Bone marrow-derived mesenchymal stem cells (BMSCs), owing to their osteoblastogenic differentiation potential, are crucial for maintaining bone homeostasis and remodeling. Nevertheless, in aging and age-related bone diseases like osteoporosis, BMSCs show significantly diminished osteogenic potential, with a concomitant increase in adipogenic differentiation. The aged BMSCs also become desensitized to BMP2 stimulation to a large extent and exhibit aberrations in BMP2 signaling. However, the molecular mechanisms facilitating this shift in lineage commitment and mediating the cellular dysfunctions remain elusive. This knowledge gap hinders the development of regenerative strategies for skeletal aging and osteoporosis. This study employed an integrative tandem mass tag (TMT)-based phosphoproteomic and total proteomic profiling on BMSCs isolated from young (6-month) and aged (15-month) C57BL/6 (B6) mice to elucidate global alterations in both protein activity and expression. The analysis identified more than 500 proteins that underwent significant alterations (BH-adjusted p-value < 0.05) either in phosphorylation or expression between young and aged BMSCs. Many lineage-specific markers also underwent changes in both phosphorylation and expression with aging. Additionally, key biological processes, including cellular metabolism, clathrin-mediated endocytosis, and nucleocytoplasmic transport mechanisms, were enriched for the deregulated proteins. Signaling proteins, ERK-1/2, had increased activating phosphorylation in the aged BMSCs, while transcription factors Lrrfip1, Ruvbl1, and Ruvbl2 also exhibited dysregulated activity and abundance in the aged BMSCs. The findings from the study adds significant mechanistic insights into how aging disrupts signal transduction, metabolism, and transcriptional program in BMSCs, contributing to age-associated loss of bone mass and reduced skeletal regenerative capabilities. Through the identification of key mediators of BMSC dysfunction seen in aging, this work offers a strong foundation in devising potential therapeutic strategies to restore diminished osteogenic potential and treat osteoporosis.
Emotion recognition from physiological signals has immense applications in healthcare and human-computer interaction. We developed an electrodermal activity (EDA)-graph signal processing pipeline that produces highly sensitive features for detecting the affective dimensions (arousal and valence) of emotions. Using the Continuously Annotated Signals of Emotion dataset, we compared our graph-based EDA features (EDA-graph) with traditional time- and frequency-domain EDA features and features derived from other signals (heart rate variability, pulse transit time, electromyography, skin temperature, and respiration) for detecting affective dimensions using machine learning regression models. The EDA-graph features showed superior performance in continuous affective dimension recognition compared to the most accurate state-of-the-art models, achieving RMSE values of 0.801 for arousal and 0.714 for valence. Furthermore, we used a variety of traditional and recently published datasets collected in laboratory and ambulatory settings to perform a comprehensive evaluation of the robust generalization capabilities of our approach across different emotional contexts. The models demonstrated exceptional performance in classifying emotional states across the datasets, achieving 98.2% accuracy in detecting positive, negative, and mixed emotions; 92.75% in discriminating between emotions (relaxed, amused, bored, scared, and neutral); and 86.54% in detecting stress vs. no stress. These results highlight the potential of a graph-based analysis of EDA in emotion recognition systems in different contexts, especially for real-world applications.
Hexavalent chromium [Cr(VI)] is the toxic form of chromium often used in industry for its hardness, bright colors, and anticorrosive properties. Cr(VI) is a known human lung carcinogen, making its inhalation an occupational hazard. Growing evidence emphasizes the neurotoxic potential of Cr(VI), though it is not linked to brain cancers. Few studies consider neurotoxicity in chromate workers, reporting impaired olfactory discrimination and an increased risk of death from mental health disorders. A major factor limiting translation of most rodent Cr(VI) studies to human populations has to do with vitamin C, which can reduce the toxic Cr(VI) to non-toxic Cr(III). Rats and mice synthesize vitamin C and are likely more resistant to Cr(VI) than humans. Here, we considered Cr(VI) neurotoxicity in guinea pigs (Cavia porcellus), which do not endogenously synthesize vitamin C. We exposed Hartley guinea pigs (both sexes) to occupationally relevant concentrations of Cr(VI) via oropharyngeal aspiration weekly for 90 days. We observed behavioral effects in the open field assay, elevated plus maze, Y-maze, and novel object recognition test during weeks 9-12 of exposure. After euthanasia, we assessed Cr accumulation and essential metal dyshomeostasis in the hippocampus. We observed significantly increased hippocampal Cr accumulation in females, while males exhibited essential metal dyshomeostasis.
