Efficient coordination of patient referral pathways in Integrated Healthcare Networks (IHNs) presents a critical challenge, as conventional optimization algorithms often fail to simultaneously optimize institutional revenue, healthcare cost containment, and patient satisfaction enhancement. These methods typically exhibit premature convergence and limited solution diversity. To address these limitations, we propose LA-NSGA-II (Local-Adaptive Non-dominated Sorting Genetic Algorithm II), a novel multi-objective evolutionary algorithm featuring two key innovations: (1) a neighborhood-enhanced local search mechanism that optimizes elite solutions along the Pareto front, and (2) an adaptive crossover-mutation operator that dynamically adjusts probabilities based on global search progression and individual fitness metrics. These synergistic mechanisms collectively address the exploration-exploitation dilemma, achieving superior convergence rates while maintaining population diversity. The algorithm is implemented within the TDC-Coordination Optimization Framework, which unifies bidirectional referrals, parallel referral pathways, and centralized scheduling operations. Empirical validation using real-world data from a regional IHN in Shenzhen demonstrates LA-NSGA-II's superiority over classical and state-of-the-art counterparts (NSGA-II-wDOS, A-NSGA-II), yielding an 8.1% increase in network revenue, 4.2% reduction in insurance expenditures, and 75% improvement in patient utility coverage. These results not only LA-NSGA-II's robustness in addressing healthcare multi-objective optimization challenges, but also provide actionable insights for operational management in integrated care systems.