Research and Publications

Pedestrian behaviour modelling in large urban networks

Pedestrian movement studies are focusing on modelling a range of phenomena starting from pedestrian flow to individual behaviours. Key aspects of each model include understanding of activity agenda and pedestrian-environment interaction. Unlike vehicles, pedestrians are more adaptable to fast decision changes, such as route and destination choices. Questions of interest include phenomena such as congestion, following psychology and street-space perception and are important to understanding evacuation and travel dynamics. Of interest in this study is analysing the impact of common urban layouts in context of pedestrian movement decisions, when various behaviours are taken into account.

The agent-based paradigm (ABM) is used as appropriate modelling tool to describe randomness and individuality of agents in typical groups. ABM helps to overcome main limitations of using Geospatial Information Systems, which are static in nature and unable to represent network changes over time, with "agents" being individuals with their own characteristics and attributes. Agents interact with both the environment and one another to represent human decision behaviour that gives rise to the phenomena described above.

Open Street Map (OSM) is an open, crowdsourced, database of geographical information (including streets, buildings, public transport routes etc.). In the current work, OSM is used to source spatial environment data, in particular of complex urban layouts including street size, interconnection degrees and destination venues of interest.


Image: Extracting flow data from pedestrian movement.


Stochastic methods for modelling Drug Delivery Systems (DDS)

(in collaboration with Sigmoid Pharma Ltd.)

As modern drug formulations become more advanced, pharmaceutical companies face the need for adequate tools to permit them to model complex requirements and to reduce unnecessary adsorption rates while increasing the dosage administered.

The aim of this research is the development and application of a general stochastic framework with agent-based elements for building drug dissolution models. The project utilises three dimensional Cellular Automata and Monte Carlo methods, to describe structural compositions and the main physico-chemical mechanisms, such as diffusion, polymer degradation and hydration. This allows tailoring of in vitro drug development processes for bespoke medical treatments providing accurate insight on controlled delivery for different drug/excipient formulations.



Image: Visualisation of the Cellular Automata simulation for DDS.




Publications

Bezbradica, M. and Vasic, J. (2015), Cellular Automata and Agent Based Modelling: Terminology, Boundaries, Interpretation, to be submitted in Journal of Computational Physics.

Bezbradica, M., Crane, M. and Ruskin, H.J. (2015), Applications of High Performance Algorithms to Large Scale Cellular Automata Frameworks used in Pharmaceutical Modelling, Journal of Cellular Automata - to appear in the 2nd quarter 2015.

Bezbradica, M. and Ruskin, H.J.(2014), Modelling Impact of Morphological Urban Structure and Cognitive Behaviour on Pedestrian Flows, Murgante, B. et al. (eds.). ICCSA 2014. Springer, LNCS 8582. pp. 268-283. DOI: 10.1007/978-3-319-09147-1_20.

Bezbradica, M., Ruskin, H.J. and Crane, M. (2014), Probabilistic Pharmaceutical Modelling: A Comparison Between Synchronous and Asynchronous Cellular Automata, Parallel Processing and Applied Mathematics; Wyrzykowski, R., Dongarra, J., Karczewski, K. and Wasniewski, J. (eds.) PPAM 2013. LNCS 8385. Heidelberg: Springer, pp 699-710.

Bezbradica, M., Ruskin, H.J. and Crane, M. (2014), Comparative Analysis of Asynchronous Cellular Automata in Stochastic Pharmaceutical Modelling, Journal of Computational Science, Vol.5 (5), pp. 834-840, DOI:10.1016/j.jocs.2014.04.010.

Perrin, D., Bezbradica, M., Crane, M., Ruskin, H.J. and Duhamel, C., High-Performance Computing for Data Analytics, 2012 IEEE/ACM 16th International Symposium on Distributed Simulation and Real Time Applications (DS-RT), Dublin, Ireland, 2012, pp. 234 - 242.

Bezbradica, M., Crane, M., and Ruskin,H. J., Parallelisation strategies for large scale cellular automata frameworks in pharmaceutical modelling, In the Proceedings of 2012 International Conference on High Performance Computing and Simulation (HPCS2012), Madrid, Spain, 2nd - 6th July 2012. Selected as one of the top 6 HPCS 2012 Outstanding Papers.

Bezbradica, M., Ruskin, H.J., and Crane, M., Modelling drug coatings: A parallel cellular automata model of ethylcellulose-coated microspheres, Proceedings of The International Conference on Bioscience, Biochemistry and Bioinformatics, ICBBB 2011, Vol.5, pp.419-424, 2011.

Bezbradica, M., Crane, M., and Ruskin, H.J., Probabilistic modelling of drug dissolution, ERCIM News, 84(42/43), January 2011.

Bezbradica, M., Ruskin, H.J., and Crane, M. Probabilistic models for dissolution of ethylcellulose coated microspheres, Proceedings of The European Simulation and Modelling conference, ESM'2010, Hasselt, Belgium, pages 408-412, October, 2010.

Posters and Presentations

For details, visit list of presentations at Mod-Sci website.