Limits...
Improving motor activity assessment in depression: which sensor placement, analytic strategy and diurnal time frame are most powerful in distinguishing patients from controls and monitoring treatment effects.

Reichert M, Lutz A, Deuschle M, Gilles M, Hill H, Limberger MF, Ebner-Priemer UW - PLoS ONE (2015)

Bottom Line: Accordingly, both amplitude (d=1.16) and frequency (d=1.04) showed alterations, indicating reduced and decelerated motor activity.Differences between MD and HC in gestures (d=0.97) and walking (d=1.53) were found by data analysis from the wrist sensor.Sample size was small, but sufficient for the given effect sizes.

View Article: PubMed Central - PubMed

Affiliation: Department of Sports and Sports Science, Karlsruhe Institute of Technology (KIT), Baden-Wuerttemberg, Germany; Department of Psychiatry and Psychotherapy, Central Institute of Mental Health, Mannheim, Baden-Wuerttemberg, Germany.

ABSTRACT

Background: Abnormalities in motor activity represent a central feature in major depressive disorder. However, measurement issues are poorly understood, limiting the use of objective measurement of motor activity for diagnostics and treatment monitoring.

Methods: To improve measurement issues, especially sensor placement, analytic strategies and diurnal effects, we assessed motor activity in depressed patients at the beginning (MD; n=27) and after anti-depressive treatment (MD-post; n=18) as well as in healthy controls (HC; n=16) using wrist- and chest-worn accelerometers. We performed multiple analyses regarding sensor placements, extracted features, diurnal variation, motion patterns and posture to clarify which parameters are most powerful in distinguishing patients from controls and monitoring treatment effects.

Results: Whereas most feature-placement combinations revealed significant differences between groups, acceleration (wrist) distinguished MD from HC (d=1.39) best. Frequency (vertical axis chest) additionally differentiated groups in a logistic regression model (R2=0.54). Accordingly, both amplitude (d=1.16) and frequency (d=1.04) showed alterations, indicating reduced and decelerated motor activity. Differences between MD and HC in gestures (d=0.97) and walking (d=1.53) were found by data analysis from the wrist sensor. Comparison of motor activity at the beginning and after MD-treatment largely confirms our findings.

Limitations: Sample size was small, but sufficient for the given effect sizes. Comparison of depressed in-patients with non-hospitalized controls might have limited motor activity differences between groups.

Conclusions: Measurement of wrist-acceleration can be recommended as a basic technique to capture motor activity in depressed patients as it records whole body movement and gestures. Detailed analyses showed differences in amplitude and frequency denoting that depressed patients walked less and slower.

No MeSH data available.


Related in: MedlinePlus

Spectral analysis of acceleration (frequency and amplitude) of MD patients at the beginning of treatment (MD) and healthy controls (HC) captured with the vertical axes of the chest-accelerometer.Mean amplitude values (in au1; grey and black horizontal resp.) as well as the group centroid frequency (in Hz; grey and black vertical line resp.) for the whole spectrum of walking (1–3Hz) are depicted in Fig 1a, 1b, respectively. 1arbitrary unit: values are based on g but attenuated due to the computation of the spectrum and the nonstationarity of the data.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4401688&req=5

pone.0124231.g001: Spectral analysis of acceleration (frequency and amplitude) of MD patients at the beginning of treatment (MD) and healthy controls (HC) captured with the vertical axes of the chest-accelerometer.Mean amplitude values (in au1; grey and black horizontal resp.) as well as the group centroid frequency (in Hz; grey and black vertical line resp.) for the whole spectrum of walking (1–3Hz) are depicted in Fig 1a, 1b, respectively. 1arbitrary unit: values are based on g but attenuated due to the computation of the spectrum and the nonstationarity of the data.

Mentions: Second, mean daily motor activity (acceleration values measured in milli-g) over the 24 h period as well as the hourly means of motor activity were calculated using SPSS version 21. Third, raw activity data was segmented into 60-s periods and subjected to Fourier-based spectral analysis using Vision Analyser version 1.04 software (Brain Products, Gilching, Germany, http://www.brainproducts.com). For further statistical analysis, the mean amplitude (see Figs 1a and 2a) and the centroid frequency (dividing the area of the spectrum into two equal parts; see Figs 1b and 2b) were computed within the frequency bands from 1 to 3 Hz. The frequency range 1–3 Hz was used as it covers the main motor activity within our groups (see Figs 1 and 3) and enabled us to focus gross motor activity, like walking and running. The amplitude of the spectral analysis is an arbitrary unit (au): the values are based on g but attenuated due to the computation of the spectrum and the nonstationarity of the data.


