Driving behaviour of drivers with Mild Cognitive Impairment and Alzheimer’s Disease: A Driving Simulator Study

Abstract

The objective of this research is the analysis of the driving performance of
drivers with Alzheimer's disease (AD) and Mild Cognitive Impairment (MCI), on the basis of a
driving simulator experiment, in which healthy “control” drivers and impaired drivers drive in
different driving scenarios, following a thorough neurological and neuropsychological
assessment of all participants. The driving scenarios include driving in rural and urban areas
in low and high traffic volumes. The driving performance of drivers impaired by the ...

Full text

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G.
DRIVING BEHAVIOUR OF DRIVERS WITH MILD COGNITIVE IMPAIRMENT
AND ALZHEIMER’S DISEASE: A DRIVING SIMULATOR STUDY
Dimosthenis Pavlou (Corresponding author)
Research Assistant
National Technical University of Athens
Department of Transportation Planning and Engineering
5 Heroon Polytechniou st., GR-15773 Athens
Tel: +302107721380, Fax: +302107721454
E-mail: dpavlou@central.ntua.gr
Eleonora Papadimitriou, PhD
Research Associate
National Technical University of Athens
Department of Transportation Planning and Engineering
5 Heroon Polytechniou st., GR-15773 Athens
Tel: +302107721380, Fax: +302107721454
E-mail: nopapadi@central.ntua.gr
Constantinos Antoniou
Associate Professor
National Technical University of Athens
School of Rural and Surveying Engineering
9 Heroon Polytechniou st., GR-15780 Athens
Tel: +302107722783, Fax: +302107722629
E-mail: antoniou@central.ntua.gr
Panagiotis Papantoniou
Research Assistant
National Technical University of Athens
Department of Transportation Planning and Engineering
5 Heroon Polytechniou st., GR-15773 Athens
Tel: +302107721376, Fax: +302107721454
E-mail: ppapant@central.ntua.gr
George Yannis
Professor
National Technical University of Athens
Department of Transportation Planning and Engineering
5 Heroon Polytechniou st., GR-15773 Athens
Tel: +302107721326, Fax: +302107721454
E-mail: geyannis@central.ntua.gr
John Golias
Professor
National Technical University of Athens
Department of Transportation Planning and Engineering
5 Heroon Polytechniou st., GR-15773 Athens
Tel: +302107721276, Fax: +302107721454
E-mail: igolias@central.ntua.gr
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 2
Sokratis G. Papageorgiou
Associate Professor
Attikon General University Hospital
University of Athens Medical School, Department of Neurology
75 Mikras Asias str., GR-11527, Athens, Greece
Tel: +302107289404, Fax: +302107216474
E-mail: sokpapa@med.uoa.gr
Word count: 5080 + 1 figure + 5 tables = 6580
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 3
ABSTRACT
The objective of this research is the analysis of the driving performance of drivers with
Alzheimer’s disease (AD) and Mild Cognitive Impairment (MCI), on the basis of a driving
simulator experiment, in which healthy “control” drivers and impaired drivers drive in
different driving scenarios, following a thorough neurological and neuropsychological
assessment of all participants. The driving scenarios include driving in rural and urban areas
in low and high traffic volumes. The driving performance of drivers impaired by the
examined pathologies (AD and MCI) is compared to that of healthy controls by means of
Repeated Measures General Linear Modeling techniques. In this paper a sample of 75
participants is analyzed. Various driving performance measures are examined, including
speed, lateral position, steering angle, headway, reaction time at unexpected events etc., some
in terms of their mean values and some in both their mean values and their variability. The
results suggest that the two examined cerebral diseases do affect driving performance, and
there are common driving patterns for both cerebral diseases, as well as particular
characteristics of specific pathologies. More specifically, drivers with these cerebral diseases
drive at lower speeds and with larger headway compared to healthy drivers. Moreover, they
appear to have difficulties in positioning the vehicle on the lane. Cerebral diseases also
appear to significantly affect reaction times at incidents.
Key-words: driving performance; driving simulator; Mild Cognitive Impairment;
Alzheimer’s disease
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G.
