Neural respiratory drive (NRD) reflects the output of central respiratory centers in the brainstem to respiratory muscles and is modulated by chemoreceptor input, mechanoreceptor feedback, and higher cortical influences [16][17]. NRD increases in response to hypoxia, hypercapnia, and increased mechanical load, serving as a compensatory mechanism to maintain ventilation.

Because direct measurement of central respiratory output is not feasible in clinical practice, surrogate measures such as diaphragm EMG are used [12]. sEMG provides an indirect but clinically accessible measure of NRD by quantifying electrical activity in respiratory muscles [1].

Studies have demonstrated strong correlations between sEMG signals from parasternal intercostals and invasive measures of diaphragm activity, supporting their use as non-invasive surrogates [11][14]. However, interpretation requires consideration of factors such as electrode placement, tissue impedance, and muscle geometry [9].

The balance between respiratory load and muscle capacity is a central determinant of respiratory function [13]. Load includes airway resistance, lung compliance, and metabolic demand, while capacity reflects muscle strength and endurance.

When load exceeds capacity, NRD increases to compensate. This leads to recruitment of accessory muscles and increased work of breathing. Persistent imbalance results in muscle fatigue and eventual ventilatory failure [18].

sEMG provides a direct measure of this imbalance by quantifying both the intensity and distribution of muscle activation. Elevated accessory muscle activity reflects increased load and reduced diaphragm efficiency, making it a potential biomarker of disease severity [3].

Accessory respiratory muscles—including the sternocleidomastoid, scalene, and intercostals—are minimally active during quiet breathing but become progressively engaged as respiratory demand increases [13]. Recruitment follows a hierarchical pattern in which diaphragm activation initially increases, followed by contribution from the intercostal muscles, and ultimately recruitment of accessory neck muscles as demand continues to rise. This sequence reflects an adaptive strategy that prioritizes mechanical efficiency and fatigue resistance. However, increased reliance on accessory muscles is energetically inefficient and is associated with greater oxygen consumption [19].

sEMG enables detailed characterization of these recruitment patterns, providing insight into compensatory mechanisms and facilitating early detection of respiratory distress.

sEMG involves placing electrodes on the skin surface to detect electrical activity generated by muscle fibers. Accurate electrode placement is critical to ensure signal fidelity and minimize contamination [20]. However, signal acquisition is subject to several important limitations, including crosstalk from adjacent muscles, cardiac interference (ECG contamination), attenuation by subcutaneous adipose tissue, and motion artifacts during breathing, all of which can affect signal amplitude and timing [2][9]. High-density sEMG arrays have been proposed to improve spatial resolution and reduce crosstalk, thereby enhancing signal specificity [21].

Raw sEMG signals require processing to extract meaningful information. This typically involves band-pass filtering to remove noise, rectification to convert the signal into a unidirectional waveform, and smoothing to generate a linear envelope that reflects the temporal pattern of muscle activation. More advanced methods, such as wavelet analysis and independent component analysis, can further improve signal quality and enable separation of overlapping physiological signals [2][10].

Normalization is essential for comparison across subjects; however, commonly used approaches such as maximal voluntary contraction (MVC) are often not feasible in critically ill patients, limiting standardization in this population [1].

Machine learning and pattern recognition techniques are increasingly applied to sEMG data. These approaches enable identification of complex patterns associated with fatigue, coordination, and disease progression [10][22]. Such methods hold promise for improving predictive accuracy but require large datasets and standardization.

Several quantitative features are commonly derived from sEMG signals, each with distinct physiological meaning. The root mean square (RMS) reflects the amplitude of muscle activation and is proportional to motor unit recruitment and firing frequency; clinically, higher RMS indicates increased respiratory effort and neural drive. The integrated EMG (iEMG) represents total electrical activity over time and provides an estimate of the cumulative work of breathing during a respiratory cycle. In contrast, median frequency, a frequency-domain parameter, decreases with muscle fatigue as a result of slowing muscle fiber conduction velocity. Signal efficiency describes the relationship between electrical activity and mechanical output, with reduced efficiency suggesting neuromechanical uncoupling or developing fatigue.

