Projects

The Biodynamics Laboratory at KU is helping to put KU at the forefront of research in Bioengineering, Neurophysiology and Neural Engineering. Our focus is on biomechanics and motor control with clinical applications.

Our research topics/questions are:

  • How does the brain control muscle activity and movement?
  • Balance recovery strategies from falls or disturbances on young and old adults.
  • Muscle fatigue and contraction analysis using nonlinear time series analysis and maximum likelihood method respectively.
  • The effect of Parkinson's Disease on motion control and balance.
  • The effect of manipulative techniques (as implemented by a physical therapist) on Parkinson's Disease symptoms.
  • Development of mathematical and/or ADAMS models that simulate human tasks such as: arm movement, quiet standing, balance recovery, among others.
  • Center of Pressure changes due to somatosensory and visual deficiencies.
  • Balance improvement through medical devices that enhance somatosensory feedback at the feet.

Other priorities in the lab are:

  • Encourage engineering education in middle and high school students by conducting summer camps such as Project Discovery and Explore @ Lawrence: Who Knew?
  • Improving teaching material of classes through the usage of active learning activities and/or flipped classrooms.

Current Projects

Somatosensory and Visual Deficiency on Quiet Standing for Healthy Adults

It has been observed that as people age or suffer from a neurological disease, their balance feedback systems deteriorate. Two of these feedback systems are the somatosensory and vision. This study is looking into how these deficiencies affect healthy subjects, so a better understanding of them is obtained, and therefore better treatments on people who suffer from these deficiencies can be designed. The somatosensory deficiencies are introduced in healthy subjects by placing different thicknesses of foam under their feet. The importance of this study is that if it is possible to show that foam under people's feet reflect somatosensory deficiencies, it would be possible to test new balance treatments on healthy adults standing on foam; instead of on people who have somatosensory and/or visual deficiencies. This reduces the risk on people when trying new technologies and adds a safety step.

Some of the latest results, observations or updates on the study are:

  • (Spring of 2018) Data shows that healthy subjects under different levels of somatosensory deficiency reflect a larger and more disorganized Center of Pressure (COP). It was possible to detect these differences using Machine Learning techniques; however, to get a better assessment, we need to record data from more subjects. During this machine learning assessment, linear and nonlinear measures on the COP time series were used as features.
  • (Fall of 2019) The pool of data was doubled, meaning that improved statistics can be performed!
  • (Spring 2020) Using the new pool of data, it was found that rembling, trembling, sample entropy and fuzzy entropy measures statistically differentiate between different levels of foam (i.e. somatosensory deficiency levels)

If you are interested in continuing this study, please contact Dr. Luchies.

Effect of Vibratory Excitation on the Somatosensory Feedback in Balance Assessment

Mechanical vibration under the feet has shown to have an impact on the balance of multiple populations (healthy, diabetic, Parkinson's, etc.). This makes mechanical vibration under the feet a possible inexpensive and non-invasive balance treatment. The Biodynamics Lab is working on the development and testing of a vibrating system that can achieve the already published results and others, so a better understanding can be achieved. Even though mechanical vibration under the feet has shown positive results, there are still many questions to be answered that are holding the full potential of this inexpensive and non-invasive treatment.

Some of the latest results, observations or updates on the study are:

  • (Fall of 2019) The vibrating system has been manufactured. Now that the system has passed required quality checks, it can be used on human subjects.
  • (Spring 2020) The vibrating system's ability to detect sensing thresholds was studied and validated.

If you are interested in continuing this study, please contact Dr. Luchies.

Previous Projects

A Physiology-Based Approach for Detecting Vibration Perception Threshold in the Plantar Foot
Faculty: Carl Luchies (PhD), Sara Wilson (PhD), Lisa Friis (PhD)
Student: Brett Whorley
August 2019 - May 2020

Stochastic-resonance-based vibration therapies have demonstrated the potential to improve balance in persons with somatosensory deficiencies to help prevent fall incidents. These vibrations must remain below vibration perception threshold (VPT) to be safe and effective. Several concerns exist regarding current approaches of detecting VPT, including inconsistent unit scales, limited knowledge of the physiological reliability of the methods, or potential effects they may have on standing balance. Recent assumptions that threshold detection tests have no impact on subsequent vibratory stimulations warrant further investigation. The purpose of this study was (a) to develop a new modified 4-2-1 VPT detection method (M421) based on existing approaches and underlying physiological principles, and (b) to identify potential effects the M421 may have on balance during or after threshold testing. To address the need for greater comparability between patient populations and across vibration systems, a common scale for expressing VPT was also established. Our results indicate that, among healthy adults, the M421 test does not significantly alter balance during or following threshold testing, and that a single trial conducted on both feet is comparable to separate tests of each foot. M421 demonstrates repeatable results and can be completed efficiently. Future studies will seek to further validate M421 through direct comparisons against existing methods to determine the optimal approach for detecting VPT prior to stochastic vibration interventions.