Tetracenomycins are anticancer polyketides that arrest cancer cell proliferation via binding to the large mammalian ribosomal subunit near the polypeptide exit channel. The tetracenomycins are natural products that many members of the actinomycete family produce. The first goal of this study was to improve the biosynthesis of tetracenomycin analogs via metabolic engineering. The second goal was to probe more deeply into the antiproliferative activity of tetracenomycin aglycones. The tetracenomycins were assessed via several assays, including cell viability assays, clonogenic assays, and flow cytometry apoptosis assays. The data suggest that tetracenomycins C and X inhibit cell proliferation and arrest cell growth, supporting their cytostatic action mechanism. In addition, tetracenomycins C and X induced degeneration of 3D spheroid cultures and exhibited concentration-dependent inhibition of cell survival and colony formation in clonogenic assays. This work demonstrates that tetracenomycins act mainly as cytostatic rather than apoptotic agents.
Introduction: Direct pulp capping (DPC) aims to preserve the vitality of the dental pulp by placing a protective biocompatible material over the exposed pulp tissue to facilitate healing. There are several calcium-silicate materials that have been designed to promote mineralization and the regulation of inflammation. These have strong potential for the repair and regeneration of dental pulp. Among them, Biodentine (BD) and EndoSequence RRM Putty (ES) have been found to promote in vitro and in vivo mineralization while minimizing some of the limitations of the first-generation calcium-silicate-based materials. Theracal-LC (TLC), a light-cured, resin-modified calcium-silicate material, is a newer product with potential to improve the clinical outcomes of DPC, but existing studies have reported conflicting findings regarding its biocompatibility and ability to support pulpal healing in direct contact with the pulp. A comprehensive assessment of the biocompatibility and pulpal protection provided by these three capping materials has not yet been performed. Aim: We aimed to quantify the inflammatory response, dentin bridge formation, and material adaptation following DPC using three calcium-silicate materials: ES, BD, and TLC. Materials and Methods: DPC was performed on the maxillary first molar of C57BL/6 female mice. Maxilla were collected and processed at 1 and 21 days post-DPC. The early inflammatory response was measured 24 h post-procedure using confocal imaging of anti-Lys6G6C, which indicates the extent of neutrophil and monocyte infiltration. Reparative mineralized bridge formation was assessed at 21 days post-procedure using high-resolution micro-computed tomography (micro-CT) and histology. Lastly, the homogeneity of the capping materials was evaluated by quantifying voids in calcium-silicate restorations using micro-CT. Results: DPC using TLC induced less infiltration of Lys6G6C+ cells at 24 h than BD or ES. BD promoted higher volumes of tertiary dentin than TLC, but TLC and ES showed no significant differences in volume. No differences were observed in material adaptation and void spaces among the three capping materials. Conclusions: All three materials under investigation supported pulp healing and maintained marginal integrity. However, TLC induced a lower inflammatory response on day 1 and induced similar levels of tertiary dentin to ES. These observations challenge the common perception that resin-based capping materials are not suitable for direct pulp capping. Our findings underscore the need to balance biological responses with physical properties when selecting pulp capping materials to improve long-term clinical success.
