Publications by Year: 2025

Background:Transvaginal insertion of polypropylene mesh was extensively used in surgical procedures to treat pelvic organ prolapse (POP) due to its cost-efficiency and durability. However, studies have reported a high rate of complications, including mesh exposure through the vaginal wall. Developing predictive models via supervised machine learning holds promise in identifying risk factors associated with such complications, thereby facilitating better informed surgical decisions. Previous studies have demonstrated the efficacy of anticipating medical outcomes by employing supervised machine learning approaches that integrate patient health care data with laboratory findings. However, such an approach has not been adopted within the realm of POP mesh surgery.

Objective:We examined the efficacy of supervised machine learning to predict mesh exposure following transvaginal POP surgery using 3 different datasets: (1) patient medical record data, (2) biomaterial-induced blood cytokine levels, and (3) the integration of both.

Methods:Blood samples and medical record data were collected from 20 female patients who had prior surgical intervention for POP using transvaginal polypropylene mesh. Of these subjects, 10 had experienced mesh exposure through the vaginal wall following surgery, and 10 had not. Standardized medical record data, including vital signs, previous diagnoses, and social history, were acquired from patient records. In addition, cytokine levels in patient blood samples incubated with sterile polypropylene mesh were measured via multiplex assay. Datasets were created with patient medical record data alone, blood cytokine levels alone, and the integration of both data. The data were split into 70% and 30% for training and testing sets, respectively, for machine learning models that predicted the presence or absence of postsurgical mesh exposure.

Results:Upon training the models with patient medical record data, systolic blood pressure, pulse pressure, and a history of alcohol usage emerged as the most significant factors for predicting mesh exposure. Conversely, when the models were trained solely on blood cytokine levels, interleukin (IL)-1β and IL-12 p40 stood out as the most influential cytokines in predicting mesh exposure. Using the combined dataset, new factors emerged as the primary predictors of mesh exposure: IL-8, tumor necrosis factor-α, and the presence of hemorrhoids. Remarkably, models trained on the integrated dataset demonstrated superior predictive capabilities with a prediction accuracy as high as 94%, surpassing the predictive performance of individual datasets.

Conclusions:Supervised machine learning models demonstrated improved prediction accuracy when trained using a composite dataset that combined patient medical record data and biomaterial-induced blood cytokine levels, surpassing the performance of models trained with either dataset in isolation. This result underscores the advantage of integrating health care data with blood biomarkers, presenting a promising avenue for predicting surgical outcomes in not only POP mesh procedures but also other surgeries involving biomaterials. Such an approach has the potential to enhance informed decision-making for both patients and surgeons, ultimately elevating the standard of patient care.

JMIR Form Res 2025;9:e59631

We propose a robust framework for quantitatively comparing model-predicted and experimentally measured strain fields in the human brain during harmonic skull motion. Traumatic brain injuries (TBIs) are typically caused by skull impact or acceleration, but how skull motion leads to brain deformation and consequent neural injury remains unclear and comparison of model predictions to experimental data remains limited. Magnetic resonance elastography (MRE) provides high-resolution, full-field measurements of dynamic brain deformation induced by harmonic skull motion. In the proposed framework, full-field strain measurements from human brain MRE in vivo are compared to simulated strain fields from models with similar harmonic loading. To enable comparison, the model geometry and subject anatomy, and subsequently, the predicted and measured strain fields are nonlinearly registered to the same standard brain atlas. Strain field correlations (Cv), both global (over the brain volume) and local (over smaller sub-volumes), are then computed from the inner product of the complex-valued strain tensors from model and experiment at each voxel. To demonstrate our approach, we compare strain fields from MRE in six human subjects to predictions from two previously developed models. Notably, global Cv values are higher when comparing strain fields from different subjects (Cv~0.6–0.7) than when comparing strain fields from either of the two models to strain fields in any subject. The proposed framework provides a quantitative method to assess similarity (and to identify discrepancies) between model predictions and experimental measurements of brain deformation and thus can aid in the development and evaluation of improved models of brain biomechanics.