An Investigation of Rambling-Trembling Sway Trajectories with Simulated Somatosensory Deficit
Faculty: Carl Luchies (PhD), Sara Wilson (PhD), Chun-Kai Huang (PhD)
Student: Eryn Gerber
August 2019 - May 2020

Introduction: Falls in older adults are often multifactorial, but can be primarily attributed to diminished sensation abilities and lowered processing rates due to age-linked neural degeneration. A novel method for center of pressure (COP) analysis, called rambling-trembling (RM-TR) decomposition, seeks to understand the underlying feedback loops that modulate balance and has potential to provide new, valuable information about postural sway. The purpose of this study is to investigate the effects of vision and simulated somatosensory deficit on RM-TR-derived measures of the COP during quiet standing. It was hypothesized that RM and TR parameters would show similar trends across deficit severity, but with different magnitudes, and present greater sensitivity to deficit detection compared to traditioanl COP measures

Materials and Methods: Fifty-two healthy young adults (aged 22.10 ± 1.88 years) participated in the study. Participants stood on two force plates with a standardized stance with either eyes open (EO) or eyes closed (EC). Five foam thicknesses (F0-F4, representing 0, 1/8”, 1/4", 1/2”, and 1”) were used to simulate somatosensory deficit. Force and moment data were filtered using a 10Hz lowpass Butterworth filter and used to calculate COP, RM, and TR time series for the anteroposterior (AP) and mediolateral (ML) directions, as detailed by Zatsiorsky & Duarte (1999). Relative measure sensitivity was defined by: (1) the number of significant differences between foam conditions for within- and between-measure comparisons and (2) the thinnest detectable foam thickness difference. MATLAB software was used to perform three-way analyses of variance with Tukey’s HSD post hoc tests with p<0.01 to determine statistical significance.

Results and Discussion: The EC condition exhibited significantly greater changes across foam thickness as compared to EO in all measured parameters, showing EC conditions to be more sensitive to changes in simulated somatosensory deficit. Within the EC condition, COP, RM, and TR parameters all showed positive, upward trends with increasing deficit, but with variable magnitudes. AP RM is shown to have a greater magnitude of change across deficit severity in acceleration and jerk parameters. RM is significantly greater than COP and TR in the AP, and TR significantly greater than COP in the ML (Figure 1). COP was able to differentiate the smallest foam difference, 1/8”, at one instance (ML velocity), but lacks consistency in detection abilities for larger differences, such as 1/4”. Overall, the RM time-series was able to detect the greatest number of foam level comparisons and may provide a detectability robustness not found in COP or TR.

Conclusions: Further exploration of rambling-trembling is needed, but these differences highlight the potential directionality of postural control mechanisms, linking anteroposterior movement to the central nervous system and mediolateral to the peripheral nervous system.  From a clinical perspective, rambling may serve as a robust measure of somatosensory loss-induced balance changes. Future work should continue the investigation of rambling-trembling decomposition with patient populations, such as Parkinson’s or diabetic peripheral neuropathy, in order to further understand its strengths for deficit detection and research-based analysis.

Exploring the Use of Supervised Machine Learning Algorithms to Classify Simulated Balance Deficits
Faculty: Carl Luchies (PhD), Cuncong Zhong (PhD), Huazhen Fang (PhD)
Students: Logan Sidener (MS)
July 2017 - July 2018

Machine learning is a field of computer science that uses statistical techniques to give computer systems the ability to “learn” trends in data without being explicitly programmed. While this technique has been used successfully in research for the classification of gait patterns and detecting and classifying falls, this is the first study that investigated using machine learning to classify a balance deficit.

In this study, balance deficits of varying severity were simulated using healthy subjects and altering their somatosensory feedback. A group of linear and non-linear Center of Pressure (COP) measures were extracted from the data and used as the features for several supervised machine learning algorithms. The results showed that the most severe simulated balance deficit was classified with the highest accuracy and that data collected during eyes closed trials was more effective for the classification. Additionally, the most effective algorithm changed for each classification problem.

F0 stands for no somatosensory deficiency; while F4 stands for the highest somatosensory deficiency

Single Inverted Pendulum Validation for Bi-Pedal Quiet Standing in Healthy Controls and Persons with Parkinson's Disease
Faculty: Carl Luchies (PhD), Huazhen Fang (PhD), Sara Wilson (PhD)
Students: Camilo Giraldo (MS)
August 2016 - May 2017

Center of Pressure (COP) is one of the time series that has been analyzed by engineers to assess subjects' balance. The COP time series on bi-pedal quiet standing tests (or sway) is analyzed either through posturography or mechanical modeling techniques. Posturography consists on obtaining linear and nonlinear measures out of the time series; while mechanical modeling consists on designing a mathematical model that can replicate the COP time series and learn from it.

