Second ventilatory threshold from heart-rate variability: valid when the upper body is involved?

Deep learning-based dynamic ventilatory threshold estimation from electrocardiograms

BACKGROUND AND OBJECTIVE

The ventilatory threshold (VT) marks the transition from aerobic to anaerobic metabolism and is used to assess cardiorespiratory endurance. A conventional way to assess VT is cardiopulmonary exercise testing, which requires a gas analyzer. Another method for measuring VT involves calculating the heart rate variability (HRV) from an electrocardiogram (ECG) by computing the variability of heartbeats. However, the HRV method has some limitations. ECGs should be recorded for at least 5 minutes to calculate the HRV, and the result may depend on the utilized ECG preprocessing algorithms.

METHODS

To overcome these problems, we developed a deep learning-based model consisting of long short-term memory (LSTM) and convolutional neural network (CNN) for a lead II ECG. Variables reflecting subjects' physical characteristics, as well as ECG signals, were input into the model to estimate VT. We applied joint optimization to the CNN layers to generate an informative latent space, which was fed to the LSTM layers. The model was trained and evaluated on two datasets, one from the Bruce protocol and the other from a protocol including multiple tasks (MT).

RESULTS

Acceptable performances (mean and 95% CI) were obtained on the datasets from the Bruce protocol (-0.28[-1.91,1.34] ml/min/kg) and the MT protocol (0.07[-3.14,3.28] ml/min/kg) regarding the differences between the predictions and labels. The coefficient of determination, Pearson correlation coefficient, and root mean square error were 0.84, 0.93, and 0.868 for the Bruce protocol and 0.73, 0.97, and 3.373 for the MT protocol, respectively.

CONCLUSIONS

The results indicated that it is possible for the proposed model to simultaneously assess VT with the inputs of successive ECGs. In addition, from ablation studies concerning the physical variables and the joint optimization process, it was demonstrated that their use could boost the VT assessment performance of the model. The proposed model enables dynamic VT estimation with ECGs, which could help with managing cardiorespiratory fitness in daily life and cardiovascular rehabilitation in patients.

Heart Rate Variability-Derived Thresholds for Exercise Intensity Prescription in Endurance Sports: A Systematic Review of Interrelations and Agreement with Different Ventilatory and Blood Lactate Thresholds

BACKGROUND

Exercise intensities are prescribed using specific intensity zones (moderate, heavy, and severe) determined by a 'lower' and a 'higher' threshold. Typically, ventilatory (VT) or blood lactate thresholds (LT), and critical power/speed concepts (CP/CS) are used. Various heart rate variability-derived thresholds (HRVTs) using different HRV indices may constitute applicable alternatives, but a systematic review of the proximity of HRVTs to established threshold concepts is lacking.

OBJECTIVE

This systematic review aims to provide an overview of studies that determined HRVTs during endurance exercise in healthy adults in comparison with a reference VT and/or LT concept.

METHODS

A systematic literature search for studies determining HRVTs in healthy individuals during endurance exercise and comparing them with VTs or LTs was conducted in Scopus, PubMed and Web of Science (until January 2022). Studies claiming to describe similar physiological boundaries to delineate moderate from heavy (HRVTlow vs. VTlow and/or LTlow), and heavy from severe intensity zone (HRVThigh vs. VThigh and/or LThigh) were grouped and their results synthesized.

RESULTS

Twenty-seven included studies (461 participants) showed a mean difference in relative HR between HRVTlow and VTlow of - 0.6%bpm in weighted means and 0.02%bpm between HRVTlow and LTlow. Bias between HR at HRVTlow and VTlow was 1 bpm (limits of agreement (LoA): - 10.9 to 12.8 bpm) and 2.7 bpm (LoA: - 20.4 to 25.8 bpm) between HRVTlow and LTlow. Mean difference in HR between HRVThigh and VThigh was 0.3%bpm in weighted means and 2.9%bpm between HRVThigh and LThigh while bias between HR at HRVThigh and VThigh was - 4 bpm (LoA: - 17.9 to 9.9 bpm) and 2.5 bpm (LoA: - 12.1 to 17.1 bpm) between HRVThigh and LThigh.