BACKGROUND 1
2
The task of driving requires the ability to receive sensory information, process the 3
information, and to make proper, timely judgments and responses (1, 2). Various motor, 4
visual, cognitive and perceptual deficits can affect the ability to drive. These deficits are 5
either age-related or caused by neurologic disorders and lead to reduced driver fitness and 6
increased crash risk. More specifically, diseases affecting a person's brain functioning (e.g. 7
presence of specific brain pathology due to neurological diseases as Alzheimer’s disease) 8
may significantly impair the person's driving ability (3, 4, 5, 6). These conditions have 9
obvious impacts on driving performance, but in mild cases and importantly in the early 10
stages, they may be imperceptible in one’s daily routine yet still impact one’s driving ability. 11
Furthermore, neuropsychological parameters associated with driving performance are 12
reaction time, visual attention, speed of perception and processing, and general cognitive and 13
executive functions. These parameters show considerable decline with age or at the presence 14
of cognitive impairments and are associated with the probability of accident involvement (7). 15
Relatively little is known about the competence of drivers with Mild Cognitive 16
Impairment (MCI). This constitutes a considerable gap, given that MCI is a pathological 17
condition with high prevalence in the general population as ~15% of people >65 years old are 18
affected. In addition, MCI eventually develops into dementia with a high annual rate (8). The 19
concept of MCI has been described as a cognitive state that lies between normal aging and 20
dementia (9). Persons with MCI exhibit cognitive decline beyond what is expected to be 21
normal for age, but are otherwise functioning well and do not meet criteria for dementia. 22
Research results are not conclusive on the extent to which MCI is affecting driving behaviour 23
and safety. MCI drivers seem to have statistically significant driving behaviour deviation 24
(maintaining speed, wheel stability, and lateral control) from the control driving population 25
(10). Another study tried to ascertain which cognitive features contribute to the safe driving 26
behaviour of MCI drivers. Participants drove using a driving simulator and seemed to have 27
considerable difficulties in maintaining lateral control on a road and in following the vehicle 28
ahead (11). 29
Moreover, Alzheimer’s disease (AD) is the most frequent form of dementia 30
worldwide (12). In the early stages of the disease, a variety of symptoms can be observed 31
with gradually progressive memory impairment being the most prominent symptom. 32
Additional deficits may be present, including, visuospatial deficits, impaired attention, 33
executive dysfunction and judgment, verbal fluency and confrontation naming (13). Another 34
research showed that AD drivers (especially the elderly) made many more safety errors (the 35
most common errors were lane violations) (14). Longitudinal evidence was provided for a 36
decline in driving performance over time, primarily in early-stage dementia of the Alzheimer 37
type (15). Mild AD significantly impaired simulated driving fitness, while MCI limitedly 38
affected driving performance (16). What is more, an accurate judgment of someone’s own 39
ability to drive and the resultant compensatory behaviour are prerequisites of safe driving, an 40
ability that is often impaired in dementia (17). 41
Given that the percentage of the elderly in society is increasing (18), and that the level 42
of motorization also increases (19), the investigation of the impact of these conditions on 43
driver performance becomes quite critical. It is also highlighted that relatively few studies 44
exist analyzing the effect of a specific pathology on driving performance, and even fewer 45
studies comparing different pathologies. 46
47
48
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 5
OBJECTIVES 49
50
The objective of this research is to analyze the driving performance of drivers with 51
Alzheimer’s disease (AD) and Mild Cognitive Impairment (MCI), by means of a driving 52
simulator experiment. Various driving performance measures are examined in both rural and 53
urban environment, e.g. mean speed, lateral position, steering angle, headway, reaction time 54
at unexpected events etc. The driving performance of drivers impaired by the above 55
pathologies is compared to that of healthy controls by means of Repeated Measures General 56
Linear Modeling techniques. 57
The research questions that are examined in this paper are: how MCI and AD affect 58
various measures of driving performance and how these diseases interact with road and 59
traffic parameters. 60
The paper starts a presentation of a large driving simulator experiment, in which the 61
driving performance of the impaired and healthy drivers was examined in different driving 62
scenarios, following a thorough neurological and neuropsychological assessment of all 63
participants. The existing sample size and characteristics are presented next, followed by a 64
short description of the analysis methods, dependent and independent variables. The results 65
are presented and discussed, and some concluding remarks are provided. 66
67
DRIVING SIMULATOR EXPERIMENT 68
69
Overview 70
71
This research is based on a methodological framework for the combined assessment of 72
traffic, behavioural, medical, neurological and neuropsychological parameters on driving 73
performance. In this framework, the aspects of driver behaviour and safety research 74
addressed are inherently interdisciplinary, and an experiment was designed by an 75
interdisciplinary research team including: 76
 Transportation Engineers - Department of Transportation Planning and 77
Engineering, of the National Technical University of Athens (NTUA) 78
 Neurologists - 2nd Department of Neurology, University of Athens Medical School, 79
at ATTIKON University General Hospital, Haidari, Athens 80
 Neuropsychologists - Department of Psychology, University of Athens, the 2nd 81
Department of Neurology of ATTIKON University General Hospital, Haidari, Athens 82
and the Aristotle University of Thessaloniki. 83
According to the objectives of the analysis, the experiment includes three types of 84
assessment: 85
 Medical / neurological assessment: The first assessment concerns the administration 86
of a full clinical medical, ophthalmological and neurological evaluation, in order to 87
well document the characteristics of each of these disorders (e.g. MCI, AD, PD, 88
Cerebrovascular disease (stroke) as well as other related parameters of potential 89