This study consisted on determining how well the inverse dynamics of the Single Inverted Pendulum (SIP) model could replicate the linear and nonlinear measures extracted from the experimental COP. In addition, the study determined if the accuracy of the SIP model was sensitive to using a linear vs. nonlinear differential equation, and to using zero vs. optimized initial conditions in the model. The results showed that if the SIP model is used with a linear differential equation and optimized initial conditions, it can represent at least 70% of the experimental COP time series based on certain measures. The measures that the SIP can represent depend on the subject type.
 

Average Percent between Trials of Linear and Nonlinear Measures within Defined Error for the COP calculated using the SIP, Linear Differential Equation and Optimized ICs

Red and bolded numbering means that the variability between trials was less than ±5% (α = 0.05)

The Application of Detrended Fluctuation Analysis and Adaptive Fractal Analysis on Center of Pressure Time Series in Parkinson's Disease

Faculty: Carl Luchies (PhD), Huazhen Fang (PhD), Suzanne Shontz (PhD)
Students: Melanie Weilert (MS)
August 2016 - March 2017

Current clinical assessments in Parkinson's disease (PS) are not sufficiently sensitive to predict fall risk. Although biomechanical postural sway measures have provided quantitative characterization towards the progression of postural instability (PI) associated with PD progression, these methods are still not sufficiently sensitive to allow for early detection of PD and fall risk. Thus, a need arises for new quantitative methods to be established which can further describe PI progression in PD.

Detrended fluctuation analysis (DFA) is a nonlinear technique that has shown promising results in sway studies. Adaptive fractal analysis (AFA) has not been tested in sway studies that investigate PI in PD patients. With that said, this study consisted on determining the best selection of parameters for these two nonlinear measures in sway studies, and then determining which one is more sensitive towards characterizing PI using the first derivative of the Center of Pressure (COPv).

The results showed that for both DFA and AFA, the recommended nmax and nmin are between N/6 and N/10 (N being the number of data points), and 4 to 6 samples respectively. In addition, it was proved that AFA produced the most clinically significant measures using smallest window sizes (Hfast).

Sample of AFA study of COPv on a subject that shows two regions in the log-log plot

Biomechanical Markers as Indicators of Postural Instability Progression in Parkinson's Disease

Faculty: Carl Luchies (PhD), Paul Cheney (PhD), Kelly Lyons (PhD), Lorin Maletsky (PhD), Sara Wilson (PhD)
Students: Annaria Barnds (PhD)
June 2010 - May 2015

This study analyzed the center of pressure (COP) during quiet standing and gait initiation of healthy controls, mild and moderate Parkinson's disease (PD) patients. The analysis consisted on extracting linear measures from both tasks, and performing a principal component analysis (PCA) on both tasks. PCA determines which measures explain for the most variability within subjects.

The goals of this study were:

  • Determine if linear measures could assess balance during quiet standing on mild and moderate PD patients.
  • Define a mathematical model using PCA and linear measures that could describe the progression of PD.
  • Determine if linear measures could assess balance during gait initiation on mild and moderate PD patients.

The results of this study were:

  • It was determined that trajectory, variation and peak measures of the COP were the measures with most significant difference between healthy controls and PD patients.
  • It was proofed that COP sway length in the anterior-posterior direction and linear measures of the COP velocity were the measures that explained the most variability between healthy and PD subjects.
  • For the gait initiation test, it was determined that velocity measures showed the most significant differences between healthy and PD subjects, suggesting that it could be a good tool to assess balance.

Quiet Standing Study: Transformed subject scores plotted on the principal component (PC) 1 versus PC 2 axes (top) and the relative contribution (coefficients) of each input parameter normalized to the most influential variable (bottom) for the eyes open PCA selection model. b. Transformed subject scores plotted on the principal component (PC) 1 versus PC 2 axes (top) and the relative contribution (coefficients) of each input parameter normalized to the most influential variable (bottom) for the eyes closed PCA selection model. A red star on the coefficient bar graphs indicates being one of the 2 most influential parameters for PC 1. Key: red x – HC subject score, green o – mild PD subject score, blue + - moderate PD subject score.

