Philips Healthcare is market leader in Interventional X-ray equipment. Hospitals are more and more managed as regular businesses; therefore, fast and reliable service of this equipment is crucial. The main indicator for measuring serviceability is Elapsed Time to Repair (ETTR), i.e. the time from the customer's call indicating a problem until the moment that the problem is resolved. The business objective of this project is to reduce the ETTR and therefore lower the service costs and improve the service quality, by focusing on diagnosability of the systems. The Design for Diagnostics (D4D) team in Philips Healthcare is developing and deploying a method that takes serviceability structurally into account during the design phase of a system and provides a way of diagnosing a failure in the system. However, this method has several severe constraints that limit its practical side and scalability. The goal of this project was to create a prototype that provides a solution for overcoming those problems. This solution should fulfill the following goals: ? Compositionality - to broaden the scope of performing diagnosis from group of components to system level. ? Configurability - to have one content base that will support different configurations of the systems, both for hardware and software. ? Optimized diagnostics - to improve the diagnostics by optimizing the process of identifying a failure in the system (to increase the accuracy of the diagnostics (avoid human mistakes) and to reduce the costs at the same time). A huge amount of diagnostic data is available for the iXR systems in Philips Healthcare. This project serves as a research study for the D4D team to learn what possibilities exist for the diagnostic data they have available for the iXR systems. The solution explained in this report is one possible way to improve the diagnostics on these systems and therefore lower the costs for service. The main challenge in this project is to design a data model of the iXR systems that could capture different configurations of the iXR systems and compose diagnostic data to system level. The support for different configurations and compositionality of diagnostic models is achieved by making a clear separation between the physical and the logical (diagnostic) model of the system. The optimized diagnostics are accomplished by creating a Bayesian Network from the diagnostic model of the system. For this purpose the SMILE library was used. A tool with graphical user interface (Diagnostic Reasoner) was implemented to show how the new D4D method can be applied, and prove that it solves the problems that were present by using the old D4D method. However, tools for system health management already exist on the market. We performed a thorough validation and verification of our prototype, as well as comparison to one of those tools. This process brought us to the conclusion that this project requires a lot of investment in order to bring the maturity of our tool to a level that can be applied on the systems in the field. Therefore, towards the end of the project Philips started running a process for supplier selection. This project's work helped us to deepen our insight in the requirements, design, and algorithms of a Diagnostic Reasoner, and especially a modeling tool for capturing the contents for the Diagnostic Reasoner. We learnt about the opportunities provided by the current diagnostic designs of the systems and we also realized the limitations. These insights help the D4D team to be sharp and well prepared in the supplier selection process, which is of significant important for improving the service of the medical systems in Philips Healthcare.
|Award date||4 Oct 2012|
|Place of Publication||Eindhoven|
|Publication status||Published - 2012|