CONCLUSION

HRVTlow seems to be a promising approach for the determination of a 'lower' threshold comparable to VTlow and potentially for HRVThigh compared to VThigh, although the latter needs further empirical evaluation. LoA for both intensity zone boundaries indicates bias of HRVTs on an individual level. Taken together, HRVTs can be a promising alternative for prescribing exercise intensity in healthy, male athletes undertaking endurance activities but due to the heterogeneity of study design, threshold concepts, standardization, and lack of female participants, further research is necessary to draw more robust and nuanced conclusions.

Is There Agreement and Precision between Heart Rate Variability, Ventilatory, and Lactate Thresholds in Healthy Adults?

This study aims to analyze the agreement and precision between heart rate variability thresholds (HRVT1/2) with ventilatory and lactate thresholds 1 and 2 (VT1/2 and LT1/2) on a treadmill. Thirty-four male students were recruited. Day 1 consisted of conducting a health survey, anthropometrics, and Cardiopulmonary Exercise Test (CPx). On Day 2, after 48 h, a second incremental test was performed, the Cardiopulmonary Stepwise Exercise Test consisting of 3 min stages (CPxS), to determine VT1/2, LT1/2, and HRVT1/2. One-way repeated-measures ANOVA and effect size (η_p²) were used, followed by Sidak's post hoc. The Coefficient of Variation (CV) and Typical Error (TE) were applied to verify the precision. Bland Altman and the Intraclass Correlation Coefficient (ICC) were applied to confirm the agreement. HRVT1 showed different values compared to LT1 (lactate, RER, and R-R interval) and VT1 (V̇E, RER, V̇CO₂, and HR). No differences were found in threshold 2 (T2) between LT2, VT2, and HRVT2. No difference was found in speed and V̇O₂ for T1 and T2. The precision was low to T1 (CV > 12% and TE > 10%) and good to T2 (CV < 12% and TE < 10%). The agreement was good to fair in threshold 1 (VT1, LT1, HRVT1) and excellent to good in T2 (VT1, LT1, HRVT1). HRVT1 is not a valid method (low precision) when using this protocol to estimate LT1 and VT1. However, HRVT2 is a valid and noninvasive method that can estimate LT2 and VT2, showing good agreement and precision in healthy adults.

Determination of the Respiratory Compensation Point by Detecting Changes in Intercostal Muscles Oxygenation by Using Near-Infrared Spectroscopy

This study aimed to evaluate if the changes in oxygen saturation levels at intercostal muscles (SmO2-m.intercostales) assessed by near-infrared spectroscopy (NIRS) using a wearable device could determine the respiratory compensation point (RCP) during exercise. Fifteen healthy competitive triathletes (eight males; 29 ± 6 years; height 167.6 ± 25.6 cm; weight 69.2 ± 9.4 kg; V˙O2-máx 58.4 ± 8.1 mL·kg−1·min−1) were evaluated in a cycle ergometer during the maximal oxygen-uptake test (V˙O2-máx), while lung ventilation (V˙E), power output (watts, W) and SmO2-m.intercostales were measured. RCP was determined by visual method (RCPvisual: changes at ventilatory equivalents (V˙E·V˙CO2−1, V˙E·V˙O2−1) and end-tidal respiratory pressure (PetO2, PetCO2) and NIRS method (RCPNIRS: breakpoint of fall in SmO2-m.intercostales). During exercise, SmO2-m.intercostales decreased continuously showing a higher decrease when V˙E increased abruptly. A good agreement between methods used to determine RCP was found (visual vs NIRS) at %V˙O2-máx, V˙O2, V˙E, and W (Bland-Altman test). Correlations were found to each parameters analyzed (r = 0.854; r = 0.865; r = 0.981; and r = 0,968; respectively. p < 0.001 in all variables, Pearson test), with no differences (p < 0.001 in all variables, Student’s t-test) between methods used (RCPvisual and RCPNIRS). We concluded that changes at SmO2-m.intercostales measured by NIRS could adequately determine RCP in triathletes.

