Comparison of Force Control Capabilities (FCCs) in ...edt.postech.ac.kr/homepage_data/publication_proceedings...Comparison of Force Control Capabilities (FCCs) in Patients with Motor

2013. 6. 26

Comparison of Force Control Capabilities (FCCs)in Patients with Motor Intentional Disorders (MIDs)

and Normal Controls

Hyunji Park1, Baekhee Lee1, Kihyo Jung, Ph.D.2, Byunghwa Lee, Ph.D.3, Duk L. Na, MD, Ph.D.3, Heecheon You, Ph.D.1

1Department of Industrial and Management Engineering, POSTECH2Department of Industrial Engineering, Ulsan University3Department of Neurology, Samsung Medical Center

0

2

4

6

8

10

12

0 20 40 60 80 100

Force (N)

Time (ms)

9.8NIT DT ME TT

Signal Signal

0

2

4

6

8

10

12

0 20 40 60 80 100

Force (N)

Time (ms)

9.8NIT DT ME TT

Signal Signal

AGENDA

Introduction

• Background

• Objective of the Study

Research Protocol

Results

1) Effects of Age, Gender, and Hand on normal FCCs

2) Comparison of FCCs bwn. MIDs & Normal Controls

3) Diagnostic Model for MIDs

Discussion

⇒ Important to detect MIDs in the early stage because MIDs are initial symptoms of brain-

damaged disorders

MIDs: Clinical Significance

Causes of decrease in motor skills (Holvia et al., 2012; Ward et al., 2003)

Internal factors: age ↑, skeletal muscle mass ↓, muscle strength ↓, cognitive ability ↓

External factor: brain damage

Motor intentional disorders (MIDs) Definition: Motor disorders that disrupt volitional movements (Seo et al., 2009)

Etiology: Damage in the premotor region, mainly appeared in brain-damaged patients

(e.g., vascular dementia, Parkinson’s disease, stroke) (Weintraub, 2008; Hong, 2010)

Symptom: Force control capabilities (FCC) ↓ motor skills ↓ (Seo et al., 2010)

frontal lobe

premotor region

3

Different types of MID by force control phase

Force Control Phases

⇒ Existing diagnostic approach: Behavioral observation & bedside test (Crucian et al., 2007)

⇒ Need to develop a quantitative system specialized for the diagnosis of MID

0

5

10

15

20

25

0 10 20 30 40 50 60 70 80 90

Force (N)

Time (ms)

Motor↓ hypokinesia

Motor↓ impersistence

Motor↓ perseveration

DevelopmentMaintenance

Termination

Initiation

Four phases by force control (Heilman, 2004)

4

Existing Studies on FCC

Mainly focused on force initiation and maintenance phases

Age effect on reaction time (Der and Deary, 2006)

Age and target force effects on force fluctuation (Vinoth et al., 2001)

⇒ Lack of studies on development and termination phases of FCC

⇒ Need to analyze FCC according to four force control phases

MenWomen

Rea

ctio

n Ti

me

(mse

c)

20 30 40 50 60 70 80

5

Research Objective

Comparison of Motor Skills

between Patients with MIDs and Normal Controls

by Evaluating Finger Force Control Capabilities (FCCs)

1. Analysis of FCCs in normal controls by force control phase

2. Comparison of MID patients with normal controls

3. Development of a diagnostic model for early screening of MIDs

6

Research Protocol

S1. System development ▪ H/W: Finger Touch (FT) system

▪ S/W: force evaluation interface

S2. Experiment ▪ Participants: 360 normal controls (20 ~ 70s)

▪ Evaluation of finger FCCs using FT system

S4. Diagnostic model development ▪ Binary logistic regression

▪ ROC-curve analysis

S3. Analysis ▪ Age, gender, and hand effects on FCC

▪ Patients (aMCI, svMCI, SVaD) vs. controls

7

S1. System Development

Finger Touch (FT) (SeedTech Co., South Korea)

Assessment of FCC by force control phase

Two finger dynamometers (load cells) (precision = 0.196 N, sampling rate = 30 ~ 32 Hz)

19-inch monitor (FLATRON L1940P, LG Electronics Co., South Korea)

Evaluation S/W

Finger dynamometer

H/W (FT system & monitor)

9.8 N

0 N

Target force

Hand: L Hand: R

Interface of SW

8

S2. Experiment: Design

Participants: 360 normal controls (30 males & 30 females from each of age strata 20s to 70s)

Three-factor mixed-subjects designs

Independent variables

1) Age (b-s-f): 20s ~ 70s

2) Gender (b-s-f): male, female

3) Hand (w-s-f): left, right 8 trials for each hand and force control phase

Dependent variables

Initiation time (IT; msec)

Development time (DT; msec)

Maintenance error (ME; mN)

Termination time (TT; msec)

Layout of FT system (e.g., test condition = left-hand & left-side)