At present, there are no universally accepted thresholds or reference ranges for these parameters in respiratory applications, and most studies rely on within-subject trends or relative changes rather than absolute cutoffs, limiting standardization in clinical practice.

The choice of recording site significantly influences both signal quality and clinical interpretation. The parasternal intercostals provide a reliable surrogate of diaphragm activity, with relatively low crosstalk and a strong correlation with neural respiratory drive, making them the most well-supported site for monitoring NRD. Surface recordings over the diaphragm region offer direct physiological relevance but are technically challenging due to muscle depth and interference from cardiac signals. The sternocleidomastoid is easily accessible and generates strong, high-amplitude signals, which makes it useful for detecting increased respiratory effort, although it is more susceptible to activation from non-respiratory factors. The scalene muscles reflect early accessory muscle recruitment and are sensitive indicators of increased respiratory load, but they are also prone to crosstalk and variability in electrode placement.

Overall, parasternal recordings are preferred for estimating neural respiratory drive, whereas sternocleidomastoid and scalene activity are most informative for identifying accessory muscle recruitment and clinical deterioration.

sEMG allows precise measurement of activation amplitude in accessory muscles such as the sternocleidomastoid and scalene muscles. Increased amplitude reflects greater motor unit recruitment and firing frequency, which are indicative of elevated neural respiratory drive (NRD) [9][12]. Importantly, these changes can occur even when traditional clinical parameters such as respiratory rate remain within normal ranges, highlighting the sensitivity of sEMG as an early biomarker.

Quantitative indices such as RMS amplitude and integrated EMG (iEMG) have been used to characterize muscle activity. Studies have demonstrated that RMS values of accessory muscles increase proportionally with inspiratory load, suggesting a direct relationship between sEMG amplitude and respiratory effort [10][19].

Beyond amplitude, the timing of muscle activation provides critical insight into respiratory mechanics. In healthy individuals, accessory muscles are minimally active during quiet breathing and are recruited later in the inspiratory cycle when demand increases. However, in pathological states, early activation of these muscles is often observed [14].

This early recruitment reflects a shift in the load-sharing strategy of the respiratory system and may indicate reduced diaphragm efficiency. sEMG studies have shown that patients with COPD or acute respiratory failure exhibit altered timing patterns, including prolonged activation and reduced synchrony between muscle groups [3][15].

Asynchrony between respiratory muscles can also reflect neuromechanical uncoupling, a condition in which neural drive is not effectively translated into mechanical ventilation. This phenomenon is particularly relevant in mechanically ventilated patients and is associated with increased work of breathing and poor outcomes [8].

sEMG-derived measures of accessory muscle activity consistently map to specific clinical outcomes rather than serving as nonspecific markers. Increased sternocleidomastoid and scalene activation correlates with dyspnea severity and reduced exercise tolerance in COPD, reflecting heightened neural respiratory drive (NRD) and load-capacity imbalance [3][14]. In acute and critical care settings, persistently elevated accessory muscle activity—particularly when sustained despite support—is associated with non-invasive ventilation (NIV) failure and progression toward intubation [15].

In mechanically ventilated patients, elevated baseline NRD and early accessory muscle recruitment during spontaneous breathing trials predict weaning failure, indicating insufficient respiratory reserve [1][15]. Across ICU and chronic disease populations, sustained high sEMG amplitude—reflecting unresolved physiological stress—has been associated with increased mortality, particularly when combined with evidence of fatigue or neuromechanical uncoupling [1][15].

Interpretation of elevated accessory muscle sEMG requires consideration of non-respiratory factors that can independently increase signal amplitude. Postural changes, particularly upright positioning or forward leaning, can increase baseline muscle activation. Speech and voluntary movements also transiently elevate activity in cervical and thoracic muscles. Psychological factors such as anxiety can increase respiratory drive independent of physiological load, while pain may alter breathing patterns and muscle recruitment. Obesity and increased subcutaneous tissue can both attenuate and distort signals, complicating amplitude interpretation. Additionally, variability in electrode placement and inter-electrode distance can significantly affect recorded activity. These factors should be accounted for to avoid overestimating respiratory distress based solely on sEMG amplitude.