impact on driving (e.g. use of medication affecting the Central Nervous System). 90
 Neuropsychological assessment: The second assessment concerns the administration 91
of a series of neuropsychological tests and psychological-behavioural questionnaires 92
to the participants. The tests carried out cover a large spectrum of Cognitive 93
Functions: visuospatial and verbal episodic and working memory, general selective 94
and divided attention, reaction time, processing speed, psychomotor speed etc. 95
 Driving at the simulator: The third assessment concerns the driving behaviour by 96
means of programming of a set of driving tasks into a driving simulator for different 97
driving scenarios. 98
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 6
The first and second assessments are carried out at the ATTIKON University General 99
Hospital, and their description is beyond the scope of this paper; for details the reader is 100
referred to Papadimitriou et al. (2014) (20). The third assessment, (driving simulator 101
experiment) takes place in the NTUA Road Safety Observatory and is presented in detail in 102
the following section. 103
104
Driving at the simulator 105
106
The NTUA driving simulator is a motion base quarter-cab manufactured by the FOERST 107
Company. The simulator consists of 3 LCD wide screens 40’’ (full HD: 1920x1080pixels), 108
driving position and support motion base. The dimensions at a full development are 109
230x180cm, while the base width is 78cm and the total field of view is 170 degrees. It’s 110
worth mentioning that the simulator is validated against a real world environment (21). 111
The design of the driving scenarios includes driving in different road and traffic 112
conditions, such as in a rural, urban area with high and low traffic volume, with or without 113
external distraction. More specifically, the driving simulator experiment begins with one 114
practice drive (usually 10-15 minutes), until the participant fully familiarizes with the 115
simulation environment. Afterwards, the participant drives two sessions (approximately 20 116
minutes each). Each session corresponds to a different road environment: a rural route that is 117
2.1 km long, single carriageway and the lane width is 3m, with zero gradient and mild 118
horizontal curves and an urban route that is 1.7km long, at its bigger part dual carriageway, 119
separated by guardrails and the lane width is 3.5m. Two traffic controlled junctions, one stop-120
controlled junction and one roundabout are placed along the route. 121
Within each road / area type, two traffic scenarios and three distraction conditions are 122
examined in a full factorial within-subject design. The traffic conditions examined include: 123
 Low traffic conditions - ambient vehicles’ arrivals are drawn from a Gamma 124
distribution with mean m=12sec, and variance σ2=6 sec, corresponding to an average 125
traffic volume Q=300 vehicles/hour. 126
 High traffic conditions - ambient vehicles’ arrivals are drawn from a Gamma 127
distribution with mean m=6sec, and variance σ2=3 sec, corresponding to an average 128
traffic volume of Q=600 vehicles/hour. 129
The distraction conditions examined concern undistracted driving, driving while 130
conversing with a passenger and driving while conversing with a mobile phone. 131
Consequently, in total, each session (urban or rural) includes six trials of the 132
simulated route. During each trial, 2 unexpected incidents are scheduled to occur at fixed 133
points along the drive. More specifically, incidents in rural area concern the sudden 134
appearance of an animal (deer or donkey) on the roadway, and incidents in urban areas 135
concern the sudden appearance of an adult pedestrian or of a child chasing a ball on the 136
roadway or of a car suddenly getting out of a parking position and getting in the road. The 137
hazard does appear at the same location for the same trial (i.e. rural area, high traffic) but not 138
at the same location between the trials, in order not to have learning effects. Regarding the 139
time that the hazard appears, it depends on the speed and the time to collision in order to have 140
identical conditions for the participant to react, either they drive fast or slowly. Thus, there is 141
no possibility for the incident to appear closely or more suddenly to a participant than to 142
another. 143
The experiment is counterbalanced concerning the number and the order of the trials. 144
However, rural drives were always first and urban drives were always second. This was 145
decided for the following reasons: It was observed that urban area causes more often 146
simulation sickness to the participants and thus it was decided to have the urban scenario 147
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 7
second and secondly, counterbalancing in driving area means that we would have twice as 148
much driving combinations which leads to much larger sample size requirements. 149
Finally, impaired participants are to carry out the simulator experiment while under 150
their usual medication, so that their driving performance corresponds to their everyday 151
condition, as treated by their neurologist. 152
153
ANALYSIS METHODS AND DATA 154
155
The aim of this research is to analyze and compare the driving performance of MCI, AD and 156
healthy drivers in rural and urban road environment. For that purpose, four trials of the 157
simulator experiment are selected: the undistracted driving trials in rural area and the 158
undistracted driving trials in urban area in both low and high traffic volumes. 159
The analysis method selected is the Repeated Measures General Linear Model 160
(GLM). The repeated measures GLM is the equivalent of the one-way ANOVA, but for 161
related, not independent groups. A repeated measures GLM may be based on a within-162
subjects or a mixed design (22). 163
At the present time more than 140 participants have participated in the driving 164
simulator experiment in approximately 15 months time. However, about 30 participants had 165
simulator sickness issues (a usual phenomenon in driving simulators) and didn’t complete the 166
driving trials of the experiment. For that reason they are eliminated from the study. Moreover 167
there are 35 participants of younger age (<55 years old) who are eliminated too for age 168
representativity reasons. The analysis is thus based on the existing related sample of the 169
(ongoing) simulator experiment of healthy and impaired participants of over than 55 years of 170
age who completed all of the examined four trials were selected, which consists of 75 171
