The project's application of dynamic belief network theory to ambulation monitoring and fall detection resulted in a number of successful developments. These included a substantial new software system, design and implementation of Vermahide-based pressure transducers, development and detailed investigation of numerous DBN models capable of extracting `high-level' status information from raw walking data, and investigation into the applications of the Compumedics Siesta datalogger.
Extensive research, results from previous and related projects, and assistance from supervisors and academics provided the project with the resources required to successfully meet its objectives. The applicability of DBN theory to ambulation monitoring and fall detection was investigated with positive results, especially in relation to the later DBN models.
The desired improvements and extensions over previous projects were achieved. These included implementation of a configurable generic DBN software system capable of processing data from a number of sources and sensors using specifiable DBN models and feature extraction configurations. Additionally, a graphical user interface was produced, testing and application of multiple sensors was performed, and testing of a number of DBN ambulation processing models showed that the system was capable of recognising different ambulation modes as well as stumbles and falls.