Improved Method for fatigue detection

Author: Ms Ticiana Capris STSM Period: 2021-06-07 – 2021-06-21


Hosting institution: Ss. Cyril and Methodius University

From ITC: Yes


The main objective of this STSM was to develop methods for monitoring physical fatigue. In particular, we analyze previously acquired data related to mobile sensors (eg accelerometer, gyroscope, magnetometer, GPS receiver and microphone), external sensors (electrocardiography sensors) and mobile sensors (eg temperature sensors) and apply these data and solve the fatigue detection problem. The new methods can help in the particular care of each subject, and to measure the health status of the subjects, identifying possible diseases and risks with machine learning techniques. The motivation for this kind of study is to concatenate programming knowledge with health knowledge and thus help people in general and even prevent overwork at the lowest cost to the end user. In order to obtain a new experience and share ideas with other people, I stayed the period of this short-term-scientific mission (STSM) under the supervision of Dr. Eftim Zdravevski at Ss. Cyril and Methodius University, Skopje, Macedonia, to share experiences and learn with the people in the university about the data processing techniques, including data cleaning and data imputation methods, in order to prepare the sensors’ data in good condition for the correct methodology for monitoring physical fatigue.

The second goal of STSM is to initiate and promote cooperation between the two institutions. Based on the results of our scoping survey, we are very confident that this will produce academic publications that may appear in project proposals.


During my research visit, I analyzed previously obtained information related to mobile sensors (such as accelerometers, gyroscopes, magnetometers, GPS receivers, and microphones), external sensors (ECG sensors), and mobile sensors (such as temperature sensors),and applied this data to try solve the fatigue detection problem.

During STSM, the following activities were carried out:
1. Analyzed the data received from the sensors: accelerometer, gyroscope and magnetometer sensors, electrocardiogram (ECG) sensor from one Bitalino device and temperature sensor from another Bitalino device and perform data cleaning and preprocessing.
2. Researched and developed improved methods for identifying fatigue based on existing literature.
3. Based on the research results, the software architecture of the fatigue recognition solution.
4. Investigated the possibility of applying the same method or similar methods to other physiological conditions.

During this STSM, the data imputation methods were developed and a paper with the results should be published in the future. After visualizing and interpreting the results, we discussed findings and trends, and brainstormed ideas for further improvements.


This chapter describes the results obtained during the STSM, and also introduces and details the indicators used to evaluate them. In order to evaluate the performance of the proposed method and the obtained classification model, the following four metrics are used:

where TN, TP, FP and FN stand for True Negative, True Positive, False Positive and False Negative, respectively, and

During the training and test phases, the model is tested to categorise each one of the information samples that are accessible in that step, that is, after any epoch of training the model is tested to search its capability for correctly categorise all the samples. After trying to categorise all the samples accessible (only the direct samples during the training phase, and only the test samples during the test phase), it’s measured the loss, which is a measurement of the disagreement between the classification delivered by the exemplar and its right class. This way, Validation Loss consults to the loss charge on each one of the training epochs, and Loss is the loss price after the testing phase.

The results obtained are focused in the objectives for the COST Action CA 16226 related to methods for monitoring physical fatigue. All of the results defined in the Work Plan are achieved with success and other research studies are performed during my visit.
The main results achieved during the COST Action “Indoor living space improvement: Smart Habitat for the Elderly” are:
• The knowledge obtained with new data processing techniques used at the Host Institute;
• Creation of a method for the imputation of the data acquired from the accelerometer, gyroscope,magnetometer sensors and external sensors.


Since the STSM is not long enough to completely execute the second deliverable, the STSM will only initiate the work needed for a high-quality publication and the actual writing of the publication will continue after the STSM.