Utku Şükrü Yavuz

Motor unit (MU) activity during electrically or mechanically evoked reflexes is used to examine the relationship between neural inputs and MU properties. However, variations in single-MU reflex amplitudes are not fully understood and limit their reliability in determining the input-output relation of motor neurons (MNs). Using experiments and computer simulations, we investigated (i) whether MN discharge statistics and muscle activation explain the variability of reflex amplitude estimates and (ii) whether these variations are reflected differently across distinct reflex amplitude estimation methods. We analyzed MU spike trains extracted from isometric contractions of the tibialis anterior muscle at 10 % and 20 % MVC (maximum voluntary contraction). Estimating reflex amplitudes based on the peristimulus frequencygram (PSF) at 10 % MVC, the linear regression between discharge rate (DR) and reflex amplitude was always positive, with p < 0.05 in 3 out of 6 subjects; however, the linear correlation was inconsistent at 20 % MVC. We thereby observed that inter-subject variability was associated with the coefficient of variation of the interspike intervals. Furthermore, the linear correlation between DR and peristimulus time histogram (PSTH) based reflex amplitudes was inconsistent for both contraction forces. To obtain further insights into the influence of MN properties, we simulated reflexes in a heterogeneous MN population using electrical circuit models and varied MN inputs. The simulations indicate that, besides mean input current and membrane noise, MN properties also contribute to the variability of reflex amplitude estimates. The MN heterogeneity is well captured by PSF-based reflex estimates but not by PSTH-based ones. These results show that variations in amplitude estimates of individual MU reflexes are due to complex interactions between intrinsic and extrinsic factors. As PSF-based reflex amplitude estimates reflect the MN size distribution, tracking PSF-based reflex amplitudes at fixed MVC levels across individual subjects could serve as a marker for investigating spinal adaptations under (patho)physiological conditions. Author summary: Motor neurons are specialized nerve cells that control human movement. Each motor neuron activates a specific set of muscle fibers, and the functional unit consisting of a motor neuron and muscle fibers is called a motor unit. We can observe the activity of motor neurons in humans by decomposing the electrical activity of muscles (the electromyogram) into contributions from individual motor units. Reflex responses of motor units are often used to study the input-output relation of motor neurons in humans. We used a combination of experiments and computer simulations to study the factors that influence the reflex amplitude of motor units during an excitatory reflex. We found that the reflex amplitude is non-linearly influenced by a number of intrinsic and extrinsic factors, e. g., motor neuron size, but also the muscle force. Additionally, we found that these factors have different effects on the results of the two common methods used to calculate the reflex amplitude. These results provide guidance on choosing a suitable evaluation method and on interpreting reflex experiments.

Understanding how the central nervous system controls complex movements, such as walking, remains a fundamental challenge. Although motor units (MUs) are well-studied in isometric tasks, their role in generating joint moments during functional dynamic movements is unclear, partly due to non-stationary conditions. Filling this knowledge gap is essential for studying neural control of walking at the cellular level and for guiding neurorehabilitation and robotic interventions. We developed a validated framework for decoding high-density electromyography (HD-EMG) into individual MU spike trains during walking. We assessed both static decoding and an adaptive algorithm that continuously tracks time-varying action potentials, validating the results against fine-wire intra-muscular EMG (iEMG). We then examined neuromechanical delays (NMD) across walking speeds and established MU-driven neuromusculoskeletal models to determine the biomechanical consequences of the decoded MU firing patterns. Five healthy adults walked at multiple speeds while we recorded HD-EMG, iEMG, motion capture, and ground reaction forces. Both static and adaptive decompositions yielded comparable MU spike trains. Median rate of agreement was higher and false positives were lower for the adaptive approach, whereas false negatives were lower for the static approach. Although not statistically significant (p>0.05), these trends suggest the adaptive method may better handle nonstationarities. NMD decreased with speed, indicating coherent acceleration of neural-to-mechanical transmission. MU-driven models reproduced ankle joint moments with substantially lower normalized RMSE than conventional EMG-envelope models (p<0.05), with no difference between static and adaptive models. The close match between MU-derived and measured joint moments confirms that the decoded neural drive captures functional motor control rather than numerical artifacts. This study validates MU decomposition during walking and establishes a MU-driven model-based framework for investigating spinal motor control under dynamic, functionally relevant conditions. By bridging cellular-level neural activity with biomechanical outcomes, our approach opens new avenues for advancing neuro-rehabilitation, assistive technology, and our understanding of human locomotion.

