Changes in Attention Levels While Driving a Car Estimated Using

Modelling Techniques with Features of Oculo-Motors

Minoru Nakayama

1 a

, Qian (Chayn) Sun

2 b

and Jianhong (Cecilia) Xia

3 c

Institute of Science Tokyo (Tokyo Tech.), O-okayama, Meguro-ku, Tokyo, 152–8552, Japan

RMIT University, Melborune, VIC 3000, Australia

Curtin University, Perth, WA 6102, Australia

Keywords:

Car Driving, Attention, Eye Movement, Pupil, Mental Workload, Modelling.

Abstract:

Changes in attention levels while driving a car were estimated using a modelling technique involving pupillary

changes and the frequency of saccades of 11 drivers. The driving route used in the experiment consisted of 19

sections of road divided into 5 groups: university campus, left turn, straight, right turn, and roundabout. The

sections of road with posted speed limits were divided into 6 conditional states, and model parameters were

estimated by assuming transitions across the states. The estimated model parameters were used to examine

changes in the level of attention resources used during each section of driving. The results of a comparison of

attention resources by section showed a signiﬁcant decrease, in the following order: straight and roundabout,

within campus, left turn and right turn. In addition, the relationship between NASA-TLX was evaluated after

driving and attention resources were examined, and a signiﬁcant correlation with the factor for “difﬁculty”

was conﬁrmed. The relationship between the conﬁdence interval of the change in attention resources and the

factor for “mental demand” was also conﬁrmed.

1 INTRODUCTION

Eye movements of drivers and human visual acu-

ity have been studied in order to improve the safety

of motor vehicle operation (Kapitaniak et al., 2015;

Paeglis et al., 2011; Schmitt et al., 2015; Yamani

et al., 2016). A detailed analysis of images viewed

during driving has also been developed in order to un-

derstand driving behaviour (Palazzi et al., 2019; Hu

et al., 2022). Currently, safety aspects of various in-

telligent vehicles designs which use autonomous driv-

ing systems are frequent points of discussion (Deng

et al., 2020). While human behavioural factors dur-

ing motor vehicle operation may show possible prob-

lems, they can be used to better optimise safe driving

practices, even when autonomous operating systems

are employed. In particular, the driving behaviour

of elderly motorists should be considered when ad-

dressing the issue of safe driving. The relationship

between a driver’s workload and their driving actions

is often studied, and detailed analysis of the relation-

https://orcid.org/0000-0001-5563-6901

https://orcid.org/0000-0002-5421-5838

https://orcid.org/0000-0002-2593-9423

ship is limited (Sun et al., 2016a; Sun et al., 2016b;

Nakayama et al., 2022), however. Elderly motorists

may possess signiﬁcant individual differences in abil-

ity to recognise workload levels, so the relationship

between their own impressions and behaviour-based

attention levels should be extracted.

The authors have introduced modelling techniques

in order to estimate the attention levels of drivers

(Nakayama et al., 2024a; Nakayama et al., 2024b),

though the notations used in the model should be up-

dated to recognise some of the behavioural factors of

some elderly motorists. In the previous study, factors

of road conditions and overall temporal changes were

still unclear. In order to emphasise these factors, the

calculation model for attention levels should be im-

proved.

This paper shows the features of some model pa-

rameters of the experimental driving conditions using

an updated model and a state-space modelling tech-

nique.

The following topics are addressed in this paper.

1. Estimation of attention levels across segments of

driving routes using a state-space model, which is

based on measured saccade rates of eyes and pupil

sizes of individuals while driving a car.

846

Nakayama, M., Sun, Q. C. and Xia, J. C.

Changes in Attention Levels While Driving a Car Estimated Using Modelling Techniques with Features of Oculo-Motors.

DOI: 10.5220/0013101400003911

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 18th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2025) - Volume 1, pages 846-852

ISBN: 978-989-758-731-3; ISSN: 2184-4305

2. The relationship between estimated attention lev-

els and surveyed workload scores are analysed in

order to provide an overall impression of partici-

pating drivers.

2 RELATED WORKS

Eye movement has often been analysed to assess driv-

ing behaviour in various environments, in order to

support safe motor vehicle operation (Palazzi et al.,

2019; K

ubler et al., 2021). In particular, dynamic

visual information processing ability depends on eye

movement behaviour while driving (Kapitaniak et al.,

2015; Paeglis et al., 2011). Driving ability and cogni-

tive performance are sometimes inﬂuenced by ageing.

