Advancing life cycle assessment through data science: A critical review of algorithms, tools, and data challenges

A well-executed life cycle assessment requires thorough data collection across all relevant processes, combined with advanced data analysis. Common data-related issues in life cycle assessment research include the absence of necessary data, low data quality, inconsistencies, uncertainty, and failure to account for variations over time and location. In this context, data science, the discipline of extracting meaningful insights from data, has the potential to address these challenges. While the integration of data science with life cycle assessment holds significant potential, best use cases depend on the goal of the study, as well as the data type and volume required, underscoring the necessity of reviewing the intersection of data science and life cycle assessment. This study used the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) method to identify literature addressing the use of data science elements to support life cycle assessment. It evaluated which data science techniques are appropriate for specific life cycle assessment stages or problem areas and the strengths and weaknesses of current data science applications in life cycle assessment. Key opportunities identified revolve around solutions for dealing with missing or poor-quality data, expensive/prohibitive data collection, and improving the accuracy of life cycle assessment results. The currently most feasible pathways appear to involve use of machine learning techniques, as these types of studies were the most conducted and generated tangible results. Extreme gradient boosting, random forest, and artificial neural networks were particularly prominent algorithm choices. Data collection and transferability using ontologies and semantic tools were also highlighted as important strategies for improving data flow in life cycle assessment, including the integration of a wide variety of databases and non-life cycle assessment data.