1. Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Predicting survival time for kidney dialysis patients: a data mining approach Advisor : Dr. Hsu Presenter : Zih-Hui Lin Author :AndrewKusiak a ; ∗, Bradley Dixon b, Shital Shah a Computers in Biology and Medicine 35 (2005) 311–327

2. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 2  Motivation  Objective Data transformation Decision-making algorithm Result  Conclusions  Future work Outline

3. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 3 Motivation  The cost for providing care for patients on hemodialysis due to end stage kidney disease is high.  Over 50 parameters may be monitored on a regular basis in providing kidney dialysis treatments.  Individual patient survival may depend on a complex interrelationship between multiple demographic and clinical parameters, medications, medical interventions, and the dialysis treatment prescription.

4. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 4 Objective  elicit knowledge about the interaction between many of these measured parameters and patient survival.

5. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 5 Data collection  Data collection was performed at four satellite locations of The University of Iowa Hospitals and Clinics (UIHC) ─ Dialysis patients received the treatment three times a week.. systolic and diastolic blood pressure measurement 、 the levels of sodium, bicarbonate, potassium, calcium, and glucose 、 blood flow rate 、 blood pressure……… The demographic and outcomes data set contained the patient’s date of birth, gender, and race; the date(s) of death, kidney transplant, transfer into dialysis center, and transfer out of dialysis center, and the diagnosis codes for the primary and secondary diagnoses.

6. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 6 Data transformation – 1 out of n cross-validation

7. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 7 Decision-making algorithm  The decision-tree (DT) algorithm creates a decision trees or sets of decision rules based on the concept of information gain.  The rough-set (RS) algorithm. ─ Certain rules (lower approximations) ─ Approximate rules (upper approximations) Voting scheme Lower approximation upper approximation

8. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 8 Result  RS rule ─ If (Diff_Pst_PreSupine_S = -21.7308) AND (Ca <3.40226) THEN (Survial_length=Above_med)  DT rule ─ IF(d_NA = 9) AND (Ca <= 3.346535) AND (d_ Art_S = 6) THEN (Survival_Length = Above_med)

9. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 9 Result

10. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 10 Result

11. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 11 Medical significance

12. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 12 Conclusions  Data transformation increased the classi7cation accuracy  The medical relevance of the signi7cant parameters was established.  Benefits and applications of the research results ─ Clinical studies ─ Treatment selection ─ “Model patient” selection: ─ More effective data collection process

13. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 13 Future research  A larger set could provide more meaningful results  incompleteness of data.  In this research, the limited memory of the machine.  Patient parameters were not considered over time  Predicting a “short-term performance” parameter could be investigated

14. Intelligent Database Systems Lab N.Y.U.S.T. I. M. 14 My opinion Advantage: Disadvantage: Apply: 應用於醫療上