Behavioural monitoring of aged drivers is necessary

to ensure safe operation of motor vehicles (Schmitt

et al., 2015; Yamani et al., 2016). Some studies have

been conducted to measure the driving speed and de-

viations in position of cars on the road using a global

positioning system (GPS) installed in cars, and mea-

surement of eye movements while driving (Sun et al.,

2015; Sun et al., 2018b). Also, route factors were dis-

cussed when considering elderly drivers, such as pay-

ing attention and perception of the situational envi-

ronment (Sun et al., 2016b; Sun et al., 2018c). Cogni-

tive performance is sometimes considered as a factor

affecting individual drivers. For elderly drivers, vi-

sual perception performance and cognitive functions

are often focused on through the use of eye track-

ing. Manoeuvre index, useful ﬁled of view (UFOV)

and mini-mental state examination (MMSE) of indi-

vidual drivers was measured, and the contributions of

these to driving performance has been discussed (Ball

and Owsley, 1993; Wood and Owsley, 2014; Mom-

baugh and McIntyre, 1992; Adler et al., 2005). The

factors affecting eye movement during driving have

been discussed and some contributions to the evalu-

ation of individual performance have been examined

(Yamaguchi et al., 2019).

In driving situations, the cognitive workload or at-

tention payment required by drivers for safe motor

vehicle operation has also been measured and dis-

cussed. Most assessments were focused on aspects

of viewing behaviour as mention above, as measure-

ment of cognitive workload or attention level is not

easy during driving, however. The cognitive work-

load is usually measured as overall assessment using

NASA-TLX or other metrics (Hart, 2006). Change

of the cognitive workload or attention may be recog-

nised to affect behavioural reactions. Eye tracking has

been used to assess and analyse attention and view-

ing behaviour (Underwood, 2005; K

ubler et al., 2021;

Table 1: Route segments.

No. Route label No. Route label

1 Straight on campus 11 Pass RoundAbout

2 Pass RoundAbout 12 TurnRight3

3 TurnLeft1 13 Straight two-lane2

4 Straight four-lane1 14 Turn RoundAbout

5 TurnLeft2 15 Straight two-lane3

6 TurnRight1 16 TurnLeft3

7 Straights 17 Straight four-lane2

8 TurnRight2 18 TurnLeft4

9 Pass RoundAbout 19 TurnRight+campus

10 Straight two-lane1

Group 1 [On-campus]:1,2,11,19; M

dur.

= 35.8sec.

Group 2 [Left-turn]:3,5,16,18; M

dur.

= 19.7sec.

Group 3 [Straight]:4,7,9,10,13,15,17; M

dur.

= 51.5sec.

Group 4 [Right-turn]:6,8,12; M

dur.

= 28.0sec.

Group 5 [Turn Roundabout]:14; M

dur.

= 31.1sec.

Hu et al., 2022). These contributions to driving ac-

tions were also extracted from the eye movements of

drivers (Nakayama et al., 2022).

Some modelling techniques can extract latent ac-

tivity such as attention level using a model hy-

pothesised for laboratory-based experiments in order

to conduct temporal change (Ueno and Nakayama,

2021; Dubiel et al., 2023). This technique can be

applied to ocular metrics during driving by introduc-

ing a hypothesised model and restrictions (Nakayama

et al., 2024a; Nakayama et al., 2024b). A more rea-

sonable assessment of the change in attention levels

during driving would require a detailed analyses us-

ing revised models since the model hypothesis was

insufﬁcient in the previous study.

3 METHOD

Both driving behaviour and oculo-motors of older

drivers were measured while they drove their own

cars along the assigned route around the university

campus (Sun et al., 2016a; Sun et al., 2016b; Sun

et al., 2018a).

3.1 Measurements Recorded During

Driving Experiment

In order to measure the above metrics while driv-

ing (Sun et al., 2015; Sun et al., 2018b), 11 older

participants (7 males and 4 females, aged 62 to 76,

mean=67.3) drove the course under experimental con-

ditions (Sun et al., 2016a). Informed consent was ob-

tained from all participants prior to the experiment.

The entire course was divided into 19 separate

segment paths, which are called “routes”, as shown

in Table 1, and these routes were classiﬁed into ﬁve

groups according to driving actions. Simple duration

of driving statistics are summarised in the table below.