participants (49 males). More specifically, the sample of the present study consists of: 172
 38 healthy “controls” (66.4 years old on average), 173
 14 AD patients (74.6 years old on average) and 174
 23 MCI patients (68.3 years old on average). 175
It is noted that the gender distribution of healthy and impaired drivers is currently not 176
fully similar, i.e. the proportion of females is lower in the impaired drivers group (no female 177
AD participant), which is in any case representative of the general population. On the other 178
hand, the age distributions of impaired and healthy drivers are comparable to a satisfactory 179
degree, taking into account that it is expected that impaired drivers are on average older than 180
healthy ones (see Figure 1). 181
182
FIGURE 1 Age distribution of the sample, health condition and gender distribution 183
184
The variables examined in the present research include a between-subject variable, 185
namely the presence of a disease (AD or MCI). They also include one within-subject 186
variables, namely the traffic scenario (low or high traffic volume). It is noted that area type 187
(rural, urban) is not examined as a within-subject variable, because all participants drove first 188
50
55
60
65
70
75
80
85
Age
Sample characteristics
Male
Female
AD Control
MCI
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 8
in rural area and then in urban area; this was done for practical reasons but obviously results 189
in order effects, and consequently the two area types are examined separately and not 190
comparatively. The driving performance measures examined include both longitudinal 191
control measures and lateral control measures. More specifically: 192
 Longitudinal control measures: 193
o Mean speed (mean speed of the driver along the route, excluding the small 194
sections in which incidents occurred, and excluding junction areas) 195
o Headway (time distance between the front of the simulator vehicle and the front 196
of the vehicle ahead) 197
o Reaction time at unexpected incidents (time between the first appearance of 198
the event on the road and the moment the driver starts to brake in milliseconds) 199
 Lateral control measures: 200
o Lateral position (vehicle distance from the central road axis in meters), 201
o Lateral position variability (the standard deviation of lateral position), 202
o Mean wheel steering angle (in degrees) 203
o Steering angle variability (the standard deviation of steering angle). 204
205
RESULTS 206
207
A Repeated Measures General Linear Model was developed for each one of the driving 208
performance measures considered. The analysis of variance for the within subject variables 209
(Table 1) indicated that traffic volume has a significant effect on mean speed, mean headway 210
and lateral position in both road environments, and lateral position variability and steering 211
angle variability only on rural road. Regarding the between-subject variable, the presence of a 212
disease was found to significantly affect mean speed and reaction time in both road 213
environments. The presence of a cerebral disease seems to affect mean headway, lateral 214
position variability and steering angle variability only in rural roads and lateral position only 215
in urban road environment. 216
217
TABLE 1 Tests of within and between subjects in rural and urban road environment 218
219
Rural Road
Urban Road
Tests of Within-
Subjects Contrasts
(Source Traffic)
Tests of Between-
Subjects Effects
(Source Disease)
Tests of Within-
Subjects Contrasts
(Source Traffic)
Tests of Between-
Subjects Effects
(Source Disease)
F
p-value
F
p-value
F
p-value
F
p-value
Mean speed
(km/h)
17,292
,000**
24,634
,000**
20,327
,000**
6,000
,004**
Mean headway
(sec)
69,665
,000**
14,218
,000**
9,569
,003**
,294
,746
Reaction time
(millisec)
1,785
,186
2,828
,066*
,466
,498
2,656
,078 *
Lateral position
(m)
106,116
,000**
,375
,689
5,690
,021**
2,552
,085 *
Lateral position variability
(st.dev of lateral position)
29,125
,000**
4,840
,011**
,430
,515
1,374
,262
Steering angle
(degrees)
1,368
,246
,358
,701
,051
,823
,381
,685
Steering angle variability
(st.dev of steering angle)
9,586
,003**
3,435
,038**
,037
,849
,313
,732
* significant at 90%, ** significant at 95% 220
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 9
221
Effect of cerebral diseases in rural roads 222
223
The results of the GLMs fitted to the data for the various longitudinal and lateral control 224
measures of the rural driving session, in terms of parameter estimates and their statistical 225
significance, are presented in Tables 2 and 3. Table 2 refers to the longitudinal control 226
measures in rural area, whereas table 3 refers to lateral control measures in rural area. 227
Cerebral diseases appear to have a significant effect on driver mean speed in rural 228
driving environment. AD and MCI patients drive at significantly lower mean speed compared 229
to healthy drivers, both at low and high traffic volumes. AD drivers’ speed is significantly 230
lower than the MCI drivers’ speed, in both driving environments. 231
Moreover, cerebral diseases appear to have a significant effect on mean headway in 232
rural roads but only for AD patients: they have significantly longer mean headway compared 233
to healthy drivers at both traffic environments. This is happening for MCI drivers too, but the 234
confidence level was only 85%. AD drivers have much longer mean headway compared with 235
the MCI drivers. These results are intuitive, given that lower speeds naturally result in larger 236
headways, with a given distribution of ambient traffic on the road network. It is also noted 237
that headways at low traffic volumes are longer for all driver groups, which is also intuitive. 238
Significant differences in the driving behavior of healthy and impaired drivers were 239
also identified as regards the drivers’ reaction time at unexpected incidents in rural roads 240
(sudden appearance of a deer or a donkey). In both traffic environments impaired drivers 241
have about 0.5 sec longer reaction times than the healthy ones. This difference was found to 242
be statistically significant at 90% confidence level for both impaired groups and both traffic 243
volumes, except for MCI drivers in high traffic volume who have longer reaction times than 244
the control group statistically significant at 95% confidence level. 245
246
TABLE 2 Parameter estimates of the repeated measures GLM - Longitudinal control 247
measures for rural driving environment 248
249
Parameter Estimates
Low Traffic Volume
High Traffic Volume
Dependent
Variable
B
Std.
