Clinical application of monitoring indicators of female dancer health, including application of artificial intelligence in female hormone networks

Keay, N.; Lanfear, M.; Francis, G.

Abstract Objectives The purpose of this study was to assess the effectiveness of monitoring professional female dancer health with a variety of subjective and objective monitoring methods, including application of artificial intelligence (AI) techniques to modelling menstrual cycle hormones and delivering swift personalised clinical advice. Methods Female dancers from a ballet company completed a published online dance-specific health questionnaire. Over the study period, dancers recorded wellbeing and training metrics, with menstrual cycle tracking and blood tests. For menstrual cycle hormones AI-based techniques modelled hormone variation over a cycle, based on capillary blood samples taken at two time points. At regular, virtual, clinical interviews with each dancer, findings were discussed, and personalised advice given. Results 14 female dancers (mean age 25.5 years, SD 3.7) participated in the study. 10 dancers recorded positive scores on the dance health questionnaire, suggesting a low risk of relative energy deficiency in sport (RED-S). 2 dancers were taking hormonal contraception. Apart from 1 dancer, those not on hormonal contraception reported current eumenorrhoeic status. The initiative of monitoring menstrual cycles and application of AI to model menstrual cycle hormones found that subclinical hormone disruption was occurring in 6 of the 10 dancers reporting regular cycles. 4 of the 6 dancers who received personalised advice, showed improved menstrual hormone function, including one dancer who had planned pregnancy. Conclusions Multimodal monitoring facilitated delivery of prompt personalised clinical medical feedback specific for dance. This strategy enabled the early identification and swift management of emergent clinical issues. These innovations received positive feedback from the dancers.