Improving motor activity assessment in depression: which sensor placement, analytic strategy and diurnal time frame are most powerful in distinguishing patients from controls and monitoring treatment effects.

Reichert M, Lutz A, Deuschle M, Gilles M, Hill H, Limberger MF, Ebner-Priemer UW - PLoS ONE (2015)

Spectral analysis of acceleration (frequency and amplitude) of MD patients at the beginning of treatment (MD) and healthy controls (HC) captured with the vertical axes of the chest-accelerometer.Mean amplitude values (in au1; grey and black horizontal resp.) as well as the group centroid frequency (in Hz; grey and black vertical line resp.) for the whole spectrum of walking (1–3Hz) are depicted in Fig 1a, 1b, respectively. 1arbitrary unit: values are based on g but attenuated due to the computation of the spectrum and the nonstationarity of the data.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4401688&req=5

pone.0124231.g001: Spectral analysis of acceleration (frequency and amplitude) of MD patients at the beginning of treatment (MD) and healthy controls (HC) captured with the vertical axes of the chest-accelerometer.Mean amplitude values (in au1; grey and black horizontal resp.) as well as the group centroid frequency (in Hz; grey and black vertical line resp.) for the whole spectrum of walking (1–3Hz) are depicted in Fig 1a, 1b, respectively. 1arbitrary unit: values are based on g but attenuated due to the computation of the spectrum and the nonstationarity of the data.
Mentions: Second, mean daily motor activity (acceleration values measured in milli-g) over the 24 h period as well as the hourly means of motor activity were calculated using SPSS version 21. Third, raw activity data was segmented into 60-s periods and subjected to Fourier-based spectral analysis using Vision Analyser version 1.04 software (Brain Products, Gilching, Germany, http://www.brainproducts.com). For further statistical analysis, the mean amplitude (see Figs 1a and 2a) and the centroid frequency (dividing the area of the spectrum into two equal parts; see Figs 1b and 2b) were computed within the frequency bands from 1 to 3 Hz. The frequency range 1–3 Hz was used as it covers the main motor activity within our groups (see Figs 1 and 3) and enabled us to focus gross motor activity, like walking and running. The amplitude of the spectral analysis is an arbitrary unit (au): the values are based on g but attenuated due to the computation of the spectrum and the nonstationarity of the data.

Bottom Line: Accordingly, both amplitude (d=1.16) and frequency (d=1.04) showed alterations, indicating reduced and decelerated motor activity.Differences between MD and HC in gestures (d=0.97) and walking (d=1.53) were found by data analysis from the wrist sensor.Sample size was small, but sufficient for the given effect sizes.

View Article: PubMed Central - PubMed

Affiliation: Department of Sports and Sports Science, Karlsruhe Institute of Technology (KIT), Baden-Wuerttemberg, Germany; Department of Psychiatry and Psychotherapy, Central Institute of Mental Health, Mannheim, Baden-Wuerttemberg, Germany.

ABSTRACT

Background: Abnormalities in motor activity represent a central feature in major depressive disorder. However, measurement issues are poorly understood, limiting the use of objective measurement of motor activity for diagnostics and treatment monitoring.

Methods: To improve measurement issues, especially sensor placement, analytic strategies and diurnal effects, we assessed motor activity in depressed patients at the beginning (MD; n=27) and after anti-depressive treatment (MD-post; n=18) as well as in healthy controls (HC; n=16) using wrist- and chest-worn accelerometers. We performed multiple analyses regarding sensor placements, extracted features, diurnal variation, motion patterns and posture to clarify which parameters are most powerful in distinguishing patients from controls and monitoring treatment effects.

Results: Whereas most feature-placement combinations revealed significant differences between groups, acceleration (wrist) distinguished MD from HC (d=1.39) best. Frequency (vertical axis chest) additionally differentiated groups in a logistic regression model (R2=0.54). Accordingly, both amplitude (d=1.16) and frequency (d=1.04) showed alterations, indicating reduced and decelerated motor activity. Differences between MD and HC in gestures (d=0.97) and walking (d=1.53) were found by data analysis from the wrist sensor. Comparison of motor activity at the beginning and after MD-treatment largely confirms our findings.

Limitations: Sample size was small, but sufficient for the given effect sizes. Comparison of depressed in-patients with non-hospitalized controls might have limited motor activity differences between groups.

Conclusions: Measurement of wrist-acceleration can be recommended as a basic technique to capture motor activity in depressed patients as it records whole body movement and gestures. Detailed analyses showed differences in amplitude and frequency denoting that depressed patients walked less and slower.

No MeSH data available.


Related in: MedlinePlus