Ski Mountaineering: Perspectives on a Novel Sport to Be Introduced at the 2026 Winter Olympic Games

Ski mountaineering is a rapidly growing winter sport that involves alternately climbing and descending slopes and various racing formats that differ in length and total vertical gain, as well as their distribution of downhill and uphill sections. In recent years, both participation in and media coverage of this sport have increased dramatically, contributing, at least in part, to its inclusion in the 2026 Winter Olympics in Milano-Cortina. Here, our aim has been to briefly describe the major characteristics of ski mountaineering, its physiological and biomechanical demands, equipment, and training/testing, as well as to provide some future perspectives. Despite its popularity, research on this discipline is scarce, but some general characteristics are already emerging. Pronounced aerobic capacity is an important requirement for success, as demonstrated by positive correlations between racing time and maximal oxygen uptake and oxygen uptake at the second ventilatory threshold. Moreover, due to the considerable mechanical work against gravity on demanding uphill terrain, the combined weight of the athlete and equipment is inversely correlated with performance, prompting the development of both lighter and better equipment in recent decades. In ski mountaineering, velocity uphill is achieved primarily by more frequent (rather than longer) strides due primarily to high resistive forces. The use of wearable technologies, designed specifically for analysis in the field (including at elevated altitudes and cold temperatures) and more extensive collaboration between researchers, industrial actors, and coaches/athletes, could further improve the development of this sport.

Assessment of Maximal Aerobic Capacity in Ski Mountaineering: A Laboratory-Based Study

This study aims to evaluate the agreement in maximum oxygen consumption (V˙O₂max) between a running protocol and a ski mountaineering (SKIMO) protocol. Eighteen (eleven males, seven females) ski mountaineers (age: 25 ± 3 years) participated in the study. V˙O₂max, maximum heart rate (HRmax), and maximum blood lactate concentration (BLAmax) were determined in an incremental uphill running test and an incremental SKIMO-equipment-specific test. V˙O₂max did not differ between the SKIMO and uphill running protocols (p = 0.927; mean difference -0.07 ± 3.3 mL/min/kg), nor did HRmax (p = 0.587, mean difference -0.7 ± 5.1 bpm). A significant correlation was found between V˙O₂max SKIMO and V˙O₂max running (p ≤ 0.001; ICC = 0.862 (95% CI: 0.670-0.946)). The coefficient of variation was 4.4% (95% CI: 3.3-6.5). BLAmax was significantly lower for SKIMO compared to running (12.0 ± 14.1%; p = 0.002). This study demonstrates that V˙O₂max determined with a traditional uphill running protocol demonstrates good agreement with an equipment-specific SKIMO protocol.

Fatigue Evaluation through Machine Learning and a Global Fatigue Descriptor

Research in physiology and sports science has shown that fatigue, a complex psychophysiological phenomenon, has a relevant impact in performance and in the correct functioning of our motricity system, potentially being a cause of damage to the human organism. Fatigue can be seen as a subjective or objective phenomenon. Subjective fatigue corresponds to a mental and cognitive event, while fatigue referred as objective is a physical phenomenon. Despite the fact that subjective fatigue is often undervalued, only a physically and mentally healthy athlete is able to achieve top performance in a discipline. Therefore, we argue that physical training programs should address the preventive assessment of both subjective and objective fatigue mechanisms in order to minimize the risk of injuries. In this context, our paper presents a machine-learning system capable of extracting individual fatigue descriptors (IFDs) from electromyographic (EMG) and heart rate variability (HRV) measurements. Our novel approach, using two types of biosignals so that a global (mental and physical) fatigue assessment is taken into account, reflects the onset of fatigue by implementing a combination of a dimensionless (0-1) global fatigue descriptor (GFD) and a support vector machine (SVM) classifier. The system, based on 9 main combined features, achieves fatigue regime classification performances of 0.82 ± 0.24, ensuring a successful preventive assessment when dangerous fatigue levels are reached. Training data were acquired in a constant work rate test (executed by 14 subjects using a cycloergometry device), where the variable under study (fatigue) gradually increased until the volunteer reached an objective exhaustion state.

