Deficient recovery from acute kidney injury (AKI) has immediate and long-term health, clinical and economic consequences. Pre-emptive recovery estimation may improve nephrology referral, optimize decision making, enrollment in trials, and provide key information for subsequent clinical handling and follow-up. For this purpose, new biomarkers are needed that predict outcome during the AKI episode. We hypothesized that damage pattern-specific biomarkers are expected to more closely associate to outcome within distinct subpopulations (i.e. those affected by specific pathological processes determining a specific outcome), as biomarker pleiotropy (i.e. associated to phenomena unrelated to AKI) introduced by unselected, heterogeneous populations may blur statistics. A panel of urinary biomarkers was measured in patients with AKI and their capacity to associate to normal or abnormal recovery was studied in the whole cohort or after sub-classification by AKI etiology, namely pre-renal and intrinsic AKI. A combination of urinary GM2AP and TCP1-eta best associates with recovery from AKI, specifically within the sub-population of renal AKI patients. This two-step strategy generates a multidimensional space in which patients with specific characteristics (i.e. renal AKI patients with good or bad prognosis) can be identified based on a collection of biomarkers working serially, applying pathophysiology-driven criteria to estimate AKI recovery, to facilitate pre-emptive and personalized handling.