Asymmetries are common during squats following anterior cruciate ligament reconstruction (ACLR). This study examined interlimb loading differences between squat phases at 6 months post-ACLR. Thirty-five participants performed bodyweight squats at self-selected speed and were analyzed using 3D motion capture. Vertical ground reaction force impulse (vGRFi), external knee flexion moment impulse (KFMi) and hip-to-knee flexion moment impulse ratio (HKRi) were calculated, along with interlimb ratios (ILR). Squat phase durations were also recorded. Paired t-tests and ANCOVA (controlling for time) were used to compare biomechanical variables across squat phases. Greater asymmetry was observed during ascent for vGRFi ILR (p = 0.045), KFMi ILR (p < 0.001) and HKRi ILR (p = 0.006). The ascent phase was faster than descent (p = 0.036). After adjusting for time, phase-related differences in ILRs were no longer significant. These findings suggest that greater limb and knee-specific loading asymmetries occur during the ascent phase of squats but may be influenced by movement speed. Importantly, significant knee-specific loading asymmetries persisted regardless of squat phase. At 6 months post-ACLR, addressing neuromuscular control and movement speed during rehabilitation may help reduce biomechanical imbalances during closed kinetic chain exercises.
The basic reproduction number (R_0) is an epidemiological metric that represents the average number of new infections caused by a single infectious individual in a completely susceptible population. The methodology for calculating this metric is well-defined for numerous model types, including, most prominently, Ordinary Differential Equations (ODEs). The basic reproduction number is used in disease modeling to predict the potential of an outbreak and the transmissibility of a disease, as well as by governments to inform public health interventions and resource allocation for controlling the spread of diseases. A Petri Net (PN) is a directed bipartite graph where places, transitions, arcs, and the firing of the arcs determine the dynamic behavior of the system. Petri Net models have been an increasingly used tool within the epidemiology community. However, no generalized method for calculating R_0 directly from PN models has been established. Thus, in this paper, we establish a generalized computational framework for calculating R_0 directly from Petri Net models. We adapt the next-generation matrix method to be compatible with multiple Petri Net formalisms, including both deterministic Variable Arc Weight Petri Nets (VAPNs) and stochastic continuous-time Petri Nets (SPNs). We demonstrate the method's versatility on a range of complex epidemiological models, including those with multiple strains, asymptomatic states, and nonlinear dynamics. Crucially, we numerically validate our framework by demonstrating that the analytically derived R_0 values are in strong agreement with those estimated from simulation data, thereby confirming the method's accuracy and practical utility.
Interdisciplinary research catalyzes innovation in mobile health (mHealth) by converging medical, technological, and social science expertise, driving critical advancements in this multifaceted field. Our longitudinal analysis evaluates how the NIH mHealth Training Institute (mHTI) program stimulates changes in research trajectories through a computational examination of 16,580 publications from 176 scholars (2015–2022 cohorts). We develop a hybrid analytical framework combining large language model (LLM) embeddings, Uniform Manifold Approximation and Projection (UMAP) dimensionality reduction, and Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) clustering to construct a semantic research landscape containing 329 micro-topics aggregated into 14 domains. GPT-4o-assisted labeling identified mHealth-related publications occupying central positions in the semantic space, functioning as conceptual bridges between disciplinary clusters such as clinical medicine, public health, and technological innovation. Kernel density estimation of research migration patterns revealed 63.8% of scholars visibly shifted their publication focus toward mHealth-dense regions within three years post-training. The reorientation demonstrates mHTI’s effectiveness in fostering interdisciplinary intellect with sustained engagement, evidenced by growth in mHealth-aligned publications from the mHTI scholars. Our methodology advances science of team science research by demonstrating how LLM-enhanced topic modeling coupled with spatial probability analysis can track knowledge evolution in interdisciplinary fields. The findings provide empirical validation for structured training programs’ capacity to stimulate convergent research, while offering a scalable framework for evaluating inter/transdisciplinary initiatives. The dual contribution bridges methodological innovation in natural language processing with practical insights for cultivating next-generation mHealth scholarship.
