A Study of Psychophysical Factor (Heart Rate) for Driver Fatigue Using Regression Model
Key Findings
- A regression model was successfully developed and validated for predicting heart rate changes as a marker of driver fatigue
- Model validation runs fell within the 90% prediction interval, with residual errors less than 10% compared to predicted values
- Heart rate was significantly influenced by three parameters: time exposure (driving duration), type of road, and driver gender
- The study provides a quantitative framework for predicting fatigue onset based on psychophysical monitoring
Background
Driver fatigue is consistently identified as one of the leading contributors to road traffic accidents worldwide. Fatigued driving impairs reaction time, reduces vigilance, degrades decision-making ability, and increases the likelihood of microsleep episodes—brief, involuntary periods of inattention that can have catastrophic consequences at highway speeds. Despite the well-documented risks, there are currently no laws in many jurisdictions specifically regulating driver fatigue, unlike regulations governing alcohol-impaired driving.
In Malaysia, road traffic accidents represent a significant public health burden. The country has one of the highest traffic fatality rates in Southeast Asia, with driver fatigue contributing to a substantial proportion of crashes, particularly on long-distance highway routes. Understanding the physiological mechanisms underlying fatigue and developing predictive models for fatigue onset are essential steps toward implementing effective prevention strategies.
Heart rate is a psychophysical measure that reflects the autonomic nervous system’s response to physiological and psychological stressors, including fatigue. As fatigue develops during prolonged driving, characteristic changes in heart rate and heart rate variability occur, reflecting shifts in sympathetic-parasympathetic balance. This study aimed to develop a regression model that could predict heart rate changes—as a proxy for fatigue development—based on driving conditions and driver characteristics.
Study Design and Methods
The study recruited ten subjects who participated in driving sessions under controlled conditions. Heart rate was continuously monitored and recorded using a Polar Watch heart rate monitor (measuring beats per minute). The experimental design examined the influence of several input parameters on heart rate as the output response variable.
The key input parameters investigated were time exposure (duration of continuous driving), type of road (representing different driving demands and complexity levels), and driver gender. The experimental design enabled systematic variation of these parameters to quantify their individual and combined effects on heart rate.
Design Expert 8.0.6 software was used for the regression analysis. The modelling approach aimed to establish a mathematical relationship between the input parameters and heart rate output that could predict fatigue-related heart rate changes under specified driving conditions. Model validation was performed by comparing predicted values with observed heart rate measurements from additional validation runs.
Results
Model Development
The regression model was successfully developed, establishing a quantitative relationship between the input parameters (time exposure, road type, and gender) and heart rate changes associated with driving fatigue. The model coefficients reflected the relative contribution of each parameter to heart rate variation during driving.
Model Validation
The validation process demonstrated strong predictive performance. All validation runs fell within the 90% prediction interval of the developed model, indicating that the model’s predictions were consistent with observed physiological responses. Residual errors between predicted and actual heart rate values were less than 10%, meeting the pre-specified accuracy criterion for model acceptability.
| Validation Criterion | Result | Threshold |
|---|---|---|
| Prediction interval coverage | All validation runs within 90% PI | Within 90% PI |
| Residual error | < 10% | < 10% |
| Model significance | Significant | p < 0.05 |
Significant Parameters
The analysis identified three parameters that significantly influenced heart rate during driving:
Time Exposure: Driving duration showed a significant effect on heart rate, with characteristic patterns of change as fatigue accumulated over time. This finding is consistent with the established understanding that prolonged driving leads to progressive fatigue and measurable physiological changes in cardiovascular function.
Type of Road: Different road types—with varying demands for attention, steering input, and speed regulation—produced significantly different heart rate profiles. More complex road environments that require greater cognitive processing and motor coordination were associated with distinct heart rate patterns compared to monotonous straight road driving.
Gender: Gender differences in heart rate response to driving fatigue were statistically significant. This finding aligns with known sex-based differences in cardiovascular physiology and autonomic nervous system regulation, and suggests that fatigue detection algorithms may need to account for gender-specific thresholds.
Discussion
The successful development of a validated regression model for predicting fatigue-related heart rate changes represents a meaningful contribution to the field of driver fatigue research. The model’s ability to account for driving duration, road complexity, and driver gender provides a multifactorial framework that more accurately reflects real-world driving conditions than single-parameter approaches.
The significance of road type as a predictor has practical implications for fatigue management. Monotonous driving environments (such as long, straight highways) may produce different fatigue patterns compared to complex urban driving, and prevention strategies should account for these differences. Similarly, the gender effect suggests that one-size-fits-all fatigue detection thresholds may be suboptimal.
Heart rate monitoring using wearable devices is increasingly feasible for real-world driving applications. Modern smartwatches and fitness trackers provide continuous heart rate data that could potentially be integrated with in-vehicle fatigue warning systems. The regression model developed in this study could serve as the algorithmic basis for such applications, translating raw heart rate data into fatigue risk estimates.
Public Health Implications
The findings from this study have direct applications for road safety policy and technology development in Malaysia and beyond. In-vehicle fatigue monitoring systems that incorporate heart rate data alongside other physiological and behavioural indicators (such as steering patterns and eye tracking) could provide real-time fatigue warnings to drivers, potentially preventing fatigue-related accidents.
Transportation authorities could use the model’s insights to inform regulations regarding maximum continuous driving durations, mandatory rest breaks, and driver scheduling practices for commercial transport operators. The gender-specific findings suggest that fatigue management guidelines should consider physiological differences between male and female drivers rather than applying uniform standards. Additionally, road design and infrastructure planning could consider the fatigue-inducing properties of different road types, incorporating features such as rest areas, rumble strips, and environmental variety to mitigate monotony-related fatigue on long-distance routes.
Limitations
The small sample size (n = 10) is a significant limitation that affects the generalisability and statistical power of the findings. The controlled experimental conditions, while necessary for systematic parameter variation, may not fully replicate the complexity and unpredictability of real-world driving environments. The study did not account for potentially important confounders such as sleep quality before driving sessions, caffeine intake, ambient temperature, or vehicle characteristics. Heart rate alone, while informative, provides a limited window into the multifaceted phenomenon of driver fatigue; future studies should integrate multiple physiological measures including heart rate variability, electrodermal activity, and ocular measures. Larger-scale validation studies under naturalistic driving conditions are needed before the model can be implemented in practical fatigue detection systems.
Ismail NK, Deros BM, Nuawi MZ. A Study of Psychophysical Factor (Heart Rate) for Driver Fatigue Using Regression Model. Malaysian Journal of Public Health Medicine. 2018; Special Volume (2).
License: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)