Changes in Attention Levels While Driving a Car Estimated Using Modelling Techniques with Features of Oculo-Motors

847

Saccade frequency / Relative pupil size

Pupil size

Saccade frequency

Figure 1: Changes in saccade rate and pupil size across 19

driving routes.

3.2 Experimental Measurements

The targeted data were eye movements including

pupil sizes, and ratings for NASA-TLX as a measure-

ment of cognitive workload, taken after all driving

had been completed.

3.2.1 Oculo-Motor Measurement

Both eye movement and pupil size were measured us-

ing a wearable eye tracker (Arrington, 30Hz) (Sun

et al., 2016a). This equipment can detect saccadic

eye movements in a time-line (Arrington Research,

2016).

Mean temporal changes in the two measured met-

rics (saccade frequency and pupil size) over the 19

routes are summarised in Figure 1, with conﬁdence

intervals of 95%. While observing eye movements

recorded while driving, it was noticed that drivers in

turns or corners rotated their heads before moving

their eyes. Therefore, these vestibulo-ocular reﬂexes

(VOR) were recorded as saccades. Pupil size might be

inﬂuenced by the luminance of the road. As a result,

the averaged metrics remain almost constant over the

entire route.

3.2.2 Cognitive Workload After Driving

A version of NASA-TLX

using six 21-point scales

(0-20) was employed to measure the cognitive work-

load, which consists of Mental demand (MD), Phys-

ical demand (PD), Temporal demand (TD), Perfor-

mance (OP), Effort (EF) and Frustration (FR).

Other metrics were also measured, though there

were no signiﬁcant differences between participants,

as all participants were healthy individuals.

https://humansystems.arc.nasa.gov/groups/tlx/downlo

ads/TLXScale.pdf

4 MODELLING ATTENTION

LEVELS

The estimated attention levels of drivers have been ex-

tracted using a state-space model based on both sac-

cade rates and relative pupil sizes (Dubiel et al., 2023;

Nakayama et al., 2024a; Nakayama et al., 2024b). A

deﬁnition of attention level with minor revisions is as

follows.

4.1 Model Description

Equation (1) introduces an inverse logit function

(inv logit) for six dimensional state changes in routes

(S level consists of 6 dimension) and individual fac-

tors (rID consists of 11 dimensions). However, route

factor (rRoute consists of 19 dimensions) could not

be implemented in an inverse logit function since the

range of the change was too large. This equation

is revised by introducing an inverse logit function

which normalises the factors of individual and tem-

poral changes, in order to emphasise the differences

in factors of the various routes.

The other conditions are the same as those in our

previous reports (Nakayama et al., 2024a; Nakayama

et al., 2024b). Changes in the index of the level of

attention within routes is represented by the 6 states.

Two measured metrics are simulated using base func-

tions together with attention levels. Saccade rates

(Nsac

times

: 0∼) are generated using Poisson distribu-

tions, and pupillary changes (Pupil

size

) around over-

all mean sizes are generated using Gaussian (Normal)

distributions (Dubiel et al., 2023). The validity of

model may be examined by obtaining an optimised

solution.

Attn = inv logit(S level + rPN + rID) + rRoute

(1)

State Model:

S level

∼ Normal(S level

i−1

, σ

)

(2)

Observation Model:

noise

∼ Normal(Attn, σ

noise

)

λ = exp(µ

noise

)

NSac

times

∼ Poisson(λ)

Pupil

size

∼ Normal(Attn, σ

)

4.2 Parameter Estimation

Model parameters were estimated using sampling

based on measured experimental data with the

BIOSIGNALS 2025 - 18th International Conference on Bio-inspired Systems and Signal Processing

848

-10

-5

Source of Attention level

1 2 3 4 5 6

State

Figure 2: Estimated distributions of latent attention in 6

states (S level).

[2]

-0.2

-0.1

0.0

0.1

0.2

[1] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]

Route Number

Estimated velue

Figure 3: Estimated distributions of route parameter

(rRoute).

Markov Chain Monte Carlo (MCMC) method. If the

hypothesised model is appropriate, all parameters can

be estimated to ﬁt with the experimental data. In or-

der to compensate for the data insufﬁciency of 11 par-

ticipants, 7 sets of data of observations were gener-

ated by shifting the observed period by ±1 second in

increments of 0.33 seconds in order to obtain aver-

aged metrics of the 6 states (Nakayama et al., 2024b).

