TECH4DYNAMICS

The assessment of motor function has not yet reached its full potential, either because qualitative analysis methodologies are used, namely specific questionnaires, which, despite providing important data, depend on the experience of the observer and the limited amount and accuracy of the information they provide; because of the costs associated with more advanced methodologies, which require greater expense and specialization in analysis, representing an additional barrier to obtaining a complete understanding of motor function; or because of the time needed to collect and obtain the scores, which in some cases, when the result is released, the subjects are no longer in the same condition as when they were evaluated. In response to this kind of issues, in recent years, portable technology has sought to play a more active role in the assessment of motor functions, both through wearables or specialized equipment. Bates and Sunderam (2023), Beswick et al. (2022), Bortolani et al. (2022), and Sica et al. (2021) conducted systematic reviews that highlight the different applications that this type of technology has had in movement analysis, demonstrating the growing importance of equipment development. Nevertheless, these publications still point to a lack of validation and refer mainly to populations with specific neural conditions, leaving open the need to promote further studies in this area. Inertial Measurement Units (IMUs) are low-cost sensors with wide applicability in various contexts (Bortolani et al., 2022; Camomilla et al., 2018). A significant portion of current portable technology has incorporated these sensors, allowing the transfer of the applicability of specific IMU-based equipment to devices used in everyday life (e.g., smartphones, smartwatches). This also enables real-time analysis of motor function, which could potentially alert users, for instance, to consult their doctor.

The project mission consists of develop and implement IMU-based methods for the assessment of motor function across different age groups and health conditions, especially in neurodegenerative diseases. The collection of continuous data in fine and gross motor skills and the testing of its validity associated with non-linear analysis methods opens up new frontiers in the understanding of complex human movement patterns, allowing for a more precise and personalized assessment of motor functions. In addition, this innovative approach by incorporating artificial intelligence (AI) and deep learning models into the analysis of motor function data, makes it possible to develop more effective interventions and technologies for different populations, promoting advances in the optimization of physical performance, health and rehabilitation, with applications in exercise, clinical and educational contexts. AI-powered systems can automatically analyze large datasets captured from IMUs and other wearable sensors, identifying subtle patterns and trends that may not be discernible through traditional analysis. Deep learning models, particularly convolutional neural networks (CNNs) and recurrent neural networks (RNNs), can be used to detect and classify different types of motor activities or dysfunctions by processing complex time-series data (Cheriet et al., 2023). These models not only enhance the accuracy of the analysis but also enable predictive insights, such as forecasting the progression of neurodegenerative diseases based on early motor function signals (Patel et al., 2021; Sigcha et al., 2023). Furthermore, the project will utilize AI models for the personalization of assessments. By continuously learning from individual data over time, the system can adapt to the user's unique motor patterns, thus providing highly tailored recommendations for movement improvement or rehabilitation. Another innovative layer that AI can provide is the real-time feedback mechanism, where users can receive instant guidance or alerts if abnormal motor patterns are detected, which is critical in clinical settings for early intervention.

Keywords: Inertial Sensors; Artificial Intelligence; Deep Learning; Non-Linear analysis; Motor Function Assessment; Neurodegenerative disorders

General Objective:

• The general objective of this project is to develop and implement methods based on Inertial Measurement Units and apply deep learning models to enhance the assessment and monitoring of motor function across diverse populations, facilitating accurate classification of motor dysfunctions and enabling predictions about disease progression, leading to earlier and more targeted interventions.