Screen

Fingerdynamometer

15 cm

50 cm

70 cm

1 / 7

9

4 measures as indicators of motor skills (Seo et al., 2009)

Speed of movements (unit: msec)

Initiation time (IT)

Development time (DT)

Termination time (TT)

S2. Experiment: Measures

Accuracy of movements (unit: mN)

Maintenance error (ME)

0

2

4

6

8

10

12

0 20 40 60 80 100

Forc

e (N

)

Time (ms)

Target force (9.8N)

Signal Signal

IT DT TTME

4 tasks

2 / 7

10

S2. Experiment: Initiation Phase

IT (msec): time to press the dynamometer after a visual signal

DT (msec): time to reach to the target force 9.8 N

ME (mN): average difference between the exerted and target forces

TT (msec): time to release the force from the dynamometer after a visual signal

Signal display

𝑰𝑰𝑰𝑰 = 𝒕𝒕𝒋𝒋 − 𝒕𝒕𝒊𝒊where, ti = time to present a visual signal

tj = time to press the dynamometer

3 / 7

S2. Experiment: Development Phase

IT (msec): reaction time to press the dynamometer after a visual signal

DT (msec): time to reach to the target force 9.8 N

ME (mN): average difference between the exerted and target forces

TT (msec): time to release the force from the dynamometer after a visual signal

Targetforce

𝑫𝑫𝑰𝑰 = 𝒕𝒕𝒋𝒋 − 𝒕𝒕𝒊𝒊where, ti = time to press the dynamometer

tj = time to reach 9.8 N

4 / 7

S2. Experiment: Maintenance Phase

IT (msec): reaction time to press the dynamometer after a visual singal

DT (msec): time to reach to the target force 9.8 N

ME (mN): average difference between the exerted and target forces

TT (msec): time to release the force from the dynamometer after a visual signal

“Keep press the button at 9.8N for 10 sec”

Targetforce

𝑴𝑴𝑬𝑬 =∑𝒊𝒊=𝟎𝟎𝟏𝟏𝟎𝟎𝟎𝟎𝟎𝟎𝟎𝟎 𝒇𝒇𝒊𝒊 − 𝟗𝟗.𝟖𝟖 𝑵𝑵

𝟏𝟏𝟎𝟎𝟎𝟎𝟎𝟎𝟎𝟎where, fi = finger force at measurement i

5 / 7

S2. Experiment: Termination Phase

IT (msec): time to press the dynamometer after a visual signal

DT (msec): time to reach to the target force 9.8 N

ME (mN): average difference between the exerted and target forces

TT (msec): time to release the force from the dynamometer after a visual signal

Targetforce

𝑰𝑰𝑰𝑰 = 𝒕𝒕𝒋𝒋 − 𝒕𝒕𝒊𝒊where, ti = time to present a visual signal

tj = time to release the force from the dynamometer

6 / 7

S2. Experiment: Procedure

S1. Preparation

S2. Practice

S3. Main experiment

S4. Debriefing

(3 min.)

(5 min.)

(10 min.)

(2 min.)

Duration: 20 min.

※ Tested in random orderIndex finger

7 / 7

Practice Test

Maintenance Maintenance

Development

Initiation

Termination

Development

Initiation

Termination

End4 times per each phase

16 times per each phase

Perform each test after enough practice

15

S3. Result: Effects of Age, Gender, Hand on FCCs

Age, gender, and hand effects on FCCs of normal controls

1 / 3

Initiation time (IT) Development time (DT) Maintenance error (ME) Termination time (TT)

Age (A) < .001 < .001 < .001 < .001

Gender (G) .009 .012 < .001 .003

Hand (H) .803 .008 .644 .032

A × G < .001 .314 < .001 .379

A × H .064 .756 .227 .515

G × H .310 .667 .705 .242

A × G × H .768 .851 .336 .568

NormalizedFCC

※ Shaded area: p < .001

⇒ All phases: Age ↑ FCC ↓⇒ Degree of motor skill decrease: ME > IT ≈ DT ≈ TT

0

50

100

150

200

250

20 ~30s

40 ~50s

60s 70s0

50

100

150

200

250

20 ~30s

40 ~50s

60s 70s0

50

100

150

200

250

20 ~30s

40 ~50s

60s 70s0

50

100

150

200

250

20 ~30s

40 ~50s

60s 70s

MaleFemale

16

S3. Result: Speed of Normal Controls

Age effect: age ↑ IT, DT, TT ↑(IT: F[5, 344] = 18.40, p < .001*; DT: F[5, 347] = 5.77, p < .001*; TT: F[5, 341] = 19.08, p < .001*)

2 / 3

※ MD = mean difference

Initiation (IT) Development (DT) Termination (TT)

A A A B B C

0

100

200

300

400

20s 30s 40s 50s 60s 70s

A A A A A B

0

100

200

300

400

20s 30s 40s 50s 60s 70s

A A B B C D

0

100

200

300

400

20s 30s 40s 50s 60s 70sAge group

Tim

e (m

sec)

