<p>Failure mode and effects analysis (FMEA), as a well-established risk assessment methodology, has been widely applied across diverse domains. However, current FMEA approaches present certain limitations in representing and fusing uncertain evaluation information, while the integration of team members (TMs)’ social network relationships and consensus-building mechanisms in risk analysis also requires refinement. Therefore, to bridge these research gaps and enhance FMEA reliability, a novel consensus-driven FMEA approach is proposed through the synergistic integration of Dempster-Shafer (D-S) evidence theory with the social network DeGroot (SNDG) model. First, belief structures are employed to precisely model TMs’ complex preferences to address the limitations in representing multiple emotional preferences within evaluation information. Second, a dual-layered network structure combining directed trust relationships and undirected opinion similarity is developed to dynamically determine TMs’ weights for resolving the disconnect between social trust and professional opinions. Third, the maximizing deviation method (MDM) is adopted for objectively determining the weights of risk factor (RFs), while a modified Dempster’s rule (DCR) prevents counterintuitive evidence fusion outcomes. Finally, a novel consensus-reaching process (CRP) incorporating the SNDG model is designed to ensure reliable prioritization of failure modes (FMs) through dynamic opinion adjustment. The effectiveness and superiority of the proposed approach is verified through medical waste management (MWM) case studies, sensitivity analysis, and comparative analysis, with actionable risk rankings being generated to support healthcare risk mitigation strategies. The synergistic integration of D-S evidence theory with dynamic social network consensus mechanisms establishes an innovative framework for risk assessment in complex uncertain environments, combining mathematical rigor with social adaptability.</p>