Fractal Analysis of Center of Pressure Velocity Time Series in Parkinson's Disease

Faculty: Carl Luchies (PhD), Huazhen Fang (PhD), Kelly Lyons (PhD)
Students: Joshua Harper (MS)
2015

Clinical assessment of postural instability in patients who suffer of Parkinson's disease (PD) is an unmet need. Common practices are based on linear and nonlinear measures. This study investigated the sensitivity of three measures: Absolute Average Maximal Velocity (AAMV), and Detrended Fluctuation Analysis (DFA).

These measures were extracted from the first derivative of the center of pressure (COPv) using three subject groups: healthy controls, mild and moderate PD subjects. The results showed:

  • AAMV showed significant differences between healthy controls and moderate PD patients.
  • In the DFA study, the short term scaling slope (α1) showed significant difference between healthy controls and moderate PD patients; while between PD subjects, the long term scaling slope (α2) showed significant differences.

Example construction of α1, α2 and Crossover Index in the log-log plot of fluctuation vs. scale plot

Application of Linear Stochastic Models in the Investigation of the Effects of Parkinson's Disease on the Cop Time Series

Faculty: Carl Luchies (PhD), Huazhen Fang (PhD), Sara Wilson (PhD)
Students: Chandrashekara Kaushik Gandur Balagangadhara (MS)
2015

Center of Pressure (COP) is one of the time series that has been analyzed by engineers to assess subjects' balance. The COP time series on bi-pedal quiet standing tests (or sway) is analyzed either through posturography or mechanical modeling techniques. Posturography consists on obtaining linear and nonlinear measures out of the time series; while mechanical modeling consists on designing a mathematical model that can replicate the COP time series and learn from it.

This study consisted on replicating the experimental center of pressure (COP) time series obtained from healthy controls, mild and moderate Parkinson's disease (PD) patients, using an autoregressive (AR) and autoregressive moving average (ARMA) models with a single degree of freedom (inverted pendulum). The parameters of the model: swiftness, damping, stiffness and natural frequency, were compared between subjects group.

The results of this study were:

  • The swiftness decreases as PD progresses
  • The stiffness is significantly larger in mild and moderate PD subjects than in healthy controls
  • The damping is significantly smaller in mild and moderate PD subjects than in healthy controls

Schematic or the inverted pendulum model for posture control during upright stance

Brain Control of Muscles and Movement: Synergies and Kinematics Resulting from Stimulation Applied to the Forelimb Representation of Primary Motor Cortex of Rhesus Macaques

Faculty: Carl Luchies (PhD), Paul Cheney (PhD), Terry Faddis (DE), James Stiles (PhD), Sara Wilson (PhD)
Students: Sommer Amundsen (PhD)
2013

This study consisted on determining if motion can be explained by muscle synergies. It is believed that human motion can be explained by the grouping and activation of few muscle groups. With that said, two monkeys' muscles were wired to determine if the motion that they were performing could be explained by some muscle synergy.

The main results of this study was that muscle synergy is present in arm motion, since 2-3 muscle synergies could explain 90% of the motion that was being analyzed. However, it was not possible to determine how the CNS grouped the muscles to perform the motion that was analyzed.

Diagram of monkey engaged in three-location task at: (1) home plate lever, (2) food-well, and (3) mouth. Approximate marker cluster set-up also pictured.

The Effect of Moderate Parkinson's Disease on the Biomechanics of Compensatory Backwards Stepping

Faculty: Carl Luchies (PhD), Terry Faddis (DE), Kelly Lyons (PhD), Jonathan Mahnken (PhD), Sara Wilson (PhD)
Students: Molly McVey (PhD)
January 2008 - August 2012

Postural assessment on patients who suffer from Parkinson's disease (PD) is an unmet need. One of the most common balance tests performed by doctors is the pull-test which consist on pulling back the patients by the shoulders, and then observing how many steps are required to recover balance and how it was done. Even though this test explains the balance of the subjects well, it is very subjective.

Therefore, this study consisted on analyzing the balance recovery of subjects with PD using linear measures on their center of pressure. This analyzed concluded:

  • Patients with PD took more steps to recover balance, and took longer to shift their weight from one foot to another foot.
  • Patients with PD make more anticipatory postural adjustments before taking a step, meaning that their lift-off after being pulled is longer when compared to healthy subjects.
  • Patients with PD take significantly longer to recover balance after being pulled than healthy controls.

With these observations made, it is possible to propose the standardization of the pull-test. This means that there is a possibility of obtaining the same diagnosis between doctors, which reduces the variability of therapies.

Plot illustrates the calculation of temporal parameters. RT= Reaction Time, WST= Weight Shift Time. Solid black line is load cell trace, solid blue line is EMG from stepping foot TA muscle. Solid red line is vertical force under the step foot. RT taken as time between muscle onset and disturbance onset, WST taken as time between force plate liftoff and muscle onset


For more information about older projects, please visit the Publications tab.