For a study, it was determined that the risk ratings did not reliably identify patients at the highest risk of recurring atherosclerotic cardiovascular disease (ASCVD) and required more extensive treatment. Advances in high-throughput plasma proteomics, along with machine learning approaches, provided new ways to improve risk classification in individuals. The Second Manifestations of Arterial Disease (SMART) cohort (n=870) and the Athero-Express cohort (n=700) were both studied with targeted plasma proteomics. Recurrent ASCVD was the primary outcome (acute myocardial infarction, ischaemic stroke, and cardiovascular death). In the derivation cohort (SMART), machine learning techniques with severe gradient boosting were employed to generate a protein model, which was then validated in the Athero-Express cohort and compared to a clinical risk model. Pathway analysis was used to identify particular pathways in patients with high and low C-reactive protein (CRP). In both the derivation and validation cohorts, the protein model outperformed the clinical model [area under the curve (AUC): 0.810 vs 0.750; P<0.001] and validation cohort (AUC: 0.801 vs 0.765; P<0.001), produced considerable net reclassification improvement (0.173 in the validation cohort), and was well-calibrated. Neutrophil-signaling-related proteins were linked to recurrent ASCVD in low CRP patients, in contrast to a clear interleukin-6 signal in high CRP patients. In predicting recurrent ASCVD episodes, a proteome-based risk model outperformed a clinical risk model. Low CRP individuals had neutrophil-related pathways, demonstrating the presence of a residual inflammatory risk beyond typical NLRP3 paths. The observed net reclassification improvement demonstrated the promise of proteomics in secondary prevention patients when used in conjunction with a personalized therapy approach.