Background : Pulmonary fibrosis can develop after acute respiratory distress syndrome (ARDS). The hypothesis is we are able to measure phenotypes that lie at the origin of ARDS severity and fibrosis development. The aim is an accuracy study of prognostic circulating biomarkers. Methods : A longitudinal study followed COVID-related ARDS patients with medical imaging, pulmonary function tests and biomarker analysis, generating 444 laboratory data. Comparison to controls used non-parametrical statistics; p < 0·05 was considered significant. Cut-offs were obtained through receiver operating curve. Contingency tables revealed predictive values. Odds ratio was calculated through logistic regression. Results : Angiotensin 1–7 beneath 138 pg/mL defined Angiotensin imbalance phenotype. Hyper-inflammatory phenotype showed a composite index test above 34, based on high Angiotensin 1–7, C-Reactive Protein, Ferritin and Transforming Growth Factor-β. Analytical study showed conformity to predefined goals. Clinical performance gave a positive predictive value of 95 % (95 % confidence interval, 82 %–99 %), and a negative predictive value of 100 % (95 % confidence interval, 65 %–100 %). Those severe ARDS phenotypes represented 34 (Odds 95 % confidence interval, 3–355) times higher risk for pulmonary fibrosis development (p < 0·001). Conclusions : Angiotensin 1–7 composite index is an early and objective predictor of ARDS evolving to pulmonary fibrosis. It may guide therapeutic decisions in targeted phenotypes.