MD = 27 msecMD = 55 msec MD = 102 msec

⇒ IT, DT, and TT of age 70s: 1.2 ~ 1.3 times ↑ than those of younger adults

1.2 times

1.2 times

1.3 timesYounger adults

17

S3. Result: Accuracy of Normal Controls

Age effect: age ↑ ME ↑ (F[5, 347] = 47.04, p < .001*)

Gender effect: ME of male < ME of female (F[1, 347] = 53.03, p < .001*)

A×G effect: age ↑ gender difference of ME ↑ (F[5, 347] = 7.18, p < .001*)

3 / 3

※ MD = mean difference

⇒ ME of 70s: 2.1 times ↑ than that of 20s ~ 50s

⇒ ME of female: 1.4 times ↑ than that of male

A A A AB

C

0

100

200

300

400

500

20s 30s 40s 50s 60s 70s

Mai

nten

ance

err

or (m

N) MD = 246 mN

AB

0

100

200

300

400

500

Female Male

Mai

nten

ance

err

or (m

N) MD = 85 mN

GenderAge group0

100

200

300

400

500

600

20s 30s 40s 50s 60s 70s

Mai

nten

ance

err

or (m

N) Male

Female

2.1 times1.4 times

A × G

18

60s ~ 70s of normal controls

Type of brain-damaged patients (Yoon et al., 2012)

Amnestic MCI (aMCI)

Subcortical vascular MCI (svMCI)

Subcortical vascular dementia (SVaD)

S4. Comparison: Patients vs. Controls

⇒ FCCs in all phases: controls > MID patients

⇒ Decline in motor skills: SVaD > svMCI > aMCI

Dementia

Amnestic MCI(aMCI)

Vascular MCI(svMCI)

Alzheimer’sdisease

VascularDementia

(SVaD)Severe

WeakMild

CognitiveImpairment

Initiation Development Maintenance Termination

IT (m

sec)

SVaDsvMCIaMCINormal

600

500

400

300

200

100

0

A BC

DT

(mse

c)

SVaDsvMCIaMCINormal

300

250

200

150

100

50

0

A

B

ME

(mN)

SVaDsvMCIaMCINormal

1800

1600

1400

1200

1000

800

600

400

200

0A B

C D TT (m

sec)

SVaDsvMCIaMCINormal

1400

1200

1000

800

600

400

200

0A

BC

D

× 1.6 × 1.7× 2.9

× 3.4

× 1.6

× 2.8

(severity ↑) (severity ↓)

× 1.2

× 1.4× 1.3

19

Objective: Early screening of MID patients

Approach Classification method: Binary logistic regression

Selected variables: TT, ME, IT, age (stepwise method, 𝛼𝛼in, out = 0.05)

Data set: normal controls vs. aMCI + svMCI patients

S4. Diagnostic Model: Method

Control vs. Case

Diagnosis = 0

Normal aMCI svMCI

MID

Diagnosis = 1

n = 98 n = 16 n = 14

20

Comparison of performance according to cut-off threshold

S4. Diagnostic Model: Performance

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.0 0.2 0.4 0.6 0.8 1.0

sens

itivi

ty

1 – specificity

Cut-off = 0.21

Cut-off = 0.12

No. Cut-off Confusion matrixPerformance (%)

Sensitivity Specificity Accuracy

1 0.12 93.3 69.4 75.0

2 0.21 76.7 81.6 80.5

3 0.43 70.0 98.0 91.4

n = 128Actual

Normal Patient

Predicted

Normal 68 2Patient 30 28

n = 128Actual

Normal Patient

Predicted

Normal 80 7

Patient 18 23

n = 128Actual

Normal Patient

Predicted

Normal 96 9

Patient 2 21

Cut-off = 0.43

⇒ Maximize sensitivity (> 90%) for early screening of MID

21

Assessment of FCCs: comparison of MID patients (aMCI, svMCI, SVaD) with normal

controls by force control phase (initiation, development, maintenance, termination) Decrease in motor skills according to severity: SVaD > svMCI > aMCI

⇒ ME, TT: Discriminant factors contribute to distinguish MID patients and controls

Diagnostic model development for early screening of MID(Sensitivity for MCI = 93%, sensitivity for SVaD = 100%, specificity = 69%, accuracy = 75%)

⇒ Useful for MID diagnosis in the early stage

Discussion

100

130

160

190

220

250

20s ~ 30s 40s ~ 50s 60s 70s

Nor

mal

ized

val

ue

Motor skills of controls and patientsAge effect on motor skills for normal controls

100

150

200

250

300

350

Normal aMCI svMCI SVaD

Nor

mal

ized

val

ue

ME > TT > DT ≈ ITME > TT > DT ≈ IT

MEME

TT

TT

DT ITDTIT

22

Q & A

Thank you for your attention!

23