The semantic knowledge management tool

Created on Thursday, 24 November 2016

The semantic knowledge management tool developed by VTT has multiple points of action, in all areas where semantic knowledge management support is required. As explained above, it enables the domain analysis and data mining towards the content interpretation, search and inter-linking for relevant suggestions, both based on historical data, as well as on system reasoning from the acquisition of new data. For example, the tool is applied when type of emergency is entered and if it recognises specific situations, it may provide suggestions as to what to expect, e.g. type of injuries from a given type of emergency. It could go so far as to also suggest to alert a specific hospital, which has this specialty or to call in a specific expert, e.g. CBRNE or poison centre expert, etc.

It acts in a similar way when hazards are entered, e.g. it provides various warnings and instructions for safety measures. Similarly, it can provide an interpretation of weather data and relevant warnings. Furthermore, the tool helps with suggestions for types of beds when referral diagnosis is entered. It also supports the correct allocation of tagged information. In addition, it can provide useful aide memoirs at any one time of the incident management, such as customised action cards (this again could be supported with geo-location to pick the correct action cards for the given jurisdiction), special procedures – e.g. decontamination steps, patient treatment algorithms and many more. While it is relatively straightforward to provide predetermined algorithms, the challenge is to design the linking of keywords for types of events to historical data and extract suggestions. Apart from the well-known fact that there is a wild variation of terminologies and contexts in the available incident databases that need to be aligned, the confounding factor is also that the data are unreliable because they are plagued by a lot of subjectivity and gaps in recording. Therefore, one approach to be tested would be to start with a small load of intelligent suggestions that are deemed meaningful by the users and then set up the system to be able to use its own growing database for more reliable interpretations.

Once set up, this tool has a significant potential for further development and refinement, as more reliable data will become available. By its nature, this tool can also be a standalone tool, which should be integratable in similar systems.

The semantic tool links with the database and interacts with the incident management tool, including the hospital management tool (KT), the decision support algorithms (NCSR, UCY), the patient status information tool (ESRI), and the maps (ESRI).

Partners CUH and HRT have provided anonymised real incident data and user input, i.e. the domain knowledge and advice on which information is meaningful and can be further processed and which information does not allow any conclusions to be drawn.