In the aging population, imbalance leading to falls is of critical concern; thus, it is imperative to determine and quantify neuromuscular changes because of rehabilitative balance training. (1) Background: Previous studies that have examined changes in balance due to rehabilitative training placed a focus on clinical measures (i.e., behavioral, kinetic, and kinematic outputs); however, irregularities due to abnormalities of underlying neural origin were unrevealed by the aforementioned measures. (2) Methods: Examining muscle activity was used to determine strategies pre- and post-six weeks of balance training in twenty-three healthy older adults (69.5 ± 5.7 years old) and five survivors of stroke (66.4 ± 9.48 years old). Surface electromyographic (sEMG) signals were recorded from eight of the lower limb muscles while participants performed forward walking (FW), forward tandem walking (FT-W), backward walking (BW), and backward tandem walking (BT-W) tasks. The sEMG data were then conditioned and muscle synergies were extracted using non-negative matrix factorization (NNMF). (3) Results: It was observed that muscle synergies and activation patterns changed for pre- versus post-balance training in older adults (i.e., healthy individuals and those who had suffered from stroke). (4) Conclusions: From our findings, it was indicative that muscle activation and muscle synergies could be used to quantify and inform rehabilitative balance training in older adults.
Falls are the leading cause of injury and mortality among older adults—one in four individuals 65 years old and above experiences falls. Thus, balance training interventions that improve balance ability and reduce the risk of falls are of critical importance. Through two complementary interventions, our research sought to determine the effects of Virtual Reality (VR) compared to Robotic-Assisted Balance Training (RABT) on balance ability and balance confidence in older adults. The VR intervention utilized Oculus headsets to create immersive balance exercises, while the RABT employed a multidirectional overground robotic system (NaviGAITor). Participants (aged 60–85 years old) underwent a 6-week training protocol consisting of two 30 min sessions per week. Balance ability was quantified using center of pressure (COP) parameters and the Balance Error Scoring System (BESS), while balance confidence was measured using the Activities-Specific Balance Confidence (ABC) scale. Results indicated no statistically significant differences between the training methods. However, the RABT group showed trends toward enhanced balance performance, with observed decreases in mediolateral (ML) maximum displacement during wide stance conditions and reductions in BESS errors on both firm and foam surfaces. The VR group demonstrated significant changes in ML RMS values during tandem stance (p = 0.045) and improved participants’ relationship with perceived and actual balance ability (increased correlation between BESS errors and ABC scores from R2 = 0.00 pre-training to R2 = 0.65 post-training). Balance confidence did not significantly increase in either group. These findings suggest that while RABT may trend toward improvements in objective balance parameters, VR training appears to enhance participants’ perceptual accuracy of their balance capabilities.
Preserving dental pulp vitality is crucial in pediatric and adolescent dentistry to promote long-term oral health and reduce the need for invasive procedures. Vital pulp therapy (VPT) enhances pulp healing and dentin formation through advanced pulp capping materials. While calcium hydroxide-based materials laid the foundation for VPT, calcium silicate-based materials like mineral trioxide aggregate, Biodentine, and TheraCal offer superior biocompatibility and sealing properties. Recent advancements focus on regenerative strategies that enhance biocompatibility, antibacterial efficacy, and anti-inflammatory effects. Tissue engineering approaches, including stem cells, growth factors, and peptide-based scaffolds, are being explored to improve pulp regeneration and long-term treatment success. This review highlights recent developments in VPT for pediatric and adolescent patients, emphasizing minimally invasive techniques, clinical challenges, and the potential of emerging biomaterials. Continued research into biomaterial efficacy and regenerative capabilities holds promise for advancing VPT, ensuring more effective and biologically driven treatment strategies for young patients.