Purpose

To determine how the biomechanical vulnerability of the human brain is affected by features of individual anatomy and loading.

Methods

To identify the features that contribute most to brain vulnerability, we imparted mild harmonic acceleration to the head and measured the resulting brain motion and deformation using magnetic resonance elastography (MRE). Oscillatory motion was imparted to the heads of adult participants using a lateral actuator (n = 24) or occipital actuator (n = 24) at 20 Hz, 30 Hz, and 50 Hz. Displacement vector fields and strain tensor fields in the brain were obtained from MRE measurements. Anatomical images, as well as displacement and strain fields from each participant were rigidly and deformably aligned to a common atlas (MNI-152). Vulnerability of the brain to deformation was quantified by the ratio of strain energy (SE) to kinetic energy (KE) for each participant. Similarity of deformation patterns between participants was quantified using strain field correlation (C_V). Linear regression models were used to identify the effect of similarity of brain size, shape, and age, as well as similarity of loading, on C_V.

Results

The SE/KE ratio decreased with frequency and was larger for participants undergoing lateral, rather than occipital, actuation. Head rotation about the inferior–superior axis was correlated with larger SE/KE ratio. Strain field correlations were primarily affected by the similarity of rigid-body motion.

Conclusion

The motion applied to the skull is the most important factor in determining both the vulnerability of the brain to deformation and the similarity between strain fields in different individuals.

Traumatic brain injuries (TBIs) occur from rapid head motion that results in brain deformation. Computational models are typically used to estimate brain deformation to predict risk of injury and evaluate the effectiveness of safety countermeasures. The accuracy of these models relies on validation to experimental brain deformation data. In this study, we create the first group-average biomechanical responses of the brain, including structure, material properties, and deformation response, by age and sex from 157 subjects. Subjects were sorted intro three age groups—young, mid-age, and older—and by sex to create group-average neuroanatomy, material properties, and brain deformation response to non-injurious loading using structural and specialized magnetic resonance imaging data. Computational models were also built using the group-average geometry and material properties for each of the six groups. The material properties did not depend on sex, but showed a decrease in shear stiffness in the older adult group. The brain deformation response also showed differences in the distribution of strain and a decrease in the magnitude of maximum strain in the older adult group. The computational models were simulated using the same non-injurious loading conditions as the subject data. While the models’ strain response showed differences among the models, there were no clear relationships with age. Further studies, both modeling and experimental, with more data from subjects in each age group, are needed to clarify the mechanisms underlying the observed changes in strain response with age, and for computational models to better match the trends observed across the group-average responses.

Objective: Computational models of the brain are typically validated using individual subjects from datasets of brain motion, but a comparison to an individual subject does not consider the biomechanical variation that naturally exists in the population. When data from multiple subjects is available, biomechanical corridors are constructed for the assessment of model biofidelity. However, a robust set of corridors for brain's biomechanical response due to applied head kinematics does not exist for model validation. The aim of this study was to create corridors based on a dataset of in situ brain displacement that included six specimens tested under a set of twelve loading conditions.

Methods: There were three main factors that complicated this task, including variation in head kinematics, differences in the initial position of the sensors, and the clustering of spatially scattered data. We employed various numerical and statistical methods to account for these experimental variations, with optimization and validation of the techniques conducted using the existing in situ dataset and a computational brain model.

Results: Corridors were constructed using average and standard deviation of the specimen responses in the dataset for 24 discrete locations within the brain. Peak displacement showed a variance of less than 30% for most brain sensor locations.

Conclusion: The corridors will serve as a better validation tool for assessing the biofidelity of computational brain models and will help understand inter-subject variability in brain biomechanics.