Applicability of Dmax Method on Heart Rate Variability to Estimate the Lactate Thresholds in Male Runners

Introduction

The purpose of this study was to evaluate the application of the Dmax method on heart rate variability (HRV) to estimate the lactate thresholds (LT), during a maximal incremental running test (MIRT).

Methods

Nineteen male runners performed two MIRTs, with the initial speed at 8 km·h^-1 and increments of 1 km·h^-1 every 3 minutes, until exhaustion. Measures of HRV and blood lactate concentrations were obtained, and lactate (LT₁ and LT₂) and HRV (HRVT_DMAX1 and HRVT_DMAX2) thresholds were identified. ANOVA with Scheffe's post hoc test, effect sizes (d), the bias ± 95% limits of agreement (LoA), standard error of the estimate (SEE), Pearson's (r), and intraclass correlation coefficient (ICC) were calculated to assess validity.

Results

No significant differences were observed between HRVT_DMAX1 and LT₁ when expressed for speed (12.1 ± 1.4 km·h^-1 and 11.2 ± 2.1 km·h^-1; p=0.55; d = 0.45; r = 0.46; bias ± LoA = 0.8 ± 3.7 km·h^-1; SEE = 1.2 km·h^-1 (95% CI, 0.9-1.9)). Significant differences were observed between HRVT_DMAX2 and LT₂ when expressed for speed (12.0 ± 1.2 km·h^-1 and 14.1 ± 2.5 km·h^-1; p=0.00; d = 1.21; r = 0.48; bias ± LoA = -1.0 ± 1.8 km·h^-1; SEE = 1.1 km·h^-1 (95% CI, 0.8-1.6)), respectively. Reproducibility values were found for the LT₁ (ICC = 0.90; bias ± LoA = -0.7 ± 2.0 km·h^-1), LT₂ (ICC = 0.97; bias ± LoA = -0.1 ± 1.1 km·h^-1), HRVT_DMAX1 (ICC = 0.48; bias ± LoA = -0.2 ± 3.4 km·h^-1), and HRVT_DMAX2 (ICC = 0.30; bias ± LoA = 0.3 ± 3.5 km·h^-1).

Conclusions

The Dmax method applied over a HRV dataset allowed the identification of LT₁ that is close to aerobic threshold, during a MIRT.

Heart rate variability to assess ventilatory thresholds in professional basketball players

PURPOSE

The aim of this study was to determine if heart rate variability (HRV) during incremental test could be used to estimate ventilatory threshold (VT) in professional basketball players, with sufficient precision to be used in their training. Furthermore, the second aim was to analyse the association between HRV and 3 methods of VT determination by gas analysis.

METHODS

Twenty-four professional basketball players (age: 23.4 ± 4.9 years; height: 195.4 ± 9.8 cm; body mass: 92.2 ± 11.9 kg) performed an incremental running test to exhaustion. First ventilatory threshold (VT1) was determined by ventilatory equivalent (VE) and HRV and second ventilatory threshold (VT2) was determined by 3 methods of gases analysis (V-slope, VE and gas exchange ratio (R), and HRV). Pearson's coefficient (r) was used to detect differences between data and the strength of each relationship. The mean of absolute differences and Bland-Altman analysis were used to evaluate whether there was agreement.

