Using Computer Simulation Models to Assist with Laboratory Decision Making

George Herrin, Jr., Ph.D.John Speir, Ph.D.

April 29, 2015ASCLD Symposium

Process Improvement Approaches

Process Mapping

Lean Six Sigma

Goal Setting

Process Modeling via software, e.g. Simul8

What is Simul8

Discrete event simulation software tool

Mimics operational processes to understand interaction between people, infrastructure, and work activities

Used by multiple industries such as health care, manufacturing, and public safety

Purpose of Simulation Modeling

Determine appropriate resource levels to meet desired report turnaround times

Evaluate numerous ‘what-if’ scenarios without implementation costs

Identify bottlenecks and non-beneficial changes before actual implementation

Drivers for Toxicology Model

Increasing casework volume resulting in backlogs

Need to determine number of scientists required to meet customer expectation on report delivery times

What to Avoid

Modeling Components

Activities

ResourcesProduct

Developing a ModelIdentify core questions the model should address

Identify existing performance data that can be incorporated into the model

Build a simple model incorporating major process steps

Evaluate model performance against existing empirical data

Refine model design by incorporating additional details

Design Strategies

Gather sample and process timing data from casework

Include Subject Matter Experts

Include critical business rules, e.g. peer review, timing of certain operational steps

Evaluate model to determine if output is consistent with existing performance

Model Overview

Sample Screening

Determine Extractions

Reporting

Model Input & Output

Staffing Patterns Total# Post-Mortem 6

# DUI 7Position Service Amp TB BZScientist1 2 1 Scientist2 2 1 Scientist3 1 Scientist4 1 Scientist5 1 1

Post-Mortem Estimate

# of Cases Received 1,541# of Cases Completed 1,097Minimum Process Time 20.1 daysAverage Process Time 41.6 daysMaximum Process Time 68.7 days

Utilization %Scientist1 103.5Scientist2 102.0Scientist3 60.6Scientist4 49.8Scientist5 85.6

Toxicology Model Outcomes

Model confirms baseline knowledge that existing number of scientists is insufficient

The model predicts that 4 additional scientists could improve turnaround times by approximately 10-15%

Potential bottlenecks, such as the extraction for amphetamines have been identified

Lessons Learned

Empirical data about your existing process helps in troubleshooting model design

Use an existing model or work with an experienced user of the software.

Participate and engage in the model development from the early stages

Future DirectionsContinue refining the toxicology model architecture and timing

Use the model as a tool, along with management experience, in determining how to best utilize 4 scientists who are completing training

Use model predictions in briefing paper requesting additional personnel

Evaluate model data to identify potentially beneficial changes to operational processes

Desired Outcome

Questions