Although upper limb impairment is one of the most common deficits post-stroke and contributes substantially to diminished functional independence, many survivors receive low dosages of upper limb task training in the inpatient setting. This study evaluates the feasibility and user experience of a progressive-challenge cued activity program, delivered via wearable technology, to promote upper limb activity in an inpatient rehabilitation facility (IRF) post-stroke. Participants (N = 30) wore our wearable system Souvenir, which provided vibrotactile cues to prompt activity in the more-involved arm during idle time. Compliance with the program was high (94% in the IRF), and the system successfully prompted increased activity, as evidenced by significantly higher post-cue response rates compared to pre-cue activity rates (mean difference = 35.1%, t(28) = 9.398, p < 0.001). User experience was positive, with participants reporting high usability, satisfaction, and motivation. Follow-up data collected in unstructured home settings (n = 23) demonstrated continued high compliance (96%) and favorable user experience. These findings suggest that Souvenir and its cued activity program can effectively convert idle time into therapeutic activity while minimizing caregiver burden. Future research should focus on enhancing user engagement and evaluating the clinical efficacy of this approach in improving functional outcomes post-stroke.
This study investigates the impact of subglottic stenosis (SGS) on voice production using a subject-specific laryngeal and airway model. Direct numerical simulations of fluid–structure–acoustic interaction were employed to analyze glottal flow dynamics, vocal fold vibration, and acoustics under realistic conditions. The model accurately captured key physiological parameters, including the glottal flow rate, vocal fold vibration patterns, and the first four formant frequencies. Simulations of varying SGS severity revealed that up to 75% stenosis, vocal function remains largely unaffected. However, at 90% severity, significant changes in glottal flow and acoustics were observed, with vocal fold vibration remaining stable. At 96%, severe reductions in glottal flow and acoustics, along with marked changes in vocal fold dynamics, were detected. Flow resistance, the ratio of glottal to stenosis area, and pressure drop across the vocal folds were identified as critical factors influencing these changes. The use of anatomically realistic airway and vocal fold geometries revealed that while anatomical variations minimally affect voice production at lower stenosis grades, they become critical at severe stenosis levels (>90%), particularly in capturing distinct anterior–posterior opening patterns and focused jet effects that alter glottal dynamics. These findings suggest that while simplified models suffice for analyzing mild to moderate stenosis, patient-specific geometric details are essential for accurate prediction of vocal fold dynamics in severe cases.
Degenerative cervical myelopathy (DCM) is characterized by progressive neurological dysfunction, yet the contribution of intramedullary stress and strain during neck motion remains unclear. This study used patient-specific finite element models (FEMs) of the cervical spine and spinal cord to examine the relationship between spinal cord biomechanics and neurological dysfunction. Twenty DCM patients (mean age 62.7 ± 11.6 years; thirteen females) underwent pre-surgical MRI-based modeling to quantify von Mises stress and maximum principal strains at the level of maximum spinal cord compression during simulated neck flexion and extension. Pre-surgical functional assessments included hand sensation, dexterity, and balance. During flexion, the mean intramedullary stress and strain at the level of maximum compression were 7.6 ± 3.7 kPa and 4.3 ± 2.0%, respectively. Increased intramedullary strain during flexion correlated with decreased right-hand sensation (r = −0.58, p = 0.014), impaired right-hand dexterity (r = −0.50, p = 0.048), and prolonged dexterity time (r = 0.52, p = 0.039). Similar correlations were observed with intramedullary stress. Patients with severe DCM exhibited significantly greater stress during flexion than those with mild/moderate disease (p = 0.03). These findings underscore the impact of dynamic spinal cord biomechanics on neurological dysfunction and support their potential utility in improving DCM diagnosis and management.
