Reducing potential dual-use risks in synthetic biology laboratory research: a dynamic model of analysis

Synthetic biology (SynBio) is an interdisciplinary field that includes biology, genomics, engineering, and informatics. The introduction of machine learning technology reduces the requirement for human participation in SynBio research, which may lead to increased biosafety and biosecurity concerns within the complex human-technology-management system in SynBio laboratories compared with traditional biology. The rapid development of SynBio and machine learning technology has added to the systemic complexity of management and technology subsystems, the risks of which remain unclear. To fully understand and address these risks, there is an urgent need to develop a quantitative model. In this study, three-round Delphi interviews were conducted with SynBio experts who have published articles in top journals (e.g., Nature, Science, Cell, etc.) and applied the system dynamics method to build a quantitative analysis model. The model explores the potential risk factors and their complex relationships in the SynBio research. We analyzed three subsystems (biotechnology, information technology, and management), identifying seven key risk factors. Among these, the Average Experience of Newcomers and Tolerable Error Frequency play an essential role in laboratory management. Notably, the biosafety atmosphere has the greatest impact on reducing error rate and increasing safety awareness, as well as on the infrastructure safety of SynBio laboratory. In addition, we also consider the impact of the progressiveness of machine learning in the SynBio research and find that using more advanced machine learning technology leads to lower instrument safety. This study provides a framework for quantifying potential risk factors in SynBio research and lays a theoretical foundation for future laboratory safety management and the development of a global code of conduct for interdisciplinary scientists.