Portfolio - Data Science, Statistics and Analytics
Study of Knee and Hip Prosthetic Vendors
Healthcare
Retrospective study of >2000 knee and hip replacement surgeries to determine if there were differences between prosthetic joint vendors relative to replacement type, physician preference, cost, surgical complications, severity of illness, case mix and comorbidities. The purpose of the study was to inform a decision to be made relative to if the system of 8 hospitals can standardize around fewer prosthetic vendors. The independent variables were largely nominal categorical therefore Mann-Whitney was used to determine if there were differences and Kruskal-Wallis techniques were used for Post-Hoc analysis. SPSS, Excel and MVP Stats were used for the analysis tools.
Hospitalist Versus Intensivist Effectiveness in ICUs
Healthcare
Retrospective study to determine if there is a difference of effectiveness between general hospitalist physicians and intensivist physicians relative to ICU outcomes. The independent variables studied were case mix index, severity of illness, complications, comorbidities and mortality. The dependent variables were both overall and ICU length of stay. The purpose of the study was to inform decisions on physician staffing. The independent variables were largely nominal categorical therefore Mann-Whitney was used to determine if there were differences and Kruskal-Wallis techniques were used for Post-Hoc analysis. SPSS, Excel and MVP Stats were used for the analysis tools.
Quality of Life Improvements for Knee and Hip Replacement Patients
Healthcare
Study of survey results of approximately 1000 patients how underwent hip and knee hip replacements. The survey was administered pre-surgery and then again 6 months after surgery to determine quality of life improvements. The survey was largely based on the Likert Scale and Mann-Whitney and Chi-Square techniques were used for the analysis. SPSS and MVP Stats were used for the analysis tools.
Contributing Factors to HCAHP Patient Experience Responses
Healthcare
Retrospective classification study performed to identify significant relationships between patient experience survey results (HCAHP) and possible contributing factors. Contributing factors considered were the treatment team, complications, severity of illness, diagnosis type, hospital units and the eight hospitals in the system. The dependent variables where the HCAHP top box percentages. The classification technique used was Exhaustive CHIAD analysis based on the Chi Square test. SPSS was used as the analysis tool.
HCAHP Control Chart Analysis
Healthcare
Control Chart analysis of HCAHP patient experience scores to identify special cause variation versus natural variation based on a patient sampling. The purpose of the analysis was to inform leadership of the risk of overreacting to natural variation and using it to drive improvement processes. Additionally, there was a tendency to draw a trend line through natural variation and claim patient experience was changing in a positive or negative direction. Minitab and Excel were the analysis tools.
Patient Falls Control Charts
Healthcare
Maintained and enhanced control charts for system wide patient falls. The control charts were used to identify natural cause variation and perform hospital to hospital comparisons for Total Falls Per 1000 Patients and Falls With Injury Per 1000 Patients.
Prediction Analysis and Goal Setting of Clinical Quality Measures
Healthcare
End of year prediction and confidence interval analysis of executive level clinical quality measures. The analysis was used for meaningful 3 tier goal setting relative to bonus payouts corresponding to strategic and tactical improvement process. The confidence interval analysis helped ensure the distance between goal tiers could be resolved beyond the measurement error. Excel was used as the analysis tool.
Component of Variance Analysis of Healthcare Cost
Healthcare
Retrospective study identifying variance components contributing to indirect cost by diagnosis types. The variance factors quantified hospital, unit, case mix index, severity of illness, complications and comorbidities. The purpose of the study was to identify large contributors of inappropriate cost variance. Excel was used as the analysis tool
Multiple Sclerosis Cognitive Study
Healthcare
Randomized Control Study supporting a clinical trial for an experimental treatment effect used on Multiple Sclerosis patients. All patients involved in the study had multiple sclerosis. Half of the patients were randomly selected to receive the treatment and the other half of patients were the control group who did not receive the treatment. The dependent variables studied were whole brain volume, hippocampus volume, thalamus volume, CVLT II cognitive results and SDMT cognitive results. Therefore, both continuous, ordinal and nominal statistical techniques were used. SPSS was used as the analysis tool.
Measurement System Analysis
FaultTree.accdb
Tracked the test debug process on manufactured electronic boards by assembly id, failure mode, failure symptom and root cause. Used to reduce future debug time on assemblies.
Metrology Regression Analysis
TestDev.accdb
Application to track the quoting process and all project milesones for complex test developments integrating both software and hardware designs. TestDev tracked the project against quoted cost and schedule. The database was also used to inform and dial in the accuracy of subsequent project quotes.
Process Capability Analysis
Misc
My use of Microsoft Access goes way beyond application development. I have used it extensively over 25 years for general data slinging (data discover, data manipulation, data transformation, data conversion). Many times just a component layer in a software stack.
Misc Application Development
Misc
To be honest, I do not remember all the applications and uses I’ve done with MS Access. I was an early adopter of Microsoft Access before it was released to the public in 1992 and have used it ever since (all versions). The above list represents some of the more substantial developments I have done with MS Access.
General Notes on MS Access Architectures.
Misc
Most MS Access architectures I implemented were client-server approaches with the client Access DB (user interface, queries and VBA) running on the users computer talking to a centralized backend Access DB running on a shared server or workstation. This is critcal for performance over a network and especially for mult-user implementations. This also helps reduce database corruptions on multi-user applicaitons.
Many times the backend database was not Access, but SQL Server, MS Excel or MySQL with data access configured through an OLEDB or native database driver (preferred) or ODBC. I also made extensive use of VBA. All VBA modules contain error trapping and write to system and transaction logs based on a Verbosity flag which defines the level of log detail.
