BI Support for Documenting Tumors
Content:
Further development of the tumor documentation and management system at the Universitäts KrebsCentrum Dresden (UCC).
Background:
The clinical cancer registry of UCC has developed a tumor documentation and management system (TDS). In the system all tumor board recommendations and patient master data, diagnostic and theratpeutic measures and aftercare data is stored. The parameters contained in the TDS include all standards of documentation for tumor patients that where aligned by Deutsche Krebshilfe e.V, Deutsche Krebsgesellschaft, Arbeitsgemeinschaft Deutscher Tumorzentren (ADT) and the forum Comprehensive Cancer Center. Furthermore, TDS includes advanced parameters and different functionalities to support the daily work of doctors and other staff members. There are interfaces to the study index and the tumor and normal tissue bank of UCC as well as to the regional clinical cancer registry Dresden.
Objective:
The main objective of the project is to develop a model to structure, extract and load routine data from other information systems into TDS. As a sub-goal a process model that describes the loading of pharmaceutical production data for chemotherapy treatments into TDS is developed. Within this model a new approach to transfer the information regarding the therapy cycle, chemotherapy protocols, dose reduction density, start and end of the treatment, objective of the treatment as well as the treating institution is focused. Thereby, the information should be distinctly and accurately assigned to the patient and treated tumor. The transfer should be conducted automatically in a structured manner. The assignment of pharmaceutical data to the diagnosed tumors should be facilitated with basic rules. The rules stored in the system can be used to infer possible diagnoses from given pharmaceutical combinations by considering certain probabilities.
Research Approach:
In a first step the necessary tasks to transfer data into TDS will be defined by conducting a requirements analysis. Based on the results, a process that extracts, processes and assigns all relevant data to the patient is designed. Afterwards, as a main part of the research project historic data will be analyzed to infer basic rules from existing data for the deduction of diagnoses. Suitable Data Mining techniques are selected and applied to generate rules based on the data. Afterwards, the created data-driven recommender system is evaluated by professionals from medical business. The result will be a validated expert system.
Coordination:
Prof. Dr. Andreas Hilbert
Processor:
Dipl.-Wirt.-Inf. Andreas Schieber
Cooperation:
Universitäts KrebsCentrum of the university's hospital Carl Gustav Carus
Duration:
2012/12/01 - 2013/11/30