Introduction

Keeping the body upright and stable, even under static conditions, is a complex task. From a biomechanical perspective, maintaining static balance can be conceptualized as the interplay between an individual’s center of pressure (CoP) and center of mass (CoM). CoM is the location in the body where the torques of the mass elements of the body sum to zero. CoP is the point where the total sum of a pressure field acts on a body, causing a force to act through that point. When the horizontal location of the CoM and CoP are not aligned, a torque about the CoM is produced. If the torque is not counterbalanced by another torque, the person will fall (or be forced to take a step). To keep from falling, muscles at the hip and ankle are used to adjust the CoP and produce another torque, thereby keeping the body upright and stable. The is how the body maintains balance.

The CoP is never perfectly underneath the CoM for more than a brief moment, and so adjustments are constantly being made by the body to keep the CoP and CoM closely in line. The body receives integrated feedback from the visual, vestibular, and proprioceptive sensory systems to make the necessary adjustments to the CoP at the right time and of the appropriate magnitude. Each of the three sensory systems play a role in an individual’s ability to maintain balance. How large of an influence each of the systems independently has on balance, however, is not known. The purpose of this study is to investigate how manipulating these sensory systems independently affects static balance. We hypothesize that the proprioceptive system will have the largest effect on balance for all measures of balance, and that the visual system will have the smallest effect.

Methods

Force, torque, and CoP data was collected from five participants using an AMTI Force Platform sampled at 1000 Hz for 30 s. Subjects performed one trial each of four conditions: all sensory systems in tact (EO), impaired visual feedback (EC), impaired visual and vestibular feedback (ECHT), and impaired visual and proprioceptive feedback (ECFoam). For the EC condition, subjects stood with their eyes closed and their arms held across their body. For the ECHT condition, subjects tilted their head completely forward, in addition to the constraints of the EC condition. For the ECFoam condition, subjects stood on a foam pad (height = 6.35 cm) that was placed on the force platform, in addition to the constraints of the EC condition. Data was also collected for 14 other participants who each performed one trial of the EC condition. The raw CoP data was filtered using a 4th order low-pass recursive butterworth filter. Instantaneous CoP velocity in the anteroposterior (AP) mediolateral (ML) direction was calculated, and the distance formula was used to calculate overall mean CoP velocity. Sway path length (SPL) was calculated by summing the instantaneous CoP velocities (of the combined ML and AP directions).

Data Analysis

library(purrr)
library(dplyr)
library(readr)
library(ggplot2)
library(signal)
# read in csv files
files <- list.files("../data", pattern = ".csv", full.names = TRUE)
data <- map(files, read_csv, skip = 3, col_select = 12:20)
names(data) <- basename(files)

# first row doesn't contain data, all columns are numeric
data <- map(data, ~map_dfc(.x[-1, ], as.numeric))
# Adjusting center location
data <- map(data, function(x) {
  x %>%
    as_tibble() %>% 
    mutate(
      Cx_adjust = (-My) / (Fz), 
      Cy_adjust = (Mx / (Fz))
    )
})
# need to adjust foam condition separately 
foam_cond <- stringr::str_detect(names(data), "ECFoam")
data <- map_at(data, which(foam_cond), function(x) {
  x %>% 
    mutate(
      Cx_adjust = ((-63.5 * Fx) - My) / (Fz),
      Cy_adjust = ((-63.5 * Fy) + Mx) / (Fz)
    )
})
# implement butterworth filter; srate was 1000Hz for this experiment
bf <- butter(type = "low", n = 2, W = 8/(1000/2))

# ideal cutoff for COP is around 8Hz for x and y 
# calculate inst. velocity 
data <- map(data, function(x) {
  x %>%
    mutate(
      filt_COP_x = filtfilt(bf, x = Cx_adjust),
      filt_COP_y = filtfilt(bf, x = Cy_adjust),
      x_vel = filt_COP_x - lag(filt_COP_x),
      y_vel = filt_COP_y - lag(filt_COP_y),
      cop_xy_vel = sqrt((x_vel^2) + (y_vel^2))
    )
})
p_peter <- data[["Peter.csv"]] %>% 
  ggplot(aes(x = ((filt_COP_y - mean(filt_COP_y)) * -1), 
             y = (filt_COP_x - mean(filt_COP_x)))) +
  geom_path() + 
  theme_bw() +
  labs(title = "Force Platform Stabilogram- Peter") + 
  ylab("COP in AP direction (mm)") +
  xlab("COP in ML direction (mm)") + 
  ylim(-24, 27) + 
  xlim(-15, 13) + 
  theme(
    panel.grid.minor.x = element_blank(),
    panel.grid.minor.y = element_blank()
  )

p_jason <- data[["Jason.csv"]] %>% 
  ggplot(aes(x = ((filt_COP_y - mean(filt_COP_y)) * -1), 
             y = (filt_COP_x - mean(filt_COP_x)))) +
  geom_path() + 
  theme_bw() +
  labs(title = "Force Platform Stabilogram- Jason") + 
  ylab("COP in AP direction (mm)") +
  xlab("COP in ML direction (mm)") + 
  ylim(-24, 27) + 
  xlim(-15, 13) + 
  theme(
    panel.grid.minor.x = element_blank(),
    panel.grid.minor.y = element_blank()
  )

gridExtra::grid.arrange(p_peter, p_jason, nrow = 1)

conds <- c("Alise_", "Anupriya_", "Brady_", "Mindie_", "Rhiannon_")
all_conds <- map(conds, stringr::str_which, string = names(data))

