Trimethylamine N-oxide (TMAO) has been recognized as a biomarker for the early detection of thrombosis. However, testing for TMAO typically requires expensive laboratory equipment and skilled technicians, making it unsuitable for home care pre-screening. To enable its widespread use in home applications, it is crucial to develop a scalable and sensitive device capable of catalyzing TMAO metabolism with a specific enzyme that is tailored for point-of-care use. This study presents an investigation of a MEMS-based two-tiered-tower biosensor array with a detection limit of 0.1 μM for TMAO, aiming to diagnose chronic metabolic diseases using urine or serum samples. Based on the augmented Cole-Cole model, the proposed parameters R_catalyzed, C_catalyzed, and Rp_catalyzed can predict the catalytic impedance of enzymatic activities such as the redox effects of analytes and characterize the small-signal current caused by catalysis. The proposed MEMS biosensor, integrated with a readout circuitry, demonstrates a high sensitivity of 41 ADC counts per μM TMAO (or 4.5 mV μM-1 TMAO), a response time of 1 second, a repetition rate of 98.9%, and a drift over time of 0.5 mV. The sensor effectively distinguishes TMAO based on minute capacitance changes induced by the TorA enzyme, resulting in a discernible distinction of 10.6%. These measurements were successfully compared to conventional cyclic voltammetry (CV) results, showing a variance of only 0.024%. The proposed biosensor is well-suited for pre-screening thrombosis factors for the early detection and prevention of thrombosis in point-of-care applications. The device is cost-effective, lightweight, and demonstrates excellent performance, with a conversion rate of 88% of TMAO and a selectivity rate of 97% for the by-product TMA, allowing for the prediction of cardiovascular risks.