Error
t
Sig.
B
Std.
Error
t
Sig.
Mean
speed
(km/h)
Intercept
47,907
1,207
39,699
,000
**
45,296
,993
45,595
,000
**
MCI
-6,112
1,965
-3,110
,003
**
-6,235
1,618
-3,854
,000
**
AD
-13,982
2,326
-6,012
,000
**
-13,383
1,915
-6,990
,000
**
Control
0
0
Mean
headway
(sec)
Intercept
46,634
4,759
9,799
,000
**
22,382
4,730
4,732
,000
**
MCI
12,361
7,750
1,595
,115
12,035
7,703
1,562
,123
AD
40,432
9,172
4,408
,000
**
51,314
9,116
5,629
,000
**
Control
0
0
Reaction
time
(millisec)
Intercept
923,048
153,950
5,996
,000
**
996,250
159,113
6,261
,000
**
MCI
481,918
250,715
1,922
,059
*
532,628
259,123
2,056
,043
**
AD
580,278
296,700
1,956
,054
*
446,428
266,688
1,674
,097
*
Control
0
0
* significant at 90%, ** significant at 95% 250
251
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 10
Regarding lateral position in rural area, it is worth mentioning that the width of the 252
driving lane is 3m (i.e. very narrow), so the drivers don’t have so much flexibility in 253
positioning their vehicle on the lane. Thus, there are no significant differences in lateral 254
position for the drivers. Positive values indicate driving more closely to the right border of 255
the road. 256
On the other hand, the lateral position variability seems to have differences for MCI 257
drivers in both traffic volumes. Lateral position variability is lower than that of healthy 258
controls, and this may be a result of the lower speed and their more conservative driving. 259
Finally, no statistically significant differences are observed in mean steering angle in 260
rural area, between control group and impaired drivers - a positive mean steering angle means 261
more counter-clockwise steering movements, which is in accordance with a lateral position 262
closer to the central road axis. On the other hand, there is statistically significant variability in 263
steering angle; all examined impaired drivers in high traffic volume environment have lower 264
steering angle variability. 265
266
TABLE 3 Parameter estimates of the repeated measures GLM - Lateral control 267
measures for rural driving environment 268
269
Parameter Estimates
Low Traffic Volume
High Traffic Volume
Dependent
variable
B
Std.
Error
t
Sig.
B
Std.
Error
t
Sig.
Lateral position
(m)
Intercept
1,491
,024
61,979
,000
**
1,605
,022
72,596
,000
**
MCI
,029
,039
,746
,458
,030
,036
,830
,410
AD
,010
,046
,224
,823
,014
,043
,328
,744
Control
0
0
Lateral position
variability
(st.dev of lateral
position)
Intercept
,299
,009
31,520
,000
**
,266
,009
29,164
,000
**
MCI
-,036
,015
-2,330
,023
**
-,027
,015
-1,821
,073
*
AD
,024
,018
1,310
,194
,017
,018
,994
,324
Control
0
0
Steering angle
(degrees)
Intercept
-1,793
,082
-21,993
,000
**
-1,949
,090
-21,680
,000
**
MCI
-,102
,133
-,770
,444
,230
,146
1,571
,121
AD
,045
,157
,284
,777
-,131
,173
-,757
,452
Control
0
0
Steering angle
variability
(st.dev of
steering angle)
Intercept
17,747
,307
57,739
,000
**
17,646
,260
67,984
,000
**
MCI
-,756
,501
-1,511
,135
-1,088
,423
-2,573
,012
**
AD
-,505
,592
-,853
,397
-1,558
,500
-3,114
,003
**
Control
0
0
* significant at 90%, ** significant at 95% 270
271
Effect of cerebral diseases on urban roads 272
273
The results of the GLMs fitted to the data for the various longitudinal and lateral control 274
measures of the urban driving session, in terms of parameter estimates and their statistical 275
significance, are presented in Tables 4 and 5. Table 4 refers to the longitudinal control 276
measures in urban area, whereas table 5 refers to lateral control measures in urban area. 277
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 11
In urban road environment similar statistical results with the rural area type were 278
observed, regarding the longitudinal control measures. Mean speed is significantly lower for 279
impaired drivers in urban driving environment. AD and MCI drivers seem to drive at the 280
same speed in both at low and high traffic volumes. 281
However, cerebral diseases appear not to have a significant effect on mean headway 282
in urban roads. Only MCI patients seem to have significantly longer mean headway 283
compared to healthy drivers only at high traffic environment. 284
Finally, regarding the reaction times, they appear to be improved for the impaired 285
drivers compared to the rural road. They are more closely to the reaction times of the control 286
group and have significant differences at 90% confidence level only in low traffic volume. 287
This is possibly also due to the learning effect resulting from the fact that the urban area trials 288
took place after the rural area trials for all participants 289
290
TABLE 4 Parameter estimates of the repeated measures GLM - Longitudinal control 291
measures for urban driving environment 292
293
Parameter Estimates
Low Traffic Volume
High Traffic Volume
Dependent
variable
Std.
