Kilding, A.E.
71, 1151-1158. and Buchheit, M. (2014) Monitoring training with heart-rate
Foster, C. (1998) Monitoring training in athletes with reference to variability: How much compliance is needed for valid
overtraining syndrome. Medicine and Science in Sports and assessment? International Journal of Sports Physiology &
Exercise 30, 1164-1168. Performance 9, 783-790.
Fuller, C.W., Taylor, A. and Molloy, M.G. (2010) Epidemiological study Plews, D.J., Laursen, P.B., Stanley, J., Kilding, A.E. and Buchheit, M.
of injuries in international rugby sevens. Clinical Journal of (2013) Training adaptation and heart rate variability in elite
Sport Medicine 20, 179-184. endurance athletes: opening the door to effective monitoring.
408 Modelling HRV response in Rugby Sevens

Sports Medicine 43, 773-781. AUTHOR BIOGRAPHY

Plews, D.J., Scott, B., Altini, M., Wood, M., Kilding, A.E. and Laursen, Sean WILLIAMS
P.B. (2017) Comparison of heart rate variability recording with
smart phone photoplethysmographic, Polar H7 chest strap and
Employment
electrocardiogram methods. International Journal of Sports Lecturer, Department for Health, Univer-
Physiology and Performance 1-17. sity of Bath, Bath, United Kingdom.
Suarez-Arrones, L.J., Nuñez, F.J., Portillo, J. and Mendez-Villanueva, A. Degree
(2012) Running demands and heart rate responses in men rugby PhD
sevens. The Journal of Strength & Conditioning Research 26, Research interests
3155-3159. Injury surveillance, injury prevention,
West, S., Williams, S., Cross, M., Howells, D., Mobed, R., Kemp, S. and training load monitoring, applied statis-
Stokes, K. (2017) Workload spikes combined with high
cumulative load is associated with increased injury risk in elite
tics.
rugby sevens players. British Journal of Sports Medicine 51, E-mail: [email protected]
408-408. Stephen WEST
Williams, S., Booton, T., Watson, M., Rowland, D. and Altini, M. (2017) Employment
Heart rate variability is a moderating factor in the workload- Ph.D. student, Department for Health,
injury relationship of competitive CrossFit athletes. Journal of University of Bath, Bath, United Kingdom
Sports Science & Medicine 16, 443-449. Degree
Wood, R.E., Hayter, S., Rowbottom, D. and Stewart, I. (2005) Applying
MSc
a mathematical model to training adaptation in a distance runner.
European Journal of Applied Physiology 94, 310-316. Research interests
World Medical Association. (2013) World Medical Association Injury surveillance, injury prevention,
Declaration of Helsinki. Ethical principles for medical research training load monitoring, applied statis-
involving human subjects. JAMA 310, 2191-2194. tics.
E-mail: [email protected]
Dan HOWELLS
Key points Employment
Strength and conditioning coach for Eng-
 A systems theory approach can be used to describe the land Sevens, Rugby Football Union, Lon-
variation in chronic HRV responses to training within don, UK.
Degree
elite Rugby Sevens players. BSc
 For the majority of athletes, model parameters can be Research interests
used to accurately predict future responses to training Strength and conditioning, injury preven-
stimuli. tion, training load monitoring.
 Responses that diverge from the predicted values may E-mail: [email protected]
serve as a useful flag for the investigation of changes Simon KEMP
in lifestyle factors. Employment
 Internal training load measures (sRPE) markedly out-
Medical Services Director, Rugby Foot-
ball Union, London, UK.
performed external load measures (HSD) in predict-
Degree
ing future HRV responses to training stimuli.  MA, MB, MRCGP, MSc(SEM), FFSEM
Research interests
Sports medicine, injury prevention, train-
 Sean Williams ing load monitoring.
Department for Health, University of Bath, Bath, BA2 7JY E-mail: [email protected]
United Kingdom Andrew FLATT
Employment
Assistant Professor & Graduate Program
Director, Department of Health Science &
Kinesiology, Georgia Southern Univer-
sity, Savannah, USA.
Degree
PhD
Research interests
Heart rate variability, Sports injuries,
training load monitoring.
E-mail: [email protected]
Keith STOKES
Employment
Professor, Department for Health, Univer-
sity of Bath, Bath, United Kingdom,
Degree
PhD
Research interests
Injury surveillance, injury prevention,
training load monitoring, applied physiol-
ogy.
E-mail: [email protected]