The development of human-machine interfaces is key to creating prosthetics that adapt to user intent. However, there remains much incongruity between muscle activity and the resulting torque. This study employs high-density surface electromyography (HD-sEMG) in conjunction with the Fast-ICA algorithm to decompose motor units (MUs), which play a crucial role in the generation of movement. Multilayer perceptron (MLP) architectures are evaluated for predicting joint torque from MU activity, and transfer learning is explored to enhance training performance. Results show that transfer learning enhances training efficiency and that joint torque can be accurately predicted from MU activity, thereby advancing the potential of control systems in assistive technologies.

This research explores the feasibility of combining surface electromyography (sEMG) and multi-frequency electrical impedance myography (mfEIM) for predicting joint angles and torques. Methods: We utilize a current-limited multi-frequency electrical impedance spectrometer to simultaneously perform mfEIM on the biceps brachii and triceps brachii at 14 frequencies. The same system also measures the sEMG signals using the same electrodes. Measurements are conducted while subjects perform tasks in a one-degree-of-freedom exoskeleton, which enables the measurement of elbow joint torque and angle. We train time-delay neural networks to model the relationships between mfEIM, sEMG, joint torque, and joint angle. Results and Conclusions: The results demonstrate that the sEMG signal can predict joint torque and that mfEIM can predict joint angle. Additionally, we find indications that a combination of sEMG and mfEIM enhances joint torque predictions compared to using sEMG alone. We also determine the delays between the sEMG, mfEIM signals, and the joint angle and torque, finding that the joint torque signal has a delay relative to the sEMG signal of 155ms during an isometric exercise and 125ms during dynamic tasks. Significance: These results suggest that combining sEMG and mfEIM measurements could provide additional insights during biomedical experiments.

Background: The impact of missing data on individual continuous glucose monitoring (CGM) data is unknown but can influence clinical decision-making for patients. Objective: We aimed to investigate the consequences of data loss on glucose metrics in individual patient recordings from continuous glucose monitors and assess its implications on clinical decision-making. Methods: The CGM data were collected from patients with type 1 and 2 diabetes using the FreeStyle Libre sensor (Abbott Diabetes Care). We selected 7-28 days of 24 hours of continuous data without any missing values from each individual patient. To mimic real-world data loss, missing data ranging from 5% to 50% were introduced into the data set. From this modified data set, clinical metrics including time below range (TBR), TBR level 2 (TBR2), and other common glucose metrics were calculated in the data sets with and that without data loss. Recordings in which glucose metrics deviated relevantly due to data loss, as determined by clinical experts, were defined as expert panel boundary error (εEPB). These errors were expressed as a percentage of the total number of recordings. The errors for the recordings with glucose management indicator <53 mmol/mol were investigated. Results: A total of 84 patients contributed to 798 recordings over 28 days. With 5%-50% data loss for 7-28 days recordings, the εEPB varied from 0 out of 798 (0.0%) to 147 out of 736 (20.0%) for TBR and 0 out of 612 (0.0%) to 22 out of 408 (5.4%) recordings for TBR2. In the case of 14-day recordings, TBR and TBR2 episodes completely disappeared due to 30% data loss in 2 out of 786 (0.3%) and 32 out of 522 (6.1%) of the cases, respectively. However, the initial values of the disappeared TBR and TBR2 were relatively small (<0.1%). In the recordings with glucose management indicator <53 mmol/mol the εEPB was 9.6% for 14 days with 30% data loss. Conclusions: With a maximum of 30% data loss in 14-day CGM recordings, there is minimal impact of missing data on the clinical interpretation of various glucose metrics. Trial Registration: ClinicalTrials.gov NCT05584293; https://clinicaltrials.gov/study/NCT05584293

Background: Data loss in wearable sensors is an inevitable problem that leads to misrepresentation during diabetes health monitoring. We systematically investigated missing wearable sensors data to get causal insight into the mechanisms leading to missing data. Methods: Two-week-long data from a continuous glucose monitor and a Fitbit activity tracker recording heart rate (HR) and step count in free-living patients with type 2 diabetes mellitus were used. The gap size distribution was fitted with a Planck distribution to test for missing not at random (MNAR) and a difference between distributions was tested with a Chi-squared test. Significant missing data dispersion over time was tested with the Kruskal–Wallis test and Dunn post hoc analysis. Results: Data from 77 subjects resulted in 73 cleaned glucose, 70 HR and 68 step count recordings. The glucose gap sizes followed a Planck distribution. HR and step count gap frequency differed significantly (p < 0.001), and the missing data were therefore MNAR. In glucose, more missing data were found in the night (23:00–01:00), and in step count, more at measurement days 6 and 7 (p < 0.001). In both cases, missing data were caused by insufficient frequency of data synchronization. Conclusions: Our novel approach of investigating missing data statistics revealed the mechanisms for missing data in Fitbit and CGM data.