Specific objectives

• Real-Time Data Monitoring: Develop systems for real-time collection of movement data in field and/or laboratory settings to analyze the subject's current state, with or without disorders, and the effect of interventions such as physical exercise, medication adjustments or rehabilitation.
• IMU-Based Early Detection and Monitoring: Concentrate on high-precision inertial sensors to detect motor dysfunction, focusing on early diagnostic tools in subject's with or without pre-diagnosed conditions like Parkinson’s, Alzheimer’s, and multiple sclerosis.
• Deep Learning Methods for Motor Pattern Detection: Integrate non-linear methods and advanced deep learning algorithms, such as convolutional neural networks and recurrent neural networks to analyze complex time-series data collected from IMUs. These models are trained to automatically detect subtle motor patterns and dysfunctions that may not be easily identified through traditional analysis. For instance, CNNs can extract spatial features from movement data, while RNNs can capture temporal dependencies, enabling the system to accurately classify motor dysfunctions and predict the progression of neurodegenerative.
• Validation in Special Populations: The study would prioritize sensor accuracy, reliability, and applicability in fine and gross motor skills, emphasizing the validation of technology for practical, real-world applications.

References

1. Brígida, N., Catela, D., Mercê, C., & Branco, M. (2024). Predictability and Complexity of Fine and Gross Motor Skills in Fibromyalgia Patients: A Pilot Study. Sports, 12(4), 90. https://doi.org/10.3390/sports12040090
2. Brígida, N., Catela, D., Mercê, C., & Branco, M. (2024). Variability of gross and fine motor control in different tasks in fibromyalgia patients. Retos, 54, 151-158. https://doi.org/10.47197/retos.v54.103134
3. Miranda-Oliveira, P., Branco, M., & Fernandes, O. (2023). Accuracy and Interpretation of the Acceleration from an Inertial Measurement Unit When Applied to the Sprint Performance of Track and Field Athletes. Sensors, 23(4), 1761. https://doi.org/10.3390/s23041761
4. Mercê, C., Cordovil, R., Catela, D., Galdino, F., Bernardino, M., Altenburg, M., António, G., Brígida, N., & Branco, M. (2022). Learning to Cycle: Is Velocity a Control Parameter for Children’s Cycle Patterns on the Balance Bike? Children, 9(12), 1937. https://doi.org/10.3390/children9121937
5. Altenburg, M., Farinha, C., Santos, C., Mercê, C., Catela, D., & Branco, B. (2021). Analysis of Motor Behavior based on Recurrence Analysis in Adults with Autism Spectrum and Neurotypicals in a Dynamic Balance Task: a pilot study. Cuadernos de Psicología del Deporte, 21, 233-242. http://scielo.isciii.es/scielo.php?script=sci_arttext&pid=S1578-84232021000300018&nrm=iso
6. Pereira, T. O., Abbasi, M., & Arrais, J. P. (2023). Enhancing reinforcement learning for de novo molecular design applying self-attention mechanisms. Briefings in Bioinformatics, 24(6), bbad368. https://doi.org/10.1093/bib/bbad368
7. Martins, D., Abbasi, M., Egas, C., & Arrais, J. P. (2024, 2024/09/01/). Enhancing schizophrenia phenotype prediction from genotype data through knowledge-driven deep neural network models. Genomics, 116(5), 110910. https://doi.org/https://doi.org/10.1016/j.ygeno.2024.110910

Project Coordination:

• Marco Branco, Principal Investigator
• Cristiana Mercê, Co-Principal Investigator
• Guilherme Furtado, Co-Principal Investigator

Research Team:

• David Catela, PhD
• Maryam Abbasi, PhD
• Pedro Sobreiro, PhD
• Bruno Silva, PhD

MSc/PhD students:

• Nancy Brígida, PhD student
• Mónica Sousa, PhD student
• Mafalda Bernardino, MSc student
• Pedro Caetano Raposo, MSc student
• Marta Santos, MSc Student

Related projects:

• Analysis of the Complexity and Variability of Fine and Gross Motor Tasks in Fibromyalgia Patients
• Assessment of Motor Coordination in Fine and Gross Motor Tasks in Parkinson's Patients
• Biomechanics' Contribution to Improving Sports Performance in Athletics: Development and Validation of a Dedicated Inertial Measurement Unit

Organizações Apoio/Sponsors

• Wisify Tech
• Sensing Future (https://sensingfuture.com/en/)
• INOPOL (https://inopol.ipc.pt)
• Novo Corpo