RESULTS

The results showed no significant differences in HR and oxygen consumption (VO₂) at VT1 between the 2 methods. Furthermore, no significant differences among the methods of gases analysis and HRV were observed in speed, HR, and VO₂ at VT2. Moreover, VTs estimated using HRV and gas methods were significantly correlated. Correlation in HR values was higher between R and HRV (r = 0.96) and VE and HRV (r = 0.96) than V-slope and HRV (r = 0.90).

CONCLUSION

These findings provide a practical, inexpensive approach for evaluating specific training loads when determining VT2 in basketball players. Therefore, HRV is an alternative method to determine VT2 without the application of expensive technology that limits its use to laboratories.

An investigation into the relationship between heart rate variability and the ventilatory threshold in healthy moderately trained males

BACKGROUND

During incremental exercise, heart rate variability (HRV) has been shown to display distinct stabilization and inflection points, which have been used to indirectly detect the ventilatory threshold (VT).

METHODS

Ten moderately trained males (26·5 ± 5·9 years: VO_2peak 48·7 ± 4·1 ml min^-1  kg^-1 ) performed an incremental test on a cycle ergometer until volitional exhaustion with both R-R intervals and respiratory indices recorded. HRV was quantified using both nonlinear (Poincare plot; short-term variability SD1) and spectral analysis of the R-R intervals (high-frequency component; HFp). The VT was identified using the V-slope method. The relationship between HRV parameters and the VT was assessed using both a paired t-test and Pearson's product correlation. In addition, Bland and Altman plots were used to quantify the mean difference along with a 95% confidence interval.

RESULTS

When expressed as the corresponding heart rate values, both the SD1 and the HFp stabilization points revealed a strong (r = 0·86 and 0·087, respectively) correlation with the VT. However, only for SD1 this relationship was different to the VT (t-test). The Bland-Altman plots supported these findings showing wide limits of agreement present for SD1 and the VT whilst the relationship between HFp and the VT revealed narrower limits.

CONCLUSION

There does not appear to be a relationship present between the VT and the SD1 stabilization point in moderately trained healthy males, whereas the HFp stabilization point revealed a strong relationship with the VT when expressed as heart rate.

Assessment of Heart Rate Variability Thresholds from Incremental Treadmill Tests in Five Cross-Country Skiing Techniques

The assessment of heart rate variability (HRV) thresholds (HRVTs) as an alternative of Ventilatory thresholds (VTs) is a relatively new approach with increasing popularity which has not been conducted in cross-country (XC) skiing yet. The main purpose of the present study was to assess HRVTs in the five main XC skiing-related techniques, double poling (DP), diagonal striding (DS), Nordic walking (NW), V1 skating (V1), and V2 skating (V2).Ten competitive skiers completed these incremental treadmill tests until exhaustion with a minimum of one to two recovery days in between each test. Ventilatory gases, HRV and poling frequencies were measured. The first HRV threshold (HRVT₁) was assessed using two time-domain analysis methods, and the second HRV threshold (HRVT₂) was assessed using two non-time varying frequency-domain analysis methods. HRVT₁ was assessed by plotting the mean successive difference (MSD) and standard deviation (SD) of normalized R-R intervals to workload. HRVT₁ was assessed by plotting high frequency power (HFP) and the HFP relative to respiratory sinus arrhythmia (HFP_RSA) with workload. HRVTs were named after their methods (HRVT_1-SD; HRVT_1-MSD; HRVT_2-HFP; HRVT_2-HFP-RSA). The results showed that the only cases where the proposed HRVTs were good assessors of VTs were the HRVT_1-SD of the DS test, the HRVT_1-MSD of the DS and V2 tests, and the HRVT_2-HFP-RSA of the NW test. The lack of a wider success of the assessment of HRVTs was reasoned to be mostly due to the high entrainment between the breathing and poling frequencies. As secondary finding, a novel Cardiolocomotor coupling mode was observed in the NW test. This new Cardiolocoomtor coupling mode corresponded to the whole bilateral poling cycle instead of corresponding to each poling action as it was reported to the date by the existing literature.