# subset to condition file for each person
sub <- function(data, x, pattern) {
  data[x][stringr::str_detect(names(data[x]), pattern)][[1]]
}

cond_diff <- function(df1, df2, cond_name) {
  tibble(
    cop_mean = mean(df1$cop_xy_vel, na.rm = TRUE) - mean(df2$cop_xy_vel, na.rm = TRUE),
    sd_cx = sd(df1$Cx_adjust) - sd(df2$Cx_adjust),
    sd_cy = sd(df1$Cy_adjust) - sd(df2$Cy_adjust),
    sway_path = sum(df1$cop_xy_vel, na.rm = TRUE) - sum(df2$cop_xy_vel, na.rm = TRUE),
    Condition = cond_name
  )
}

df_summary <- map_dfr(all_conds, .id = "index", function(x) {
  rbind(
   cond_diff(sub(data, x, "EC"), sub(data, x, "EO"), "Eyes Closed (EC)"),
   cond_diff(sub(data, x, "ECHT"), sub(data, x, "EC"), "Head Tilt (ECHT)"),
   cond_diff(sub(data, x, "ECFoam"), sub(data, x, "EC"), "Foam (ECFoam)") 
  )
})
# Mean of means
df_means <- df_summary %>% 
  group_by(Condition) %>% 
  summarise(
    mean_cop = mean(as.numeric(cop_mean)),
    sd_cx = mean(as.numeric(sd_cx)),
    sd_cy = mean(as.numeric(sd_cy)),
    sway_path = mean(as.numeric(sway_path))
 ) %>%
  arrange(sway_path)
df_means %>%
  ggplot(aes(x = Condition, y = (mean_cop * 100))) +
  geom_bar(stat = "identity", col = "black", fill = "darkred") +
  labs(title = "Average ΔChange in Mean CoP Velocity") +
  ylab("Change in Mean CoP Velocity (cm/s)") +
  theme_bw() +
  theme(
    panel.grid.minor.x = element_blank(),
    panel.grid.major.x = element_blank(),
    text = element_text(family = "Times", size = 12)
  )

df_means %>% 
  ggplot(aes(x = Condition, y = sd_cx)) + 
  geom_bar(stat = "identity", col = "black", fill = "darkred") +
  geom_abline(intercept = 0, slope = 0) +
  labs(title = "Average ΔChange in Standard Deviation") + 
  ylab("SD Change in CoP ML direction (mm)") +
  theme_bw() + 
  theme(
    panel.grid.minor.x = element_blank(),
    panel.grid.major.x = element_blank(),
    text = element_text(family = "Times", size = 12)
  )

df_means %>% 
  ggplot(aes(x = Condition, y = sway_path)) + 
  geom_bar(stat = "identity", col = "black", fill = "darkred") +
  labs(title = "Average ΔChange in Sway Path Length") + 
  ylab("Change in Sway Path Length (mm)") +
  theme_bw() + 
  theme(
    panel.grid.minor.x = element_blank(),
    panel.grid.major.x = element_blank(),
    text = element_text(family = "Times")
  )

Results

Average change in: mean CoP velocity, ML CoP deviation (SD) of amplitude (SD-A), and SPL were calculated for the five subjects for the EC, ECHT, and ECFoam trials. EC trial was subtracted from EO trial, ECHT trial was subtracted from EC trial, and ECFoam trial was subtracted from EC trail for each of the three measures for each subject. The ECFoam condition showed the greatest change on all three measures, while the EC condition showed the smallest change in mean CoP velocity and SPL. The ECHT condition showed the smallest change in ML CoP SD-A. Surprisingly, the EC trial lead to a decrease in ML CoP SD-A. Two stabilogram plots of the EC trial were made to visually demonstrate CoP in the ML and AP directions over a 30 s trial.

Discussion

It was hypothesized that the ECFoam and EC condition would produce the largest and smallest changes in the three measures of balance, respectively. The rationale for this hypothesis was based on the assumption that there could (and likely would be) an interaction when manipulating multiple sensory systems. In other words, even after controlling for the EC condition by subtracting this value from the ECFoam and ECHT conditions, the interaction between the effect of EC and head tilt or foam would cause a larger change compared to EC alone. ECFoam was hypothesized to produce a larger change than ECHT because prior research has shown that the vestibular system doesn’t seem to play as large of a role in maintaining balance as proprioception [1].

