A Novel Multilocus Genetic Model Can Predict Muscle Fibers Composition

The Development of a Personalised Training Framework: Implementation of Emerging Technologies for Performance

“…With an abundance of information available to the coach, recent research has focused on being able to better utilise this information to underpin decision making via prediction, either in terms of injury risk [ 193 , 194 ], postinjury recovery times [ 195 ], physiology (such as muscle fibre type [ 196 ]), training loads and fatigue [ 197 , 198 ], talent identification [ 199 ] and training plans [ 200 , 201 ]. These approaches utilise a variety of different statistical modelling techniques, including simple data analysis with a hold-out set for validation (e.g.…”

“…Similar single data input models have been utilised in sport. For example, Borisov and colleagues [ 196 ] utilised a 14 SNP model to predict muscle fibre type in 55 subjects, with a Receiver Operating Characteristic (ROC) of 81% for professional athletes, demonstrating strong concordance with muscle biopsies. Such a finding could be very useful within elite sport because muscle biopsy testing is highly invasive, limiting its use, while genetic testing is non-invasive.…”

“…As a result of the Larruskain and colleagues [ 194 ] study, it is clear that the use of individual pieces of data is likely insufficient in the prediction of complex phenotypes and outcomes, such as injury, while it perhaps is sufficient for less complex phenotypes, as demonstrated by Borizov et al [ 196 ], who used genetic information to predict muscle fibre type with success. Data mining refers to the conversion of raw data—such as that collected by the various technological and testing practices utilised in elite sport—to information which can then be analysed [ 202 ].…”

“…The identification of large numbers of SNPs that affect a given trait, along with the creation of Total Genotype Scores (TGS) for that trait, will likely improve the predictive accuracy of genetic information. However genetics will only ever serve as part of the picture: it allows us to understand predisposition, which we can use to predict outcomes—and, as we have seen, serve as a useful, but incomplete, input to statistical models [ 194 , 196 ]—but the addition of additional pieces of information, explored within this paper, should enhance the personalisation process.…”