Aki Pulkkinen

On- and Off-dynamics and Respiration Rate Enhance the Accuracy of Heart Rate Based VO2 Estimation

Current heart rate (HR) based calculations use linear steady state based function to calculate VO2 from HR. HR to VO2 relationship changes during variations in exercise intensity and non-exercise related physiological processes affecting HR especially at low intensity. PURPOSE: To examine whether addition of Onand OffVO2-dynamics (VO2HR+ON/OFF) and respiration rate (VO2HR+Resp) or both (VO2HR+Resp+ON/OFF) increase the accuracy of HR based VO2 estimation (VO2HR). METHODS: Beat-by-beat HR and breath-by-breath VO2 data from 16 male and 16 female healthy untrained adults (age of 38±9 years, weight 70±11 kg, height 172±8 cm and VO2max 44±9 ml/kg•min-1) were collected 5 minutes prior to and 15 minutes after 10min exercises at 40% and 70% VO2max and maximal stepwise test on bicycle ergometer. Two 50 min series of simulated low intensity real life tasks (RLT, mean HR 101±8 and 105±9 bpm) were also carried out. Onand OffVO2 changes were used to model VO2 dynamics. Four neural network models were constructed using 3% from the data in the learning process: VO2HR, VO2HR+Resp, VO2HR+ON/OFF, VO2HR+Resp+ON/OFF. The whole dataset was used to evaluate the accuracy of the methods using mean absolute error (MAE) between estimated and measured VO2. RESULTS: Accuracy was enhanced (p<.001) in all exercise conditions and during RLTs when ON/OFF-response information was included. Across all subjects and conditions MAE of VO2HR 3.7 ml/kg•min-1 was reduced to 3.3, 2.3 and 1.9 ml/kg*min-1 improving the accuracy by 11%, 38% and 48% using VO2HR+Resp, VO2HR+ON/OFF, VO2HR+Resp+ON/OFF methods, respectively. During maximal exercise test MAE of VO2HR 4.9 ml/kg*min-1 was reduced to 4.4, 1.8 and 1.3 ml/kg•min-1 improving the accuracy by 6.1%, 60.9% and 70.7% using VO2HR+Resp, VO2HR+ON/OFF, VO2HR+Resp+ON/OFF methods, respectively. CONCLUSIONS: Both Onand Offdynamics and respiration were able to enhance the accuracy of VO2 estimation compared with VO2HR from maximal to varying low intensity exercise. The greatest and the most consistent improvements were due to Onand Off-dynamics.

Energy Expenditure Can Be Accurately Estimated From HR Without Individual Laboratory Calibration: 586 Board #177 3:30 PM ??? 5:00 PM

Current heart rate (HR) based energy expenditure (EE) estimation methods are inaccurate. Flex-HR method is currently most accurate, but it requires individual calibration in laboratory limiting applicability for large-scale daily use under free-living conditions. A recent method utilizing RR-interval (RRI) derived data on HR, respiratory frequency and On-Off dynamics has increased the accuracy of HR-based VO2-estimation (RRIEST, Pulkkinen et. al., MSSE 36(5), 2004). PURPOSE: To evaluate whether EE during real life tasks and physical exercises can be estimated accurately using RRIEST without individual laboratory calibration. METHODS: RRI and breath-by-breath VO2 data from 16 male and 16 female healthy untrained adults (age of 38±9 years, weight 70±11 kg, height 172±8 cm and VO2max 44±9 ml/kg•min-1) were collected 5 minutes prior to, during and 15 minutes after 10-min exercises at 40% and 70% VO2max and maximal stepwise test (MAX) on bicycle ergometer. Two 50 min series of simulated low intensity real life tasks (RLT1 & RLT2, mean HR 101±8 and 105±9 bpm) were also carried out. Steady-state periods from MAX were used to construct individual (FLEXIND) and mean for all subjects (FLEXALL) equation slopes and intercepts to calculate EE. Flex HR under which EE was assumed to be at resting level was determined as the mean of the highest 1 min HR during baseline and lowest HR during MAX. RRI data was used to calculate EE with RRIEST model (Pulkkinen et. al., 2004). Accuracy was evaluated using mean absolute error (MAE) and r2 between the estimated and the measured. RESULTS: Across all subjects and conditions, MAE and r2 between measured and estimated EE for RRIEST, FLEXIND and FLEXALL were 139, 185 and 191 kcal, and 0.81, 0.77 and 0.77, respectively. RRIEST was significantly (p<0.05) more accurate during RLT1 and MAX compared with both FLEX-methods, whereas during 40% exercise FLEXIND was more accurate than RRIEST or FLEXALL. CONCLUSIONS: RRIEST provided accurate EE estimation during simulated real life tasks and physical exercises. EE can be estimated accurately using RRIEST model without individual laboratory calibration making the RRIEST method especially suitable for field use.