Presentation is loading. Please wait.

Presentation is loading. Please wait.

Applying of risk-adjusted CUSUM control chart monitoring of medical information Zi-Hsuan Chen Advisor: Jing-Er Chiu, Ph.D.

Similar presentations


Presentation on theme: "Applying of risk-adjusted CUSUM control chart monitoring of medical information Zi-Hsuan Chen Advisor: Jing-Er Chiu, Ph.D."— Presentation transcript:

1 Applying of risk-adjusted CUSUM control chart monitoring of medical information Zi-Hsuan Chen Advisor: Jing-Er Chiu, Ph.D.

2 Contents 2 Reason &PurposesLiterature ReviewResearch MethodsExpected Results

3 Research Reason 3 Reduce determine error, and quickly the failure occurs. Time conditions Risk adjustment factor Patient's heterogeneity Reason& Purposes Literature ReviewResearch MethodsExpected Results

4 Research Purposes 4 Construct risk-adjusted CUSUM control chart. Calculation of risk-adjusted CUSUM control chart ARL values. Added time conditions, the early detection of surgical failure. Reason& Purposes Literature ReviewResearch MethodsExpected Results

5 Research Framework 5 Shoulder surgery Logistic Model ARL CUSUM Cox Model ARL CUSUM Time conditions Reason& Purposes Literature ReviewResearch MethodsExpected Results

6 6 Limitations of Research The data include information from May 2010 to May 2013 Part of simulation data Biswas ed.(2008) Reason& Purposes Literature ReviewResearch MethodsExpected Results

7 CUSUM Control Chart Proposed by Page(1954) Small process shifts 7 Reason& Purposes Literature ReviewResearch MethodsExpected Results

8 Author Result Steiner et al. (2000) Since there individual differences between the patients.To find a risk-adjusted weight and standardize the basis of the risk of patients. Woodall (2006) The medical risks may come from the physicians surgical skills 、 Anesthesia practices 、 Patient physical condition The risk factor, and so on. J.S., Yu (2007) Explore of the heterogeneity patients before and after risk adjustment to the Fall rates. Applying Of CUSUM Control Chart In Medical 8 Reason& Purposes Literature ReviewResearch MethodsExpected Results

9 Logistic Regression Observed outcome can have only two possible 9 Reason& Purposes Literature ReviewResearch MethodsExpected Results Success Failure Source : Wikipedia

10 Author Result Steiner et al. (2000) Explore to the interns and experienced doctors of cardiac surgery performance,using logistic regression backward elimination) which filter out the main factor affecting the cardiac surgery. Chen Yuzhi et al.(2002) Logistic regression analysis to explore the inpatient fall over event occurred easily cause injury or complications. Biswas ed.(2008) Apply logistic regression analysis to explore the impact of kidney transplant surgeon factor is the age of the recipient, age of the donor, the donor and recipient weight ratio, donor age, disease status. Applying Of Logistic Regression In Medical 10 Reason& Purposes Literature ReviewResearch MethodsExpected Results

11 Cox PH Model Proposed by Cox(1972) Author Result Biswas ed.(2008) Apply of Cox Model in Kidney transplantation data. The primary outcome of interest was graft failure, including death with a functioning graft during the one-year period post-transplant.Note that in both risk-adjustment models, factors such as recipient age, donor to recipient weight ratio, donor age and presence of other diseases, among several other factors, are highly predictive of transplant failure. Use survival time  Applying Of Cox Model In Medical 11 Reason& Purposes Literature ReviewResearch MethodsExpected Results Use the Logistic Model Method and Cox Model Method to compare the result, which can quickly detect the patient has a shoulder joint shift or death situation.

12 Research Process 12 Collection and Simulation of patient information Notation and Assumptions Calculate and Simulation CUSUM control chart Calculate ARL Comparative analysis Conclusion Establish Cox Model Establish Cox Model Establish Logistic Model Establish Logistic Model Reason& Purposes Literature ReviewResearch MethodsExpected Results

13 Object Of Study And Assumptions Logistic Model Cox Model 13 Reason& Purposes Literature ReviewResearch MethodsExpected Results

14 Logistic Model Cox Model VariableRisk factor Explain Yt Surgery Status Shoulder surgery status : Success(0) 、 Failure (1) VariableRisk factor Explain Yt Surgical failure point in time Time of surgery from the start time until relapse Unit : month 14 Object Of Study And Assumptions Reason& Purposes Literature ReviewResearch MethodsExpected Results VariableRisk factor Explain SexFemale (0) and Male (1) Age34 – 94 year old to quantify the variables Artificial bone Devices bone before surgery : NO(0) 、 YES(1) The degree of injuryMinor (0), Medium (1), More serious(2) Bone plates at their own expense NHI payment : NO(0) 、 YES(1) Cancellous bone drugs Edible medicine : NO(0) 、 YES(1) Smoking Smoking behavior : NO(0) 、 YES(1) Menopause Female patient's menstrual status : Normal (0) 、 Menopause (1) Vegetarian food Non-vegetarian(0) 、 Vegetarian (1) Displand After surgery the patient‘s degree of deviation : 0mm ~ 19mm (quantify the variables)

15 Logistic Model in CUSUM Monitoring probability of Yt Binary data is transformed into probability Calculate the weight value Draw CUSUM control chart Steiner et al.(2000) 15 Reason& Purposes Literature ReviewResearch MethodsExpected Results

16 Cox Model in CUSUM Monitoring probability of Yt Yt of time conditions convert the probability of occurrence Calculate the weight value Draw CUSUM control chart Biswas ed.(2008) 16 Reason& Purposes Literature ReviewResearch MethodsExpected Results

17 Simulation ARL steps (Logistic Model ) Simulate number of samples N Out –of-control ⇒ ARL 1 Out –of-control ⇒ ARL 1 Randomly generated patient after surgery results To draw CUSUM control chart , and monitoring control chart 17 Steiner et al.(2000) Reason& Purposes Literature ReviewResearch MethodsExpected Results

18 Simulation ARL steps (Cox Model ) Simulate number of samples N, added time's condition Out –of-control ⇒ ARL 1 Out –of-control ⇒ ARL 1 Randomly generated patient after surgery results To draw CUSUM control chart , and monitoring control chart 18 Biswas ed.(2008) Reason& Purposes Literature ReviewResearch MethodsExpected Results

19 In CUSUM control chart, Cox model method can quickly detect the patient has a shoulder joint shift or death situation. In simulation ARL, use the Cox model to find the failure in a more timely manner. 19 Expected Results Reason& Purposes Literature ReviewResearch MethodsExpected Results

20 Collecting more variables of time, and apply to Cox model analysis of shoulder surgery. Input data in simulate method, then compare the result between simulation and actual. 20 Future Research Reason& Purposes Literature ReviewResearch MethodsExpected Results

21 - The end - 21


Download ppt "Applying of risk-adjusted CUSUM control chart monitoring of medical information Zi-Hsuan Chen Advisor: Jing-Er Chiu, Ph.D."

Similar presentations


Ads by Google