Postinhibitory excitation is a transient overshoot of a neuron’s baseline firing rate following an inhibitory stimulus and can be observed in vivo in human motoneurons. However, the biophysical origin of this phenomenon is still unknown and both reflex pathways and intrinsic motoneuron properties have been proposed. We hypothesized that postinhibitory excitation in motoneurons can be facilitated by hyperpolarization-activated inward currents (h-currents). Using an electrical circuit model, we investigated how h-currents can modulate the postinhibitory response of motoneurons. Further, we analyzed the spike trains of human motor units from the tibialis anterior muscle during reciprocal inhibition. The simulations revealed that the activation of h-currents by an inhibitory postsynaptic potential can cause a short-term increase in a motoneuron’s firing probability. This result suggests that the neuron can be excited by an inhibitory stimulus. In detail, the modulation of the firing probability depends on the time delay between the inhibitory stimulus and the previous action potential. Further, the postinhibitory excitation’s strength correlates with the inhibitory stimulus’s amplitude and is negatively correlated with the baseline firing rate as well as the level of input noise. Hallmarks of h-current activity, as identified from the modeling study, were found in 50% of the human motor units that showed postinhibitory excitation. This study suggests that h-currents can facilitate postinhibitory excitation and act as a modulatory system to increase motoneuron excitability after a strong inhibition.

Alpha band oscillations in shared synaptic inputs to the alpha motor neuron pool can be considered an involuntary source of noise that hinders precise voluntary force production. This study investigated the impact of altering muscle length on the shared synaptic oscillations to spinal motor neurons, particularly in the physiological tremor band. Fourteen healthy individuals performed low-level dorsiflexion contractions at ankle joint angles of 90° and 130°, while high-density surface electromyography (HD-sEMG) was recorded from the tibialis anterior (TA). We decomposed the HDsEMG into motor units spike trains and calculated the motor units’ coherence within the delta (1-5 Hz), alpha (5-15 Hz) and beta (15-35 Hz) bands. Additionally, torque steadiness and torque spectral power within the tremor band was quantified. Results showed no significant differences in torque steadiness between 90° and 130°. In contrast, alpha band oscillations in both synaptic inputs and force output decreased as the length of the TA was moved from shorter (90°) to longer (130°), with no changes in delta and beta bands. In a second set of experiments, evoked twitches were recorded with the ankle joint at 70° and 130°, revealing longer twitch durations in the longer muscle lengthen condition compared to the shorter. These experimental results, supported by a simple computational simulation, suggest that increasing muscle length enhances the muscle’s low-pass filtering properties, influencing the oscillations generated by the Ia afferent feedback loop. Therefore, this study provides valuable insights into the interplay between muscle biomechanics and neural oscillations.

Understanding how motor units (MUs) contribute to skeletal mechanical force is crucial for unraveling the underlying mechanism of human movement. Alterations in MU firing, contractile and force-generating properties emerge in response to physical training, ageing or injury. However, how changes in MU firing and twitch properties dictate skeletal muscle force generation in healthy and impaired individuals remains an open question. In this work, we present a MU-specific approach to identify firing and twitch properties of MU samples and employ them to decode musculoskeletal function in vivo. First, MU firing events were decomposed offline from high-density electromyography (EMG) of six lower leg muscles involved in ankle plantar-dorsi flexion.We characterized their twitch responses based on the statistical distributions of their firing properties and employed them to compute MU-specific activation dynamics. Subsequently, we decoded ankle joint moments by linking our framework to a subject-specific musculoskeletal model. We validated our approach at different ankle positions and levels of activation and compared it with traditional EMG-driven models. Our proposed MU-specific formulation achieves higher generalization across conditions than the EMG-driven models, with significantly lower coefficients of variation in torque predictions. Furthermore, our approach shows distinct neural strategies across a large repertoire of contractile conditions in different muscles. Our proposed approach may open new avenues for characterizing the relationship between MU firing and twitch properties and their influence on force capacity. This can facilitate the development of targeted rehabilitation strategies tailored to individuals with specific neuromuscular conditions.