Error
t
Sig.
B
Std.
Error
t
Sig.
Mean
speed
(km/h)
Intercept
33,677
,899
37,454
,000
**
30,372
,692
43,870
,000
**
MCI
-4,854
1,733
-2,801
,007
**
-3,713
1,334
-2,783
,007
**
AD
-4,357
2,435
-1,789
,079
*
-4,636
1,875
-2,473
,017
**
Control
0
0
Mean
headway
(sec)
Intercept
48,628
5,149
9,444
,000
**
23,784
2,309
10,302
,000
**
MCI
7,538
9,944
,748
,461
11,989
4,449
2,695
,009
**
AD
4,340
13,944
,311
,757
7,266
6,252
1,162
,250
Control
0
0
Reaction
time
(millisec)
Intercept
1294,487
66,621
19,431
,000
**
1284,224
62,967
20,395
,000
**
MCI
198,056
115,973
1,708
,092
*
139,062
121,353
1,146
,257
AD
296,187
165,711
1,787
,078
*
209,693
170,515
1,230
,224
Control
0
0
* significant at 90%, ** significant at 95% 294
295
Regarding lateral position in urban area, MCI patients appear to drive at longer 296
distance from the central road axis compared to healthy drivers, both at high and at low 297
traffic volumes (statistically significant at 90% confidence level). AD drivers in high traffic 298
volume have significant differences in lateral position too. This is observed only in urban 299
road environment and it’ is worth mentioning, that the width of the driving lane is 3,5m, there 300
are 2 lanes in the bigger part of the route, so there are opportunities for overtaking and there 301
are choices in positioning the vehicle on the road. It seems that in urban areas the high traffic 302
volume makes the conditions more complex for the impaired drivers and leads them to drive 303
more closely to the right border of the road. Especially for AD drivers there is significant 304
increase in the variability of the lateral position in high traffic volume (in contrast with all 305
other cases). 306
Statistically significant differences are not observed for mean steering angle, or for 307
the variability in the steering angle between control group and impaired drivers. 308
309
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 12
TABLE 5 Parameter estimates of the repeated measures GLM – Lateral control 310
measures for urban driving environment 311
312
Parameter Estimates
Low Traffic Volume
High Traffic Volume
Dependent
Variable
B
Std.
Error
t
Sig.
B
Std.
Error
t
Sig.
Lateral position
(m)
Intercept
2,961
,103
28,864
,000
**
3,064
,103
29,690
,000
**
MCI
,305
,184
1,756
,099
*
,326
,185
1,762
,083
*
AD
,171
,278
,616
,541
,514
,279
1,839
,071
*
Control
0
0
Lateral position
variability
(st.dev of lateral
position)
Intercept
1,560
,098
15,839
,000
**
1,522
,099
15,351
,000
**
MCI
,210
,190
1,107
,273
,195
,191
1,021
,312
AD
,171
,267
,640
,525
,482
,268
1,797
,078
*
Control
0
0
Steering angle
(degrees)
Intercept
6,967
,203
34,374
,000
**
7,336
,294
24,963
,000
**
MCI
,136
,391
,348
,729
-,379
,566
-,670
,506
AD
,546
,549
,996
,324
,196
,796
,246
,807
Control
0
0
Steering angle
variability
(st.dev of
steering angle)
Intercept
22,872
,753
30,365
,000
**
22,463
1,328
16,918
,000
**
MCI
,368
1,452
,254
,801
1,646
2,559
,643
,523
AD
-,821
2,040
-,402
,689
,102
3,596
,028
,978
Control
0
0
* significant at 90%, ** significant at 95% 313
314
DISCUSSION 315
316
This paper analyzed the driving performance of drivers with cerebral diseases, with focus on 317
the comparative assessment of AD and MCI pathologies. Relatively few studies exist 318
analyzing the effect of a specific pathology on driving performance, and even fewer studies 319
comparing different pathologies. The majority of these studies indicate serious deterioration 320
in driving performance of drivers with a cerebral disease compared to healthy drivers. 321
The research questions examined in this paper are: how the examined pathologies 322
affect various measures of driving performance and how they interact with road and traffic 323
parameters. For this purpose, four trials were selected from a large driving simulator 324
experiment including twelve trials in total, namely those concerning undistracted driving in 325
rural and urban areas with low or high traffic volume. These four trials were based on a 326
mixed (within- and between-subject) counterbalanced design. Both longitudinal and lateral 327
control measures are examined, e.g. speed, lateral position, steering angle, headway, reaction 328
time at unexpected events etc. by means of Repeated Measures General Linear Modeling 329
techniques. This research in progress is one of the few which attempt to compare different 330
pathologies in terms of their effect on driving performance. 331
Summarizing the results, AD and MCI drivers were found to drive at significantly 332
lower speeds compared to the healthy control group drivers, both at low and at high traffic 333
volume. AD drivers in rural environment have even lower mean speed compared to the MCI 334