Elevation in left ventricular (LV) myocardial stiffness is a key remodeling-mediated change that underlies the development and progression of heart failure (HF). Despite the potential diagnostic value of quantifying this deterministic change, there is a lack of enabling techniques that can be readily incorporated into current clinical practice. To address this unmet clinical need, we propose a simple protocol for processing routine echocardiographic imaging data to provide an index of left ventricular myocardial stiffness, with protocol specification for patients at risk for heart failure with preserved ejection fraction. We demonstrate our protocol in both a preclinical and clinical setting, with representative findings that suggest sensitivity and translational feasibility of obtained estimates.

Strategic delivery of hydrogels to the newly formed myocardial infarction (MI) is an area of active investigation and offers high target specificity for releasing a small molecule therapeutic payload. This study examined the effects of delayed post-MI delivery (pigs, 3 days post-MI) of a shear-thinning hydrogel which encapsulated and released a small molecule matrix metalloproteinase inhibitor. The results demonstrated the feasibility and efficacy of targeted delivery of a shear-thinning injectable hydrogel containing a small molecule matrix metalloproteinase inhibitor to attenuate post-MI remodeling.

Endothelial dysfunction, defined as a reduction in the bioavailability of nitric oxide (NO), is a risk factor for the occurrence and progression of various vascular diseases. This study investigates the effect of endothelial dysfunction on age-related changes in aortic extracellular matrix (ECM) microstructure and the relationship between microstructural adaptation and the mechanical response. Here, we used groups of NOS3 knockout (KO), NOS3 heterozygotes (Het), and wild-type (WT) B6 mice (controls) to study changes in hemodynamic parameters, collagen fiber organization, and both active and passive aortic mechanics using biaxial pressure myography over a time course from 1.5 to 12 mo. Our results show that homeostatic levels of passive circumferential stress and stretch were preserved in KO mice by remodeling adventitial collagen fibers toward a more predominantly circumferential direction with age, rather than by increased fibrosis, in response to hypertension induced by endothelial dysfunction. However, passive aortic stiffness in KO mice was significantly increased owing to geometrical changes, including significant increases in wall thickness and decreases in inner diameter, as well as by ECM microstructural reorganization, during this maladaptive vascular remodeling. Furthermore, long-term NO deficiency significantly increased smooth muscle cell (SMC) contractility initially, but this effect was attenuated with age. These findings improve our understanding of microstructural and mechanical changes during the maladaptive vascular remodeling process, demonstrating a role for adventitial collagen fiber re-orientation in the response to hypertension

We revisit the thermodynamic analysis of an isothermal ideal gas mixture enclosed within a cylinder and separated from the surrounding atmosphere by a movable and frictionless piston. When equilibrium conditions based on the chemical potentials of one or more species in the mixture are not satisfied at all times, which occurs for example for a chemical reaction with finite and non-zero reaction rates in the forward and reverse directions and for mass transfer of one species across a permeable membrane occurring at a finite and non-zero rate, an irreversibility is necessarily introduced into the system with a resulting increase in the entropy of the Universe. Consequently, when the piston is set in motion, it cannot oscillate indefinitely. The piston must again come to rest despite there not being any mechanical dissipative mechanisms, i.e. friction or viscous dissipation, nor a thermal dissipative mechanism, i.e. irreversible heat transfer, operating within the system. Only when the system is reversible, such that the entropy of the Universe remains constant at all times, will the piston oscillate indefinitely. 'Chemical damping,' or an irreversibility arising from nonequilibrium conditions on the chemical potential, provides another dissipative mechanism that has not yet been analyzed before.

In general, the present disclosure is directed to systems and methods of evaluating a subject's risk of one or more complications associated with pelvic organ prolapse surgery. The method comprising: obtaining, by a computing system comprising one or more computing devices, sample data associated with the subject; obtaining, by a computing system comprising one or more computing devices, medical record data associated with the subject; inputting, by the computing system, the sample data into a machine-learned surgical model; receiving, by the computing system as an output of the machine-learned surgical model, one or more predictions of post-surgical complications of mesh exposure through a vaginal wall associated with the subject; and performing a pelvic organ prolapse repair surgery on the subject.