Driving has become essential in transporting people from one place to another. However, prolonged driving could cause muscle fatigue, leading to drowsiness and microsleep. Electromyography (EMG) is an important type of electro-psychological signal that is used to measure electrical activity in muscles. This work classifies and predicts muscle fatigue from trapezius muscle of 10 healthy subjects. The EMG signals and the time when muscle fatigue was experienced by the subjects were recorded. The mean frequency and median frequency of the EMG signals were extracted. For classification of muscle fatigue in non-fatigue and fatigue condition, six machine learning models were used: Logistic Regression, Support Vector Machine, Naïve Bayes, k-nearest Neighbour, Decision Tree and Random Forest. From the value of median frequency and slope coefficient of median frequency, mathematical model was developed with respect to driver’s physical factors. The results show that both the median and mean frequency are lower when fatigue conditions exist. In term of the classification performance, the highest accuracy for classifying muscle fatigue due to prolonged driving was obtained by the Random Forest classifier with 85.00%, using both the median and mean frequency of the EMG signals. This method of using the mean and median frequency will be useful in classifying driver’s non-fatigue and fatigue conditions and predict muscle fatigue during prolonged driving. The significant factor influencing muscle fatigue of the driver was Body Mass Index (BMI). This study successfully developed mathematical model of second order polynomial of muscle fatigue and BMI (p<0.05 and the R2 = 0.85). The model was successfully validated where the residual errors compared between predicted values and actual values were less than 10%.

In the case of amputees, the development of cybernetic hands that closely resemble the functions of real hands is essential for comfort and functionality purposes. Controlled by intrinsic and extrinsic muscles, the human thumb plays a major role in differentiating hand gestures. For those who have lost their intrinsic hand muscles, any information about muscle activities that can be obtained from the extrinsic muscles is essential to control the thumb. Thus, focusing on transradial amputees, this research investigates the relationship between extrinsic muscles to characterise thumb posture. A High-Density surface Electromyogram (HD-sEMG) device and a portable thumb force measurement system were used to collect forearm HD-sEMG signals from a total of 17 subjects. For the flexion motion, the subjects were asked to repetitively place their thumb at rest before exerting 30% of their individual maximum voluntary contraction (MVC) on a load cell by following a designated trajectory presented on a developed graphical user interface (GUI). The measurement system was set to four different postures namely zero degrees, thirty degrees, sixty degrees, and ninety degrees. Feature extraction was then performed by extracting the absolute rectified value (ARV), root mean square (RMS), mean frequency (MNF) and median frequency (MDF) values of the forearm HD-sEMG signals before being classified using four different classifiers namely linear discriminant analysis (LDA), support vector machine (SVM), k-Nearest Neighbour (KNN), and TREE-based classifier. The results revealed that the LDA classified RMS and ARV-RMS features, which were extracted from both posterior and anterior hand sides successfully achieved the highest correctly classified percentage of 99.7%. The findings of the study are significant for the development of a dedicated model-based control framework for prosthesis hand development to be used by transradial amputees in the near future.