[Paper Review] Design of statistical quality control procedures using genetic algorithms

Aristides T. Hatjimihail, Theophanes T. Hatjimihail|arXiv (Cornell University)|Jan 1, 1994

Clinical Laboratory Practices and Quality Control21 references3 citations

TL;DR

This paper proposes a genetic algorithm (GA)-based approach to design near-optimal statistical quality control (SQC) procedures for clinical laboratories, overcoming limitations of traditional algebraic or enumerative methods. Using the deterministic crowding GA, the method optimizes control rules to minimize false rejection while maximizing detection of critical random and systematic errors, demonstrating superior performance over 45 conventional QC procedures in a case study.

ABSTRACT

In general, we cannot use algebraic or enumerative methods to optimize a quality control (QC) procedure so as to detect the critical random and systematic analytical errors with stated probabilities, while the probability for false rejection is minimum. Genetic algorithms (GAs) offer an alternative, as they do not require knowledge of the objective function to be optimized and search through large parameter spaces quickly. To explore the application of GAs in statistical QC, we have developed an interactive GAs based computer program that designs a novel near optimal QC procedure, given an analytical process. The program uses the deterministic crowding algorithm. An illustrative application of the program suggests that it has the potential to design QC procedures that are significantly better than 45 alternative ones that are used in the clinical laboratories.

Motivation & Objective

• To address the challenge of designing optimal statistical quality control (SQC) procedures that balance high error detection with low false rejection rates.
• To overcome the limitations of algebraic and enumerative methods in optimizing complex, multi-parameter QC procedures.
• To develop an interactive, GA-based tool that enables the design of near-optimal QC procedures tailored to specific analytical processes.
• To evaluate the performance of GA-generated QC procedures against established conventional methods in clinical laboratory settings.

Proposed method

• The study employs the deterministic crowding genetic algorithm (GA) to search large parameter spaces for optimal control rules.
• The GA evolves populations of QC procedures by applying selection, crossover, and mutation operations to minimize false rejection while maximizing error detection.
• The optimization objective function is defined by user-specified probabilities for detecting critical random and systematic errors.
• The method uses a fitness function that penalizes high false rejection rates and rewards high error detection probabilities.
• An interactive computer program was developed to implement the GA-based design, allowing users to input process-specific parameters.
• The approach does not require analytical derivatives or knowledge of the objective function's gradient, making it suitable for complex, non-linear optimization problems.

Experimental results

Research questions

• RQ1 Can genetic algorithms effectively design statistical quality control procedures that outperform conventional methods in detecting critical analytical errors?
• RQ2 To what extent can a GA-based approach reduce the probability of false rejection while maintaining high error detection rates?
• RQ3 How does the performance of GA-generated QC procedures compare to 45 commonly used conventional QC procedures in clinical laboratories?
• RQ4 Can the deterministic crowding GA efficiently explore the large, discrete parameter space of control rule combinations in SQC design?

Key findings

• The GA-based method successfully designed near-optimal QC procedures that significantly outperformed 45 conventional QC procedures in terms of error detection and false rejection rates.
• The proposed approach achieved a substantial reduction in false rejection probability while maintaining or improving detection of critical random and systematic errors.
• The interactive GA-based program demonstrated practical feasibility for real-world clinical laboratory applications.
• The results suggest that genetic algorithms are a viable and powerful alternative to traditional optimization techniques in statistical quality control design.