As hypothesized, the ECFoam condition had the largest change in all three measures of balance. The EC condition had the smallest change in two measures of balance, mean CoP velocity and SPL. It was the ECHT condition, however, that technically led to the smallest change in ML CoP SD-A; although “change” doesn’t tell the full story in this case, because the EC condition actually caused a decrease in SD-A. As previously mentioned, these results are likely due to an interaction between the manipulation of multiple sensory systems.

A study by Adamo et al. found a significant two way interaction between the manipulation of surface and vision. He noted that the mean difference in sway velocity between EO and EC on a foam surface was about 10.5 fold greater than the mean difference in sway velocity between EO and EC when standing on a firm surface [2]. This suggests that even after subtracting off the EC difference in our data, the residual is not simply due to the foam or head tilt manipulations alone, but rather an interaction between EC and the respective condition.

A study by Isableu found that the alteration of ankle proprioception had a greater destabilizing effect in subjects exhibiting the smallest CoP displacements when standing in a normal proprioception conditon [3]. Another way to say this is that the less the subjects swayed in the firm condition, the more likely they swayed on the foam condition. If this were the case, then our current study with a sample of young, healthy, mostly female (n = 4) fitness enthusiasts could partially explain the large effect of the ECFoam condition; the participants may have superior balance with very small CoP displacements under a normal, EO condition, thereby displaying greater sway velocity than what would typically be observed under an altered proprioceptive condition. However, a correlation was not performed for our study, so it is unknown if this was the case or not.

Moghadam and colleagues [4] found that the ECFoam condition compared to the EC conditon increased the mean CoP velocity and ML CoP SD-A by 1.13 cm/s (75% increase) and 3.6 mm (180% increase). The changes coincide with the current study, although not as drastic. Again, one explanation for the drastic changes in balance measures for the ECFoam condition could be due to how well-balanced the students are in the current study during non-proprioceptive-altering condtions.

The same study found that the EC condition compared to the EO condition increased the mean CoP velocity and ML CoP SD-A by 0.12 cm/s (7% increase) and 0.3 mm (16% increase). The change in mean CoP velocity is relatively in line with our findings. The ML CoP SD-A, however, did not decrease like in our study. Moghadam also performed retests for each of the conditions performed and reported the coefficent of variation (CV) for the nine balance measures that were calculated. ML CoP SD-A had the highest reported CV, 20.24%. The high degree of variation within this measure could be one explanation for the observed results.

The literature suggests the vestibular system plays less of a role than what was observed in the current study. Hansson and colleagues found a 0.11 cm/s increase in mean CoP velocity when comparing an impaired vestibular system condition with an EC condition. However, Hansson had the subjects perform the impaired vestibular condition with an extended neck (among other vestibular conditions), and so this may not be an appropriate comparison. Wu [6] had similar findings to Hansson’s, concluding that the vestibular system plays less of a role than the other two systems. Inconsistencies with our findings and the literature is not very surprising, though, given that some researchers suggest the existence of preferential sensorimotor tactics in response to perturbations [7].

Limitations

There are several potential limitations of this study worth noting. 1) The conditions were not performed in a random order by the subjects, which could potentially create a systematic effect. This could also lead to a learning effect across the trials. Some researchers even suggest not performing multiple conditions or trials in the same day because there can even be a learning effect for eyes open standing balance trials. 2) The small sample may not reflect what would typically be seen in larger samples. Additionally, the sample in the present study may be quite different from other young, healthy adult samples, given the exercise science background and general interest in health and fitness of the present subjects. 3) The trial length may not have been long enough. Some studies have shown that the G-coefficients can change drastically for certain balance measures depending on the trial length [5].

Conclusion

Balance is a complex task involving the visual, vestibular, and proprioceptive sensory systems. The purpose of this study was to investigate how manipulating these systems affects three measures of balance: mean CoP velocity, ML CoP SD-A, and SPL. It was found that the ECFoam condition had the largest changes in all three measures compared to EO, and the EC condition had the smallest changes for mean CoP velocity and SPL compared to EO, and there was a decrease in the ML CoP SD-A compared to EO, but there is generally large variation within this measure of balance. These findings could provide insight to creating effective rehabilitation programs that involve manipulating the sensory systems to improve balance.

References

  1. Hansson EE, et al. Acta oto-laryngologica, 130.12, 1358-1363, 2010.
  2. Adamo DE, et. al. J of Vest Res, 23.1, 33-40, 2013.
  3. Isableu B, et al. Exp Brain Res, 174.4, 763-768, 2006.
  4. Moghadam M, et al. Gait and Posture, 33.4, 651-655, 2011.
  5. Doyle RJ, et al. Gait and Pos, 25.2, 166-171, 2007.
  6. Wu G, et al. Brain Res Bull, 74, 69-365, 2006.
  7. Kluzik J, et al. Exp Brain Res, 162, 474-489, 2005.