This data extension technique provides 7 times the

data of the original measurements. A sampling using

the Markov Chain Monte Carlo (MCMC) method was

conducted as 4 chains and 6,000 iterations (including

2,000 burn-ins) using the converged condition

R < 1.1

for all parameters.

Distributions of estimated parameters for common

latent activity levels (S level) are illustrated in Figure

2, for route parameters (rRoute) in Figure 3 and for

individual factor parameters (rID) in Figure 4. Using

equation (1), attention levels (Attn) for each of the

routes are summarised in Figure 5. Changes in atten-

tion levels may depend on the parameters of the route

(rRoute), and the 5 route groups are indicated in Fig-

ure 6 using coloured lines. As the ﬁgure shows, the

rID[1]

rID[2]

rID[3]

rID[4]

rID[5]

rID[6]

rID[7]

rID[8]

rID[9]

rID[10]

rID[11]

-2 -1 0 1 2 3

Participant number

Estimated velue

Figure 4: Estimated distributions of individual parameter

(rID).

Level of Attention

Route number

Figure 5: Mean attention levels (Attn) over routes with con-

ﬁdence intervals of 95%.

levels for [3] Straight and [5] Turn RoundAbout are

higher than the others, and the levels for [2] Left-turn

and [4] Right-turn are lower than the others. This sug-

gests that the level of attention for turns made while

driving is the lowest in order to devote these resources

to overall operation of the motor vehicle. In this sce-

nario, the results show that the level of attention de-

creases with the amount of the cognitive workload.

Attention levels paid while driving along each

route are compared using one-way ANOVA with a

factor of the 5 route groups. This route group factor is

statistically signiﬁcant (F(4,1219)=112.6, p < 0.01).

In order to extract the relationship between the groups

of routes, the sub-effect test in Tukey method is ap-

plied. In the results, there are signiﬁcant differences

between [1] On-campus and others, [2] Left-turn and

others except [4] Right-turn, and [4] Right-turn and

[5] Turn RoundAbout. This suggests that the atten-

tion level for turns made while driving is the lowest in

order to devote these resources to overall operation of

the motor vehicle.

However, the results and discussions depend on

the hypothesised model, which is deﬁned as be-

havioural processing, and the validity assessment is

Changes in Attention Levels While Driving a Car Estimated Using Modelling Techniques with Features of Oculo-Motors

849

Route number

Level of Attention

on-Campus

Left-turn

Straight

Right-turn

turn RoundAbout

Figure 6: Mean attention levels (Attn) across route groups.

NASA-TLX score

1:MD 2:PD 3:TD 4:OP 5:EF 6:FR

Figure 7: NASA-TLX score measurements.

not easy to evaluate. Therefore, the results may have

some limitations in explaining the change in workload

under actual driving conditions.

5 MENTAL WORKLOAD

EVALUATION

The results of the assessment of the measurement of

cognitive workload are summarised in a box-plot, as

shown in Figure 7. Individual ratings correlate with

metrics of oculo-motors (Nakayama et al., 2022).

In this section, the contribution of the attention

level to the ratings (Nakayama et al., 2024b), and

the dependency of the estimation models examined.

Correlation coefﬁcients between the ratings and mean

attention levels across all participants in the 5 route

groups are summarised in Figure 8 as a bar graph.

The levels of signiﬁcance (p < 0.05, p < 0.10) are

illustrated in the ﬁgure using dotted lines.

Signiﬁcant coefﬁcients with a factor for Frus-

tration (FR) are conﬁrmed across the 5 groups of

routes. Both coefﬁcients of factors for Performance

(OP) and Effort (EF) show signiﬁcant tendencies (p <

0.10). Therefore, most participants perceived some

of the cognitive workload during the driving experi-

ment. In particular, individual factor (r ID) also cor-

relates signiﬁcantly with the factor rating for Frustra-

tion (FR). Since the attention level is estimated using

equation(1), which consists of a function for individ-

ual factor (rID), the distribution of estimated individ-

p<0.05

p<0.10

on-Campus Left-turn Straight Right-turn turn RoundAbout

Groups of routes

Correlation coefficient

Figure 8: Correlation coefﬁcients between attention levels

and NASA-TLX scores.

