Location tracking: technology, methodology and applications Marina L. Gavrilova SPARCS Laboratory Co-Director Associate Professor University of Calgary
Interests and affiliations SPARCS Lab Co-Founder and Director BT Lab Co-Founder and Director Computational Geometry and Applications Founder and Chair since 2001 ICCSA Conference series Scientific Chair (since 2003) Transactions on Computational Science Journal Springer Editor-in-Chief Research topics: optimization, reliability, geometric algorithms, data structures representation and visualization, GIS, spatial analysis, biometric modeling
Talk outline 1. Calgary Health Region RTLS Competition 2. Medical personnel tracking project description 3. Methodologies 4. Outcomes
CHR RTLS Competition Set-up CHR RTLS Call 7 large vendors responded Combination of software and hardware provider Evaluated by a committee for 3 months Decision made to approach a selected vendor for a limited trial Parameters Hardware, medical equipment, personnel Laptops, PDAs, and wireless devices Room-level accuracy Soft or hard thresholds Transmission coverage Security alerts
CHR RTLS Competition Criteria General Overall Vision for RTLS WiFi / RTLS experience Case Studies Provided Trial possibility Software Hardware Software Integration tools Security measures Adjustable for local settings Support/upgrades AutoCad, 802.11 devices Physical characteristics (size, weight) Adjustable Radio Frequency Battery replacement Functionality
Research methodology: four phases Four phases of research: 1. Wi-Fi RFID Technology purchase and integration (initial trial phase: 20 units) at W21 site. 2. Initial data collection (over 15-day period) and validation through independent observers. 3. Tracking of temporal-spatial data related to nurses and MDs using RFID technology: location in time of doctors and nurses, contact with patients, use of medical devices, use of computers, use of hand-washing facilities, etc. 4. Location tracking of specific procedures where time involved in patient care can be more efficiently utilized, such as hand-washing behavior
Research methodology: proposed approaches Innovative approaches to data analysis and visualization developed in SPARCS Lab: Use of sophisticated topology-based methods for data representation and analysis (such as clustering, path planning, risk analysis, dependencies trends) Use of hierarchical weighted tree-based data structure with varied LOD (level of detail) for fast search and dynamic data updates Utilization of recently developed spatial analysis tools (autocorrelation, regression) for analysis of spatio-temporal trends and patterns Utilization of adaptive methods for data visualization (to improve space and time efficiency); Use of advanced interface design methods for improved visual reports and easy decision-making
Example: topology-based data structures to store information Voronoi Diagram Raster Method Potential Method
Interpolation Engine Cleanup Operation Error Metric Refine G Q Example: adaptive tree-based data structure DE M Wavelet Error Analysis Triangle Quad Tree Data structure Render
Example: converting Height field data into 3D topological mesh Pixel value (z) is used as Height Map Vertices are generated as points in 3D A Mesh is triangulated 200 255 150 100 100 255 255 200 200 150 200 100
Example 3D data visualization using adaptive LOD Marina L. Gavrilova
Example: Risk Analysis using Spatial Neighborhood Properties and Clustering Methods INCIDENTS DELAUNAY TRIANGULATION CLUSTERS SHIP ROUTE INTERSECTIONS CLUSTERING OF HIGH-RISK AREAS REDUCED VISIBILITY-GRAPH MINIMIZING RISK AT SEA Priyadarshi Bhattacharya and Marina Gavrilova, SPARCS Lab, Department of Computer Science, University of Calgary e-mail: {pbhattac, marina}@cpsc.ucalgary.ca
Example: Clustering and data filtering Original dataset Crystal output (Th = 2.5) Original dataset Crystal output (Th = 2.4)
Example: Path planning and risk avoidance Clearance = 12 Clearance = 7 Clearance = 8 Clearance = 0 Clearance = 0 Clearance = 0
Example: Path planning with constraints and multiple overlays Path follows shipping lanes wherever possible
Example: Spatio-temporal data analysis and visualization Average Tonnage of Tracks in each Grid Cell Average Tonnage of Incidents Average track counts Accident point counts
Research methodology: expected outcomes Expected outcomes: Knowledge outcomes, where research will produce new knowledge that is relevant to decision-making and policy-setting in health care; Improved patient-centred outcomes, particularly as a result of research that relates to the patient experience; Enhancement of processes in the complex clinical environment, which in turn will produce improvements in outcomes such as provider well-being, patient satisfaction, and improved patient-care policies; Cost and time saving outcomes Efficient resource utilization outcomes.
Questions? E-mail. marina@cpsc.ucalgary.ca Web www.cpsc.ucalgary.ca/~marina