Construction And Empirical Study Of Recovery Model Of Customized Cold-Water Immersion After Exercise Based On XGBoost Algorithm

Background:Cold water immersion has been widely used in sports practice as an effective measure to relieve exercise-induced fatigue.However,there is still no evidence of developing the optimal cold water immersion protocol as a commonly used recovery method.The human body is a complex system,and individual characteristics can affect the recovery effect of cold water immersion.Therefore,the optimal cold water immersion model should be customized.Objective:The overall aim of this study was to construct a model of customized cold water immersion recovery after exercise.In order to achieve this overall objective,this study was divided into four sub-studies,and the sub-study objectives were as follows:[Study 1]To explore the effects of different cold water immersion durations on recovery after exercise.[Study 2]To construct a customized cold water immersion(CCWI)recovery model based on different recovery purposes after exercise.[Study 3]To verify the applicability of the CCWI recovery model after exercise in a hot environment.[Study 4]Iterate and optimize the post-exercise CCWI recovery model based on different recovery purposes.seated restMethods:[Study 1]170 male exercise enthusiasts were randomly assigned to either seated recovery or 5～20 minutes of cold water immersion(13℃)after incremental load cycling to exhaustion.The effects of different cold water immersion durations on recovery after exercise were compared by measuring the cardiovascular system,body temperature,subjective scales,nervous system,and performance indicators.[Study 2]According to the data of Study 1,combined with the influencing factors of the cold water immersion effect,the XGBoost algorithm of machine learning was used to establish a CCWI recovery model after exercise based on three different purposes:endurance recovery,lower limb explosive power recovery,and parasympathetic nerve activation.[Study 3]Ten male exercise enthusiasts were randomly crossed over to seated recovery(CON),general cold water immersion recovery(GCWI),and three CCWI recovery with different recovery purposes(CCWI-EE,CCWI-VJP,CCWIRMSSD)after incremental load cycling to exhaustion in a hot environment(35-38℃).The recovery effect of three CCWI models after exercise in a hot environment was explored by measuring the cardiovascular system,body temperature,subjective scales,nervous system,and performance indicators.The metabolic characteristics of CWI promoting recovery were explored through the non-targeted metabolomics analysis of urine.[Study 4]The 40 pieces of cold water immersion data after exercise in a hot environment obtained in Study 3 were included in the CCWI recovery model constructed in Study Ⅱ,and the training samples and test samples were extended to iteratively and optimize the three post-exercise CCWI recovery models for different purposes.Results:[Study 1]Heart rate(HR),systolic blood pressure(SBP),mean arterial pressure(MAP),blood lactic acid(LAC),ear temperature(Te),RPE,and thermal sensation(TS)increased significantly(P<0.05).The HR of the control group remained significantly higher than the baseline value at 55 min after exercise(P<0.05);all cold water immersion groups had no significant difference from the baseline value at 55 min after exercise(P>0.05).Mean skin temperature(Tms)in all cold water immersion groups was significantly lower than baseline at 55 min after exercise(P<0.05).The rating of perceived exertion(RPE)of the control,8min,and 9min cold water immersion groups were significantly lower than the baseline after 25 min of exercise(P<0.05).There was no significant difference in the TS of the control group and the 17-min cold water immersion group between 25 min after exercise and baseline(P>0.05),and the other groups were significantly lower than the baseline(P<0.05).The RMSSD,SDNN,and SDSD of the 9min,14min,and 17min cold water immersion groups post-test were significantly higher than pre-test(P<0.05).The TP,HF,and LF of subjects in the 11min and 17min cold water immersion groups post-test were significantly higher than the pre-test(P<0.05).The second incremental load cycling duration was significantly lower than the first(P<0.05).There was no significant difference in the second vertical jump height between the first and the second in the control and 6-minute cold water immersion groups(P>0.05).The second vertical jump height of the other groups was significantly lower than that of the first(P<0.05).There was no significant difference in the second maximum vertical jump height between the first and the second in the 6-minute cold water immersion groups(P>0.05).The other groups’ second maximum vertical jump height was significantly lower than that of the first(P<0.05).The body fat ratio could affect the mean skin temperature(P<0.05).[Study 2]The results of the prediction model for the duration of incremental load cycling showed that the characteristic variables with significant influence included the pre-test duration of incremental load cycling,basal metabolic rate,muscle mass,age and body age,and the coefficient of determination of the test sample was R2=0.868.The vertical jump height prediction model results showed that the significant influence characteristics were pre-measured vertical jump height,immersion time,environment temperature,and thermal sensation,R2=0.850.The results of the RMSSD prediction model showed that the significant influence characteristics were pre-measured RMSSD,muscle mass,immersion duration,and basal metabolic rate,R2=0.279.