Hematoma expansion (HE) is an independent predictor of poor outcomes and a modifiable treatment target in intracerebral hemorrhage (ICH). Evaluating HE in large datasets requires segmentation of hematomas on admission and follow-up CT scans, a process that is time-consuming and labor-intensive in large-scale studies. Automated segmentation of hematomas can expedite this process; however, cumulative errors from segmentation on admission and follow-up scans can hamper accurate HE classification. In this study, we combined a tandem deep-learning classification model with automated segmentation to generate probability measures for false HE classifications. With this strategy, we can limit expert review of automated hematoma segmentations to a subset of the dataset, tailored to the research team’s preferred sensitivity or specificity thresholds and their tolerance for false-positive versus false-negative results. We utilized three separate multicentric cohorts for cross-validation/training, internal testing, and external validation (n = 2261) to develop and test a pipeline for automated hematoma segmentation and to generate ground truth binary HE annotations (≥3, ≥6, ≥9, and ≥12.5 mL). Applying a 95% sensitivity threshold for HE classification showed a practical and efficient strategy for HE annotation in large ICH datasets. This threshold excluded 47–88% of test-negative predictions from expert review of automated segmentations for different HE definitions, with less than 2% false-negative misclassification in both internal and external validation cohorts. Our pipeline offers a time-efficient and optimizable method for generating ground truth HE classifications in large ICH datasets, reducing the burden of expert review of automated hematoma segmentations while minimizing misclassification rate.
The simulation of gas flow in the bronchial tree using computational fluid dynamics (CFD) has become a useful tool for the analysis of gas flow mechanics, structural deformation, ventilation, and particle deposition for drug delivery during spontaneous and assisted breathing. CFD allows for new hypotheses to be tested in silico, and detailed results generated without performing expensive experimental procedures that could be potentially harmful to patients. Such computational techniques are also useful for analyzing structure–function relationships in healthy and diseased lungs, assessing regional ventilation at various time points over the course of clinical treatment, or elucidating the changes in airflow patterns over the life span. CFD has also allowed for the development and use of image-based (i.e., patient-specific) models of three-dimensional (3D) airway trees with realistic boundary conditions to achieve more meaningful and personalized data that may be useful for planning effective treatment protocols. This focused review will present a summary of the techniques used in generating realistic 3D airway tree models, the limitations of such models, and the methodologies used for CFD airflow simulation. We will discuss mathematical and image-based geometric models, as well as the various boundary conditions that may be imposed on these geometric models. The results from simulations utilizing mathematical and image-based geometric models of the airway tree will also be discussed in terms of similarities to actual gas flow in the human lung.
Postural imbalance is a leading cause of injury in older adults. Our study investigated the effectiveness of virtual reality (VR)-based interventions on balance ability in this population. Here, we examined 21 older, healthy adults (75.8 ± 5.2 years old). Participants performed 6 weeks of balance training, twice per week for 30 min; the experimental group donned an Oculus VR headset during the training while control participants did not. To assess balance ability, a force platform measured displacement of the center of pressure (COP) during quiet standing in double-leg, tandem, and single-leg stances with eyes closed pre- and post-assessment. COP measurements included mediolateral (ML) and anterior–posterior (AP) directions for root mean square (RMS), peak-to-peak displacement (MAXD), total excursion (TE), and 95% confidence area ellipse (AE) parameters. Post-training assessments showed improvements (significant decreases) in the COP parameters. Control group COP parameters improved in various stances ranging from a 3% to 40% decrease on average. The VR group improved MAXD, TE, and 95% AE ranging from a 5% to 47% decrease, on average, across various stances post- compared to pre-training. VR-based exercise training programs may encourage older adults to engage in mobility exercises, leading to a reduced risk of falls or injuries.
