TITLE: The use of statistical learning to predict health and performance outcomes in Servicewomen.
SESSION DESCRIPTION: This session will explore the use of supervised and unsupervised statistical learning methods to predict health and performance outcomes in Servicewomen. The session will provide an overview of statistical learning techniques, as well as their merits and limitations, with a focus on models used both for inference and prediction; techniques for model validation will also be discussed. A range of health and performance outcomes relevant to Servicewomen will be highlighted. The session aims to highlight the evidence where statistical learning prediction models can be used to highlight individual risk and improve outcomes in Servicewomen as well as to identify where research is lacking. Case studies where prediction models have been successfully implemented to help inform military decision making and / or aid performance will also be provided. The aim of the session will be to improve understanding of how statistical learning techniques can be used to improve the health and performance of Servicewomen.
BACKGROUND: Servicewomen are at increased risk of musculoskeletal injuries and, on average, have poorer physical performance than their male counterparts. With the recent opening of combat roles to women in the United Kingdom, United States, Australia, and other nations, women are employed in more physically arduous roles and are required to complete the most physically arduous military courses. Arduous training courses may come with increased risk of musculoskeletal injuries and other health outcomes including menstrual disturbances. These roles also require greater levels of physical performance. The ability to understand predictive factors and / or make individual predictions for success in a training course, meeting a physical employment standard, or developing a health outcome will help militaries better manage the health and performance of Servicewomen. Statistical learning methods have long been used for statistical inference and the same models can also be used to predict outcomes in new observations.
SPEAKERS AND PRESENTATION TITLES:
- Prof A Scientist, Chair, British Army [20 minutes]. Statistical learning methods to predict physical employment standards performance from routine physical performance tests.
- Dr J Smith, US Army [20 minutes]. Factors affecting menstrual function in Servicewomen: a statistical learning approach for identifying menstrual dysfunction.
- Prof J Doe, British Army [20 minutes]. A comparison of machine learning classification models to predict arduous training course success in men and women.
- Dr NE Body, University College London [ 20 minutes]. The use of statistical learning to predict musculoskeletal injuries during basic military training.
- Discussion [10 minutes].
MILITARY IMPACT: With the employment of women expanding across many militaries, coupled with recent increased interest in statistical learning methods, this session is a highly topical thematic session. Militaries typically collect routine health and performance data across selection, training, and through career that could be used to better manage the health and performance of Service personnel; statistical learning methods provide one means to gain insight from these data. The impact of this work extends far beyond military employment, with implications for athlete, arduous occupation, and wider female, populations. Practical recommendations on risk factors for, and the use of prediction models to detect, health and performance outcomes will be presented within each talk based on the best available evidence. Where evidence is lacking, areas for novel future research will be identified.
SUBMIT THEMATIC SESSION