TLX1

TLX2

TLX3

TLX4 TLX5 TLX6

Correlation coefficient

Number of state

Figure 9: Correlation coefﬁcients between attention levels

during routes and NASA-TLX scores.

ual factors may affect the ratings.

In addition, the range of the conﬁdence interval

for attention level (Attn) correlates with ratings for

Mental demand (MD). The mean of MD does not cor-

relate directly with attention level. The conﬁdence in-

terval may deviate along the driving route, so it is in-

teresting that the range of the interval correlates with

the ratings for MD.

Regarding changes in correlation coefﬁcients

along with segments of the routes, most coefﬁcients

are maximised slightly during the ﬁnal route driven

(No.19). Once again, these coefﬁcients are almost

completely dependent on the ﬁrst term in equation(1)

including individual factor (rID).

Dependency of change in the 6 states while driv-

ing the routes is examined by comparing the corre-

lation coefﬁcients. The changes in coefﬁcients for

attention level and ratings of cognitive workload are

summarised in Figure 9. As mentioned above, only

one rating for FR is a signiﬁcant coefﬁcient. For

other factors, coefﬁcients change from route to route,

though these coefﬁcients are not signiﬁcant. More de-

tailed analyses including the revision of the model,

will be a subject of our further study.

BIOSIGNALS 2025 - 18th International Conference on Bio-inspired Systems and Signal Processing

850

6 SUMMARY

A procedure for estimating driver’s attention levels

over a driven course was developed using a state-

space modelling technique with saccade rates and

pupillary changes. In order to consider the interac-

tion of the model’s parameters, the attention estima-

tion model was revised. The estimated attention levels

are assessed along the routes driven. As a result, the

estimated attention level decreased during routes with

turns, such as Left-turn and Right-turn, in comparison

with the straight route.

Also, driver’s ratings for cognitive workload, such

as the Frustration factor, correlate with mean atten-

tion levels over all 5 route groups when surveyed after

driving as been completed. Some statistical informa-

tion regarding changes in levels of attention correlate

with some of the ratings for cognitive workload fac-

tors.

Examination of the contribution of route factors to

attention levels will be a subject of our further study.

REFERENCES

Adler, G., Rottunda, S., and Dysken, M. (2005). The older

driver with dementia: an updated literature review.

Journal of safety research, 36(4):399–407.

Arrington Research (2016). ViewPoint EyeTracker User-

Guide.

Ball, K. and Owsley, C. (1993). The useful ﬁeld of view

test: A new technique for evaluating age-related de-

clines in visual function. Journal of the American Op-

tometric Association, 64(1):71–79.

Deng, T., Yan, H., Qin, L., Ngo, T., and Manjunath, B. S.

(2020). How do drivers allocate their potential at-

tention? driving ﬁxation prediction via convolutional

neural networks. IEEE Trans. ITS, 21(5):2146–2154.

Dubiel, M., Nakayama, M., and Wang, X. (2023). Mod-

elling attention levels with ocular responses in a

speech-in-noise recall task. In Proc. ETRA 2023,

pages 89:1–89:7. ACM.

Hart, S. G. (2006). NASA-task load index (NASA-TLX);

20 years later. In Proceedings of the human factors

and Ergonomics Society 50th Annual meeting, pages

904–908.

Hu, Z., Lv, C., Hang, P., Huang, C., and Xing, Y.

(2022). Data-driven estimation of driver attention us-

ing calibration-free eye gaze and scene features. IEEE

Trans. Industorial Electronics, 69(2):1800–1808.

Kapitaniak, B., Walczak, M., Kosobudzki, M., J

zwiak,

Z., and Bortkiewicz, A. (2015). Application of eye-

tracking in drivers testing: a review of research. Inter-

national Journals of Occupational Medicine and En-

vironmental Health, 28(6):941–954.

ubler, T. C., Fuhl, W., Wagner, E., and Kasneci, E. (2021).

55 rides: Attention annotated head and gaze data dur-

ing naturalistic driving. In Spencer, S. N., editor,

Proceedings of ACM Symposium on Eye-Tracking Re-

search & Applications (ETRA 2021), pages 17(1–8),

New York, USA. ACM.

Mombaugh, T. N. and McIntyre, N. J. (1992). The

mini-mental state examination: A comprehensive re-

view. Journal of the American Geriatrics Society,

40(9):922–935.