The pathophysiology of acute respiratory failure involves a combination of increased airway resistance, reduced lung compliance, and impaired gas exchange. These factors increase the work of breathing and necessitate compensatory increases in NRD [6].

As NRD increases, the diaphragm initially augments its activity. However, when the diaphragm reaches its functional limits, accessory muscles are recruited to sustain ventilation. This compensatory mechanism is reflected in increased sEMG amplitude across multiple muscle groups [13].

sEMG studies have identified characteristic patterns associated with acute respiratory failure, including increased amplitude of diaphragm and accessory muscle signals, prolonged activation duration, and reduced coordination between muscle groups. Together, these changes reflect both increased respiratory load and reduced efficiency of the respiratory system [2].

Importantly, these sEMG signatures often precede changes in traditional clinical parameters such as oxygen saturation or arterial blood gases. This temporal advantage suggests that sEMG could serve as an early warning tool for impending respiratory failure [1].

Elevated NRD and accessory muscle activity have been associated with increased risk of intubation and mortality. Patients who fail non-invasive ventilation (NIV) often exhibit persistently high sEMG activity despite support, indicating ongoing respiratory distress [15].

From a mechanistic perspective, sustained high NRD reflects a failing compensatory response. As respiratory muscles fatigue, the system becomes unable to maintain ventilation, leading to rapid clinical deterioration [23].

Patient-ventilator asynchrony occurs when there is a mismatch between patient effort and ventilator support. This mismatch can lead to increased work of breathing, discomfort, and prolonged ventilation [24]. sEMG provides a direct measure of patient effort, allowing detection of asynchrony that may not be apparent from ventilator waveforms alone. For example, delayed or ineffective triggering can be identified by comparing sEMG activity with ventilator cycles [8].

Mechanical ventilation can lead to both underloading and overloading of respiratory muscles. Underloading may result in muscle atrophy, while overloading can lead to fatigue [5].

sEMG enables real-time monitoring of muscle activity, allowing clinicians to titrate ventilator support to maintain an optimal level of muscle engagement. This approach aligns with the concept of “lung and diaphragm-protective ventilation” [1].

Weaning failure is a complex process involving multiple factors, including respiratory muscle weakness, excessive load, and impaired coordination [4]. sEMG studies indicate that failure is associated with elevated baseline neural respiratory drive, early recruitment of accessory muscles, and persistently high signal amplitude during spontaneous breathing trials, all of which reflect insufficient respiratory capacity to sustain independent ventilation [1][15]. These findings suggest that sEMG could be incorporated into weaning protocols to improve prediction accuracy and guide clinical decision-making.

COPD is characterized by airflow limitation and hyperinflation, which impair diaphragm function and increase reliance on accessory muscles [18].

sEMG studies have demonstrated that patients with COPD exhibit elevated NRD even at rest, reflecting chronic respiratory stress [14]. This persistent elevation is associated with increased dyspnea, reduced exercise capacity, and higher mortality [3].

In neuromuscular diseases such as ALS, progressive weakness of respiratory muscles leads to ventilatory failure [25], sEMG can detect early changes in muscle activation patterns, including reduced amplitude and altered timing. These changes may precede clinical symptoms, suggesting a role for sEMG in early diagnosis and monitoring.

The ability of sEMG to provide continuous, non-invasive monitoring makes it particularly valuable in chronic disease. Longitudinal changes in NRD and muscle recruitment patterns may provide insight into disease progression and prognosis.

Elevated NRD reflects severe physiological stress and an inability of the respiratory system to meet metabolic demands. Persistent high NRD is associated with muscle fatigue, ventilatory failure, and death [16].

Studies have demonstrated associations between sEMG-derived measures and mortality in ICU patients, COPD populations, and neuromuscular disease [1][15].

sEMG could be incorporated into risk stratification models to identify high-risk patients and guide early intervention. However, further validation is needed.