drivers, but in urban roads their speed is approximately the same. As would be expected, this 335
reduced speed results under given ambient traffic conditions in increased headways, both at 336
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 13
low and at high traffic volumes in rural roads, however in urban environment there are 337
statistically significant differences in mean headways only for MCI drivers in high traffic 338
volume. 339
Analyzing the reaction times of the impaired drivers at unexpected incidents, it is 340
observed that MCI and AD drivers have significantly longer reaction times in rural road in 341
both traffic volumes compared with the control group. In urban area, they have longer 342
reaction times, but only in low traffic volume this difference is significant. Compared with 343
each other, MCI drivers seem to have slightly better reaction times than the AD group in 344
most cases. These longer reaction times of impaired drivers are likely to be confirmed by 345
their neurological and neuropsychological assessments (at the present time the medical and 346
neuropsychological database is under preparation in order to be finalized and used in future 347
statistical analyses, and thus it is not available). 348
Analyzing the lateral control measures it is observed that in rural area there are more 349
statistically significant differences between the driving groups except for lateral position 350
because of the very narrow lane in rural area. More specifically, MCI patients drive more 351
closely to the right border of the road in urban area and in both traffic volumes, whereas AD 352
drivers only in high traffic volume in urban area. Regarding the variability of this measure, a 353
significantly higher variability is highlighted for AD divers in high traffic volume in urban 354
area. It seems that the more complex is the driving environment the more the AD drivers 355
have difficulty in maintaining the position of the vehicle on the lane. Finally, in rural area 356
both impaired groups have low steering variability in high traffic volume that is a result of 357
their low speed and conservative driving. 358
The effect of the sample representativity is something that needs to be highlighted; the 359
age and gender distributions of the impaired and control populations seem balanced at the 360
present time, however sample representativity should be improved in the next steps of the 361
ongoing experiment. The larger proportion of female drivers in the control group is 362
representative, as the proportion of female AD or MCI patients is low in the general 363
population. On the other hand, the average age of the examined groups should be totally 364
balanced, in order to eliminate the possibility that the differences of the diving behaviour 365
between the examined groups are a result of age distribution. 366
Finally, the results are to be considered within the limiting context of driving 367
simulator studies - driving performance is known to be more accurately and reliably 368
estimated by means of on-road studies. However, the relative effects of impaired vs healthy 369
drivers are known to be quite identifiable in simulator studies. 370
371
ACKNOWLEDGEMENT 372
373
This research was carried out within the framework of the Operational Program "Education 374
and Lifelong Learning" of the National Strategic Reference Framework (NSRF), namely the 375
Research Funding Program: THALES. Investing in knowledge society through the European 376
Social Fund, and the Action: ARISTEIA (Action’s Beneficiary: General Secretariat for 377
Research and Technology), co-financed by by the European Union (European Social Fund -378
ESF) and Greek national funds. More information available at: DRIVERBRAIN: 379
http://www.nrso.ntua.gr/driverbrain and DISTRACT: http://www.nrso.ntua.gr/distract. 380
381
REFERENCES 382
383
1. Waller, JA. Physician’s role in highway safety. Functional impairment in driving. N. Y. 384
State J Med, 80, 1980, pp. 1987-1991. 385
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 14
2. Freund B, Gravenstein S, Ferris R, Burke BL, Shaheen E. Drawing clocks and driving 386
cars: Use of Brief Tests of Cognition to Screen Driving Competency in Older Adults. J 387
Gen Intern Med. 20 (3), 2005, pp. 240-244. 388
3. Wood, J. M., C. Worringham, et al. Quantitative assessment of driving performance in 389
Parkinson's disease. J Neurol Neurosurg Psychiatry 76(2), 2005, pp.176-80. 390
4. Cordell, R., H. C. Lee, et al. Driving assessment in Parkinson's disease - a novel 391
predictor of performance?. Mov Disord 23(9), 2008, pp 1217-1222. 392