Nakayama, M., Sun, Q. C., and Xia, J. C. (2022). Es-

timation of older driver’s cognitive performance and

workload using features of eye movement and pupil

response on test routes. In Proc. IV 2022, pages 155–

160. IEEE.

Nakayama, M., Sun, Q. C., and Xia, J. C. (2024a). Car driv-

ing temporal cognitive workload estimation using fea-

tures of eye tracking. In Proc. Eye Tracking Research

& Applications (ETRA) 2024, pages 35:1–3. ACM.

Nakayama, M., Sun, Q. C., and Xia, J. C. (2024b). Using

oculo-motor features to estimate cognitive workload

while driving a test route. The Transactions of Human

Interface Society, 26(3):273–276.

Paeglis, R., Bluss, K., and Atvars, A. (2011). Driving expe-

rience and special skills reﬂected in eye movements.

In Proceedings of SPIE, pages 1–7.

Palazzi, A., Abati, D., Calderara, S., Solera, F., and Cuc-

chiara, R. (2019). Predicting the driver’s focus of at-

tention: The DR(eye)VE project. IEEE Trans. PAMI,

41(7):1720–1733.

Schmitt, K.-U., Seeger, R., Fischer, H., Lanz, C., Muser,

M., Walz, F., and Schwarz, U. (2015). Saccadic eye

movement performance as an indicator of driving abil-

ity in elderly drivers. Swiss Medical Weekly, 145:1–

10.

Sun, Q. C., Xia, J. C., Falkmer, T., and Lee, H. (2016a). In-

vestigating the spatial pattern of older drivers’ eye ﬁx-

ation behaviour and associations with their visual ca-

pacity. Journal of Eye Movement Research, 9(6):2:1–

16.

Sun, Q. C., Xia, J. C., Foster, J., Falkmer, T., and Lee, H.

(2018a). Driving manoeuvre during lane maintenance

in older adults: Associations with neuropsychological

scores. Transportation Research Part F, 53:117–129.

Sun, Q. C., Xia, J. C., Foster, J., Falkmer, T., and Lee, H.

(2018b). A psycho-Geoinformatics approach for in-

vestigating older adults’ driving behaviours and un-

derlying cognitive mechanisms. European Transport

Research Review, 10(2)(36):1–11.

Sun, Q. C., Xia, J. C., He, J., Foster, J., Falkmer, T., and Lee,

H. (2018c). Towards unpacking older drivers’ visual-

motor coodination: A gaze-based integrated driving

assessment. Accident Analysis & Prevention, 113:85–

96.

Sun, Q. C., Xia, J. C., Nadarajah, N., Falkmer, T., and Lee,

H. (2016b). Assessing drivers’ visual-motor coordi-

nation using eye tracking, GNSS and GIS: a spatial

turn in driving psychology. Journal of Spatial Science,

61(2):299–316.

Sun, Q. C., Xia, J. C., Nadarajah, N., Foster, J., and

Lee, H. (2015). Positioning eye ﬁxation and vehi-

cle movement: Visual-motor coordination assessment

Changes in Attention Levels While Driving a Car Estimated Using Modelling Techniques with Features of Oculo-Motors

851

in naturalistic driving. In Veenendaal, B. and Kealy,

A., editors, Research@Locate15, pages 90–99. CEUR

Workshop Proceedings. http://ceur-ws.org.

Ueno, T. and Nakayama, M. (2021). Estimation of visual at-

tention using microsaccades in response to vibrations

in the peripheral ﬁeld of vision. In Proceedings of

ETRA2021, pages 1–6. ACM.

Underwood, G. (2005). Cognitive Processes in Eye Gudid-

ance. Oxford University Press, New York, USA.

Wood, J. M. and Owsley, C. (2014). Gerontology view-

point: Useful ﬁeld of view test. Gerontology,

60(4):315–318.

Yamaguchi, K., Nakayama, M., Sun, Q., and Xia, J. (2019).

Prediction of cognitive performance of drivers using

eye ﬁxation behaviours. In Proceedings of 23rd Inter-

national Conference Information Visualisation, pages

186–190, London, UK.

Yamani, Y., Horrey, W. J., Liang, Y., and Fisher, D. L.

(2016). Age-related differences in vehicle control

and eye movement patterns at intersections: older and

middle-aged drivers. PLOS ONE, 11:1–17.

BIOSIGNALS 2025 - 18th International Conference on Bio-inspired Systems and Signal Processing

852