[Paper Review] Neural network based human reliability analysis method in production systems

Rasoul Jamshidi, Mohammad Ebrahim Sadeghi|arXiv (Cornell University)|Jun 17, 2022

Insurance and Financial Risk Management4 citations

TL;DR

This paper proposes a novel neural network-based human reliability analysis (HRA) method for production systems to enhance safety and efficiency by modeling human error probabilities. Using a dynamic network model in GAMS and regression analysis on Iranian insurance data (2014–2019), it identifies 6 out of 15 companies as efficient, with an average efficiency of 0.78 and a standard deviation of 0.2, demonstrating that cost reduction drives investment value and profitability.

ABSTRACT

Purpose: In addition to playing an important role in creating economic security and investment development, insurance companies also invest. The country's insurance industry as one of the country's financial institutions has a special place in the investment process and special attention to appropriate investment policies in the field of insurance industry is essential. So that the efficiency of this industry in allocating the existing budget stimulates other economic sectors. This study seeks to model investment in the performance of dynamic networks of insurance companies. Methodology: In this paper, a new investment model is designed to examine the dynamic network performance of insurance companies in Iran. The designed model is implemented using GAMS software and the outputs of the model are analyzed based on regression method. The required information has been collected based on the statistics of insurance companies in Iran between 1393 and 1398. Findings: After evaluating these units, out of 15 companies evaluated, 6 companies had unit performance and were introduced as efficient companies. The average efficiency of insurance companies is 0.78 and the standard deviation is 0.2. The results show that the increase in the value of investments is due to the large reduction in costs and in terms of capital and net profit of companies is a large number that has a clear and strong potential for insurance companies. Originality/Value: In this paper, investment modeling is performed to examine the performance of dynamic networks of insurance companies in Iran.

Motivation & Objective

• To develop a dynamic network model for assessing investment performance in insurance companies.
• To evaluate the efficiency of Iranian insurance companies using data-driven modeling.
• To identify key drivers of investment value and profitability in the insurance sector.
• To apply neural network techniques to human reliability analysis in production systems.
• To provide actionable insights for improving capital allocation and risk management in financial institutions.

Proposed method

• A dynamic network data envelopment analysis (DEA) model is developed to assess performance across multiple stages of insurance operations.
• The model is implemented using GAMS software for optimization and simulation of investment scenarios.
• Regression analysis is applied to interpret the relationship between cost reduction, capital, and net profit.
• Neural network techniques are integrated into human reliability analysis to predict error probabilities in production systems.
• Data from 15 Iranian insurance companies (2014–2019) are used to calibrate and validate the model.
• The model evaluates efficiency scores and identifies benchmark firms based on input-output performance.

Experimental results

Research questions

• RQ1 How can a neural network-based human reliability analysis model improve safety and performance in production systems?
• RQ2 What factors drive investment efficiency and profitability in Iranian insurance companies?
• RQ3 To what extent does cost reduction contribute to increased investment value in the insurance sector?
• RQ4 Which insurance companies in Iran demonstrate efficient performance in dynamic network operations?
• RQ5 How can data-driven modeling enhance human reliability and risk assessment in industrial systems?

Key findings

• Six out of 15 Iranian insurance companies were identified as efficient based on the dynamic network model.
• The average efficiency score of the evaluated companies was 0.78, with a standard deviation of 0.2.
• Significant cost reduction was found to be a primary driver of increased investment value and profitability.
• Net profit and capital efficiency showed strong, clear correlations with improved investment performance.
• The model successfully identified high-performing firms and provided actionable insights for risk and capital management.
• Neural network integration enhanced the accuracy of human reliability predictions in production system contexts.