Unilateral neglect (UN) is a common post-stroke neurocognitive deficit linked to interhemispheric interactions, though mechanisms remain unclear. This study evaluated the 'bimodal balance recovery' model in UN, exploring its relationship with interhemispheric connectivity and proposing a stratification framework for patient categorization. Thirty stroke patients with UN and 15 healthy controls were recruited. Interhemispheric signal propagation (ISP) was assessed using transcranial magnetic stimulation-electroencephalography. UN severity was quantified using a battery of paper-and-pencil tasks, and overall patient functioning was evaluated using the Activities of Daily Living Scale, Fugl-Meyer Assessment, and Berg Balance Scale. Analyses of the Schenkenberg Line Bisection Test, Albert's Cancellation Task, and Ota's Cancellation Task indicated that quadratic models provided a better fit than linear regressions. A novel metric, the PenPCA Index, was derived using Principal Component Analysis (PCA) to assess the complex relationship between ISP and UN. This index demonstrated a bimodal relationship with ISP, effectively distinguishing between negative and positive correlations with the PenPCA Index. This study introduces the PenPCA Index, underscores the bimodal UN-ISP relationship, and offers a stratified assessment framework for stroke patients.