Class Introduction
This was the final session of a 3-week Health Data Analytics course, covering data validation, visualization, and statistical process control. John led the discussion on data validation concepts including measurement systems analysis, validity, reliability, and reproducibility, using practical examples to explain the importance of accurate data collection for international publications and regulatory submissions. The session covered various data visualization techniques including bar graphs, line charts, histograms, and Pareto charts, with detailed explanations of how to interpret and create these charts using Excel. John also discussed the concepts of common cause and special cause variation in statistical process control, explaining how to identify and respond to different types of variation in healthcare data. The participants, including Najeh, Joy, and Victoria, actively engaged with questions about scatter plots and correlation, with John clarifying that opposing variables can still show correlation in scatter plot diagrams.
Health Data Analytics Training Final
John conducted the final session of a three-week health data analytics training program, reviewing key concepts including sampling methods, measures of dispersion, and data validation techniques. The session covered the importance of data accuracy and reliability, explaining how validity and reliability are measured through concepts like reproducibility and repeatability. John illustrated these concepts using examples of data collection in healthcare settings and emphasized the critical role of accurate data in scientific publications and organizational decision-making.
Data Collection Reproducibility Standards
John and Najeh discussed reproducibility and repeatability in data collection, establishing that results should be at least 90% similar to be acceptable, with differences above 70-80% requiring rejection and root cause analysis. They examined a specific example involving a smoking cessation nurse and 10 patients, where a data collector interviewed patients to verify if they received advice, comparing the results against the medical records in Cerner. The discussion concluded by counting the number of agreements between the data collector's findings and the standard records, finding 9 out of 10 cases matched correctly.
Related Offerings
Data Collector Measurement Analysis
John presented a measurement systems analysis of data collectors, discussing accuracy, repeatability, and reproducibility. Data collector number one had 90% accuracy but only 80% repeatability when collecting data at different times, which falls within acceptable but cautionary standards. When comparing different data collectors, they found only 50% agreement in results, prompting John to recommend conducting a root cause analysis to identify potential issues like training gaps or language barriers before making any changes to the data collection process.
Data Visualization Training Session
John taught a session on data visualization, explaining how to transform complex data into easily understandable charts and graphs. He covered different types of visualizations including bar charts, line graphs, and run charts, with detailed explanations of how to interpret trends, shifts, and runs in data. The session included practical examples using Excel to demonstrate creating these visualizations and analyzing data patterns, with John asking for confirmation of understanding after explaining key concepts like shifts and trends.
Histograms and Pareto Charts
John explained the concept of histograms, emphasizing their use in displaying frequency distributions rather than making direct comparisons like bar graphs. He demonstrated how to create a histogram from patient waiting time data, showing the distribution of waiting times between 60-124 minutes, with most patients waiting between 70-79 minutes. John also introduced Pareto charts as a prioritization tool based on the 80-20 rule, explaining how they combine bar graphs and line charts to show where focus should be directed, using needle-stick injuries across 11 departments as an example.
Pareto Chart for Needle Stick Injuries
John demonstrated how to create and interpret a Pareto chart using medical department data, showing how to identify the "vital few" that represent 80% of the problem. He explained the calculation method for percentages and cumulative frequencies, using emergency departments, orthopedics, surgery, and gastroenterology data to illustrate how the chart helps prioritize which departments to focus on to address the majority of needle stick injuries. John emphasized that focusing on these key departments would solve 80% of the problem, making it more efficient to concentrate resources on the vital few rather than addressing all departments equally.
Data Visualization Techniques Training
John taught the class about three data visualization techniques: Pareto charts for prioritization (using 80-20 ratios), pie charts for showing proportions in small datasets, and scatterplot diagrams for analyzing correlation between two variables. He explained that scatterplots can show positive correlation (when both variables move in the same direction) or negative correlation (when they move in opposite directions), and demonstrated how to interpret these relationships using examples. When Najeh asked about bidirectional relationships in scatterplots, John clarified that both variables should either increase or decrease for proper correlation analysis.
Statistical Process Control Training
John taught the class about statistical process control, focusing on common cause and special cause variations. He explained how to identify these variations using control charts and run charts, using examples like hospital readmission rates and daily commute times. John also addressed a question from Najeh about scatter plot diagrams and correlation, clarifying that variables don't need to have the same units. The class will have a module quiz next week after covering inferential statistics, which will count toward their overall Health Data Analytics exam score.
Ready to build expertise in healthcare quality and data analytics?