5. Cubo, E., P. Martinez Martin, et al. What contributes to driving ability in Parkinson's 393
disease. Disabil Rehabil, 2009. 394
6. Frittelli, C., D. Borghetti, et al. Effects of Alzheimer's disease and mild cognitive 395
impairment on driving ability: a controlled clinical study by simulated driving test. Int J 396
Geriatr Psychiatry 24(3), 2009, pp. 232-238. 397
7. Lunsman, M., Edwards, J. D., Andel, R., Small, B. J., Ball, K. K., Roenker, D. L. What 398
predicts changes in Useful Field of View test performance? Psychology and Aging, 23, 399
2008, pp. 917-927. 400
8. Winblad, B., Palmer, K., Kivipelto, M., Jelic, V., Fratiglioni, L., Wahlund, L.O., et al. 401
(2004). Mild cognitive impairment – beyond controversies, towards a consensus: report 402
of the International Working Group on Mild Cognitive Impairment. Journal of Internal 403
Medicine, 256 (3),240–246. 404
9. Petersen, R.C., Smith, G.E., Ivnik, R.J. Tangalos, E.G., Schaid, D.J., Thibodeau, S.N., 405
Kokmen, E., Waring, S.C., Kurland, L.T. (1995). Apolipoprotein E status as a predictor 406
of the development of Alzheimer’s disease in memory impaired individuals. Journal of 407
the American Medical Association, 16, 1274-1278. 408
10. Wadley, V.G., Okonkwo, O., Crowe, M., Vance, D.E., Elgin, J.M., Ball, K.K., Owsley, 409
C. (2009). Mild cognitive impairment and everyday function: an investigation of driving 410
performance. Journal of Geriatric Psychiatry and Neurology, 22(2), 87-94. 411
11. Kawano, N., Iwamoto, K., Ebe, K., Suzuki, Y., Hasegawa, J., Ukai, K., Umegaki, H., 412
Iidaka, T., Ozaki, N. (2012). Effects of mild cognitive impairment on driving 413
performance in older drivers. Journal of the American Geriatrics Society, 60(7), 1379-414
1381. 415
12. Evans, D.A., Funkenstein, H.H., Albert, M.S., Scherr, P.A., Cook, N.R., Chown, H.J., 416
Hebert, L.E., Hennekens, C.H., Taylor, J.O. (1989). Prevalence of Alzheimer’s disease in 417
a community population of older persons: higher than previously reported. J. Am. Med. 418
Assoc. 262, 2551 – 2556. 419
13. Zec, R.F. (1993). Neuropsychological functioning in Alzheimer’s disease. In R. W. 420
Parks, R. F. Zec, & R. S. Wilson (Eds.), Neuropsychology of Alzheimer’s disease and 421
other dementias (pp. 3–80). New York: Oxford University Press. 422
14. Dawson, J.D., Anderson, S.W., Uc, E.Y., Dastrup, E., Rizzo, M., (2009). Predictors of 423
driving safety in early Alzheimer disease. Neurology, 72, 521–527. 424
15. Duchek, J.M., Carr, D.B., Hunt, L., Roe, C.M., Xiong, C., Shah, K., Morris J.C. (2003). 425
Longitudinal Driving Performance in Early-Stage Dementia of the Alzheimer Type. 426
Journal of the American Geriatrics Society, 51, 1342–1347. 427
16. Frittelli, C., Borghetti, D., Iudice, G., Bonanni, E., Maestri, M., Tognoni, G., Pasquali, 428
L., Iudice, A. (2009). Effects of Alzheimer’s disease and mild cognitive impairment on 429
driving ability: a controlled clinical study by simulated driving test. International 430
Journal of Geriatric Psychiatry, 24, 232–238. 431
17. Johansson, K., & Lundberg, C. (1997). The 1994 International Consensus Conference on 432
Dementia and Driving: A brief report. Alzheimer Disease and Associated Disorders, 11, 433
62–69. 434
TRB 2015 Annual Meeting Paper revised from original submittal.

Pavlou D., Papadimitriou E., Antoniou C., Papantoniou P., Yannis G., Golias J.,
Papageorgiou S.G. 15
18. Baldock, M. R. J., Mathias, J., McLean, J., & Berndt, A. Visual attention as a predictor 435
of on-road driving performance of older adults. Australian Journal of Psychology, 59, 436
2007, pp. 159-168. 437
19. Yannis G., Antoniou C., Papadimitriou E., Katsohis D., When may road fatalities start to 438
decrease?, Journal of Safety Research,Vol. 42, Issue 1, 2011, pp. 17-25. 439
20. Papadimitriou E., Pavlou D., Papantoniou P., Yannis G., Golias J., Papageorgiou S.G., 440
"Results from a driving simulator study on performance of drivers with cerebral diseases 441
in rural roads", Proceedings of the 93rd Annual meeting of the Transportation Research 442
Board, Washington, January 2014 443
21. Nikas M., "Comparative analysis of young drivers’ behaviour in normal and simulated 444
conditions in interurban road", Graduate diploma thesis, Department of Transportation 445
Planning and Engineering, School of Civil Engineering, NTUA, January 2014 446
22. Caird, J.K., Horrey, W. Twelve practical and useful questions about driving simulation. 447
In Fisher, D.L., Rizzo, M., Caird, J.K., & Lee, J.D. (Eds). (2011). Handbook of Driving 448
Simulation for Engineering, Medicine, and Psychology. Boca Raton, FL:, 2011, CRC 449
Press. 450
TRB 2015 Annual Meeting Paper revised from original submittal.