Blade damage diagnosis is a critical requirement for ensuring operational safety and reliability in aero-engines and gas turbines. Blade tip timing (BTT) is a well-established non-contact vibration monitoring technique. However, most existing damage detection methods focus on single-blade response analysis, which suffers from diagnostic latency and can only identify faults at advanced stages when vibration deviations become significant. To enable early damage detection, this study proposes a system-level approach that analyzes the coupled dynamics of the bladed disk. The methodology incorporates a vector autoregressive (VAR) based modal parameter identification technique and introduces the blade dominant mode as a diagnostic parameter for incipient damage identification. The framework is validated through numerical simulations and compressor test rig experiments. The results demonstrate that single-blade responses exhibit interference from multiple bladed disk modes, introducing uncertainty in traditional resonance frequency identification. This limitation constrains the effectiveness of conventional frequency-shift methods for early-stage damage detection. In contrast, bladed disk mode analysis reveals a distinct monotonic decrease in the blade dominant mode frequencies of damaged blades, with shifts significantly exceeding normal variations observed in healthy blades. By utilizing this sensitive parameter, the proposed damage identification framework achieves precise localization of incipient blade damage. The method demonstrates robust performance under compressor test conditions, representing a significant improvement over existing BTT techniques for proactive fault prevention in rotating machinery.