This paper describes the principles behind the radio-frequency (RF) sensing of bacterial biofilms in pipes and heat exchangers in a dairy processing plant using an electromagnetic simulation. Biofilm formation in dairy processing plants is a common issue where the absence of timely detection and subsequent cleaning can cause serious illness. Biofilms are known for causing health issues and cleaning requires a large volume of water and harsh chemicals. In this work, milk transportation pipes are considered circular waveguides, and pasteurizers/heat exchangers are considered resonant cavities. Simulations were carried out using the CST studio suite high-frequency solver to determine the effectiveness of the real-time RF sensing. The respective dielectric constants and loss tangents were applied to milk and biofilm. In our simulation, it was observed that a 1 µm thick layer of biofilm in a milk-filled pipe shifted the reflection coefficient of a 10.16 cm diameter stainless steel circular waveguide from 0.229 GHz to 0.19 GHz. Further sensitivity analysis revealed a shift in frequency from 0.8 GHz to 1.2 GHz for a film thickness of 5 µm to 10 µm with the highest wave reflection (S11) peak of ≈−120 dB for a 6 µm thick biofilm. A dielectric patch antenna to launch the waves into the waveguide through a dielectric window was also designed and simulated. Simulation using the antenna demonstrated a similar S11 response, where a shift in reflection coefficient from 0.229 GHz to 0.19 GHz was observed for a 1 µm thick biofilm. For the case of the resonant cavity, the same antenna approach was used to excite the modes in a 0.751 m × 0.321 m × 170 m rectangular cavity with heat exchange fins and filled with milk and biofilm. The simulated resonance frequency shifted from 1.52 GHz to 1.54 GHz, for a film thickness varying from 1 µm to 10 µm. This result demonstrated the sensitivity of the microwave detection method. Overall, these results suggest that microwave sensing has promise in the rapid, non-invasive, and real-time detection of biofilm formation in dairy processing plants.
A computational neuromuscular control system that generates lung pressure and three intrinsic laryngeal muscle activations (cricothyroid, thyroarytenoid, and lateral cricoarytenoid) to control the vocal source was developed. In the current study, LeTalker, a biophysical computational model of the vocal system was used as the physical plant. In the LeTalker, a three-mass vocal fold model was used to simulate self-sustained vocal fold oscillation. A constant /ə/ vowel was used for the vocal tract shape. The trachea was modeled after MRI measurements. The neuromuscular control system generates control parameters to achieve four acoustic targets (fundamental frequency, sound pressure level, normalized spectral centroid, and signal-to-noise ratio) and four somatosensory targets (vocal fold length, and longitudinal fiber stress in the three vocal fold layers). The deep-learning-based control system comprises one acoustic feedforward controller and two feedback (acoustic and somatosensory) controllers. Fifty thousand steady speech signals were generated using the LeTalker for training the control system. The results demonstrated that the control system was able to generate the lung pressure and the three muscle activations such that the four acoustic and four somatosensory targets were reached with high accuracy. After training, the motor command corrections from the feedback controllers were minimal compared to the feedforward controller except for thyroarytenoid muscle activation.
Emotion recognition, a rapidly evolving domain in digital health, has witnessed significant transformations with the advent of personalized approaches and advanced machine learning (ML) techniques. These advancements have shifted the focus from traditional, generalized models to more individual-centric methodologies, underscoring the importance of understanding and catering to the unique emotional expressions of individuals. Our study delves into the concept of model personalization in emotion recognition, moving away from the one-size-fits-all approach. We conducted a series of experiments using the Emognition dataset, comprising physiological and video data of human subjects expressing various emotions, to investigate this personalized approach to affective computing. For the 10 individuals in the dataset with a sufficient representation of at least two ground truth emotion labels, we trained a personalized version of three classical ML models (k-nearest neighbors, random forests, and a dense neural network) on a set of 51 features extracted from each video frame. We ensured that all the frames used to train the models occurred earlier in the video than the frames used to test the model. We measured the importance of each facial feature for all the personalized models and observed differing ranked lists of the top features across the subjects, highlighting the need for model personalization. We then compared the personalized models against a generalized model trained using data from all 10 subjects. The mean F1 scores for the personalized models, specifically for the k-nearest neighbors, random forest, and dense neural network, were 90.48%, 92.66%, and 86.40%, respectively. In contrast, the mean F1 scores for the generic models, using the same ML techniques, were 88.55%, 91.78% and 80.42%, respectively, when trained on data from various human subjects and evaluated using the same test set. The personalized models outperformed the generalized models for 7 out of the 10 subjects. The PCA analyses on the remaining three subjects revealed relatively little facial configuration differences across the emotion labels within each subject, suggesting that personalized ML will fail when the variation among data points within a subject’s data is too low. This preliminary feasibility study demonstrates the potential as well as the ongoing challenges with implementing personalized models which predict highly subjective outcomes like emotion.
