Project 23: WizWhy: an association rule tool f o r biomedical research Organization: WizSoft Inc. Abstract Primary Contact: Abraham Meidan Address: 3 Beit-Hillel St. We have developed a new association rules algorithm, Tel-Aviv 67017 that is used not just for revealing all the if-then rules, Israel but for issuing predictions as well. This algorithm Eniail: abraham@wizsoft.com avoids the problem of overfitting by calculating the URL: http://www.wizsoft.com/ error probability of each rule. Onr package, WizWhy for Windows 95/NT, that is based, on this algorithm is used in many biomedicine research groups. Duration: 5 years Number of People: 4 Keywords: association-rules Tools Developed: WizWhy Project 24: Data mining applied to in vitro fertilization, tumor and diabetic patient data Organization: Bavarian Research Center for KnowledgeAbstract Based Stems Primary Contact: Oliver Hogl We currently develop a technology to combine Address: FORWISS knowledge acquisition from domain experts and data Am Weichselgarten 7 mining thus allowing an easy and focused approach to D-91058 Erlangen intelligent knowledge discovery. By enabling the Germany domain expert to enter data mining queries in his/her domain terminology and by returning findings in the Email: ovhogl@forwiss.uni-erlangen.de URL: same terms as well as measuring the interestingness of findings, user support is provided throughout the http://www.forwiss.uni-edangen.de/fg-we/ knowledge discovery process. In the biomedical area a project on the discovery of interesting dependencies in Duration: 5 years patient data for the purpose of quali/y management has Number of People: 3 been carried out within several clinics. In this sector, Tools: data mining precedes confirmative analyses by Developed: Knowledge Discovery Assistant (research generating first hypotheses. In addition to that, a prototype) number of studies concerning the analysis of medical Purchased: SPSS, Pilot Decision Support Suite data, e.g. in the fields of in vitro fertilization, Academic Disciplines: AI, Knowledge Acquisition, and classification of tumors, fusion of sensor data, and Knowledge Based Systems diabetic patient profiling have been worked on with Funding Sources: 50% state funding, local university hospitals. By fusing complex medical 50% industry projects data from various Sources and analyzing them, new discoveries, e.g. decision trees fi)r tumors, could be made. Keywords: clinical quality management, tumor classification, in vitro fertilization, neurophysiological sensor data, diabetic patient profiling Project Related Publications: Hausdorf, Carsten and MOiler, Michael, "A Theory of Interestingness for Knowledge Discovery in Databases Exemplified in Medicine", In Lavrac, Nada and Keravnou, Elpida and Zupan, Blaz, First International Workshopon Intelligent DataAnalysis in Medicine and Pharmacology (IDAMAP-96), Budapest, 31- 36, 1996. Miiller, Michael, "Interestingness and Knowledge Discovery in Databases", Ph.D. Thesis, University ErlangenN0xnber~, 1998.

/lp/association-for-computing-machinery/data-mining-applied-to-in-vitro-fertilization-tumor-and-diabetic-b0yXeJmJh3