[Study 3]HR,SBP,MAP,LAC,Te,RPE,and TS increased significantly after cycling(P<0.05).During the follow-up period after cold water immersion,the HR of the CON group remained significantly higher than the baseline value at 55 min after exercise(P<0.05),and the GCWI,CCWI-EE,CCWI-VJP,and CCWIRMSSD groups had no significant difference from the baseline value at 55 min after exercise(P>0.05).LAC in the CON and CCWI-EE groups was not significantly different from the baseline after 45 minutes(P>0.05);there was no significant difference from the baseline value at 55 minutes after exercise in the CCWI-VJP group(P>0.05),GCWI and CCWI-RMSSD remained significantly higher than baseline values at 55 min after exercise(P<0.05).In the CON group,TS was not significantly different from the baseline after 25 min of exercise(P>0.05),and GCWI had no significant difference from the baseline after 35 min(P>0.05);there was no significant difference from the baseline value at 55 minutes after exercise in the CCWI-VJP group(P>0.05),and both CCWI-EE and CCWI-RMSSD groups remained significantly lower than baseline values at 55 min after exercise(P<0.05).In the CCWIRMSSD group,RMSSD,SDNN,SDSD,TP,HF,and LF were significantly higher in the posttest than in the pre-test(P<0.05).The duration of incremental load exercise in the CON group was significantly lower after the test than before the test(P<0.05),and the CCWI-EE group was significantly higher in the post-test than in the pre-test(P<0.05).The post-test vertical jump height of the GCWI,CCWI-EE,and CCWI-RMSSD group was significantly lower than that of the pre-test(P<0.05),but there was no significant difference between CON and CCWI-VJP groups(P>0.05).The untargeted metabolomics results showed that 5083 metabolites with HMDB annotation were detected.There were 26 differential metabolites and 17 differential metabolic pathways between CON-Post and CON-Pre groups.GCWI-Post and CON-Post groups had 21 differential metabolites and 11 differential metabolic pathways.There were 31 differential metabolites and 5 differential metabolic pathways between the CCWI-EE-Post and CON-Post groups.There were 31 differential metabolites and 23 differential metabolic pathways between the CCWI-VJP-Post and CON-Post groups.There were 4 differential metabolites and 2 differential metabolic pathways between the CCWI-RMSSD-Post and CONPost groups.There were 24 differential metabolites and 7 differential metabolic pathways between CCWI-EE-Post and GCWI-Post groups.There were 14 differential metabolites and 6 differential metabolic pathways between CCWI-VJP-Post and GCWI-Post groups.There were 13 differential metabolites and 5 differential metabolic pathways between CCWI-RMSSD-Post and GCWI-Post groups.[Study 4]The results of the iterative and optimized for the duration of incremental load cycling showed that the characteristic variables with significant influence included the pre-measured duration of incremental load cycling,age,body water content,muscle mass,and environmental temperature,and the coefficient of determination of the test sample was R2=0.891.The iterative and optimized vertical jump height prediction model results showed that the most influential characteristic variables were the pre-measured vertical jump height,muscle mass,mean arterial pressure,body age,and environmental temperature,R2=0.831.The iterative and optimized RMSSD prediction model results showed that the most influential characteristic variables were pre-measured RMSSD,basal metabolic rate,visceral fat grade,age,and BMI,with R2=0.353.Conclusion:(1)Cold water immersion after exercise can effectively reduce HR,Tms,RPE,and TS,promote parasympathetic activation and improve endurance performance,but it also reduces reaction time and vertical jump performance.However,the optimal duration of cold water immersion varies for different recovery purposes.Within a specific range,increased immersion duration can further improve endurance exercise performance and promote parasympathetic activation,but it also causes a downward trend in vertical jump performance.In addition,body fat percentage also affects the recovery effect of cold water immersion after exercise.(2)The three prediction models established by the machine learning XGBoost algorithm based on the recovery of incremental load cycling time,the recovery of vertical jump ability,and parasympathetic activation are credible.The effect of cold water immersion after exercise will be affected by the duration of immersion and individual characteristics.Twenty-four characteristic variables were input into the three models,among which the duration of cold water immersion was input into 5 to 20 minutes.Customized cold water immersion durations could be obtained by comparing the predictive values of the output of different cold water immersion durations after exercise for different recovery purposes.(3)Compared with the GCWI,the three CCWI models based on XGBoost can promote recovery after exercise in a hot environment.The CCWI-EE model can promote the recovery of HR,BLA,TS,and incremental load cycling time after exercise in a hot environment.CCWIVJP model can promote the recovery of BLA and VJP after exercise in a hot environment.CCWI-RMSSD model can further promote the recovery of TS and parasympathetic activation after exercise in a hot environment.(4)The three prediction models after iteration and optimization for different recovery purposes were credible,among which the prediction accuracy of the iterative and optimized CCWI-EE and CCWI-RMSSD prediction models were improved.Moreover,the importance of environmental temperature as a characteristic variable in the three prediction models increased.