A Review of AI and Machine Learning Contribution in Predictive Business Process Management (Process Enhancement and Process Improvement Approaches)

TL;DR

This survey investigates how AI and ML contribute to predictive BPM through two main avenues: process enhancement (predictive monitoring and analytics on ongoing processes) and process improvement (process redesign and corrective actions). It adopts a PRISMA-inspired six-step methodology and presents a bibliometric analysis showing a strong focus on process improvement while highlighting a growing PBPM literature and gaps in integrated approaches. The paper details substantial ML methods across drift/anomaly detection, remaining time and next-activity prediction, constraint monitoring, performance forecasting, and simulation-based evaluation, then covers integrated enhancement-improvement strategies with emphasis on pattern similarity, NLP-driven redesign, and Lean Six Sigma automation. Key findings suggest integrated, AI-driven BPM lifecycle support offers superior applicability, but challenges remain in data availability, explainability, and handling complex constructs like loops and parallelism. The work points to opportunities for generative models, unsupervised learning, few-shot/zero-shot learning, and synthetic data to advance robust, explainable BPM automation.

Abstract

Purpose- The significance of business processes has fostered a close collaboration between academia and industry. Moreover, the business landscape has witnessed continuous transformation, closely intertwined with technological advancements. Our main goal is to offer researchers and process analysts insights into the latest developments concerning Artificial Intelligence (AI) and Machine Learning (ML) to optimize their processes in an organization and identify research gaps and future directions in the field. Design/methodology/approach- In this study, we perform a systematic review of academic literature to investigate the integration of AI/ML in business process management (BPM). We categorize the literature according to the BPM life-cycle and employ bibliometric and objective-oriented methodology, to analyze related papers. Findings- In business process management and process map, AI/ML has made significant improvements using operational data on process metrics. These developments involve two distinct stages: (1) process enhancement, which emphasizes analyzing process information and adding descriptions to process models, and (2) process improvement, which focuses on redesigning processes based on insights derived from analysis. Research limitations/implications- While this review paper serves to provide an overview of different approaches for addressing process-related challenges, it does not delve deeply into the intricacies of fine-grained technical details of each method. This work focuses on recent papers conducted between 2010 and 2024. Originality/value- This paper adopts a pioneering approach by conducting an extensive examination of the integration of AI/ML techniques across the entire process management lifecycle. Additionally, it presents groundbreaking research and introduces AI/ML-enabled integrated tools, further enhancing the insights for future research.

Figures (4)

• Figure 1: Research Scope$^{1}$
• Figure 2: Research Methodology
• Figure 3: Number of published papers in 2010-2024
• Figure 4: Keyword network links and clusters in this context