CSPH is a critical stage in chronic liver disease, as it signals a heightened risk of life-threatening complications. Patients diagnosed with CSPH are often prescribed non-selective beta-blockers (NSBBs), which have been shown to reduce the risk of disease progression. However, confirming CSPH currently relies on hepatic venous pressure gradient (HVPG) measurement, an invasive procedure that is available only in specialised centres.
Existing non-invasive guidelines, known as the Baveno VII criteria, use liver stiffness measurements (LSM) and platelet counts to estimate the likelihood of CSPH. While widely adopted, these criteria leave around 40–50% of patients in an indeterminate "gray zone," making it difficult for clinicians to decide on treatment.
This study developed the Elastography and Learning Machine (ELM) Score, a Random Forest-based machine-learning model that combines liver stiffness, spleen stiffness, and routine clinical information, including platelet count, Child-Pugh score, age, sex, and liver disease cause.
The model was designed to work across multiple elastography technologies, including vibration-controlled transient elastography (VCTE), two-dimensional shear-wave elastography (2D-SWE), and point shear-wave elastography (p-SWE).
The study analysed data from 1,435 patients with compensated ACLD who underwent paired HVPG measurements alongside liver and spleen stiffness assessments. Of these, 943 patients were used to train the model, 150 for internal validation, and 342 patients from seven independent centres for external validation.
The ELM Score achieved strong diagnostic performance during external validation, recording an area under the curve (AUC) of 0.91 and a Brier Score of 0.13. Using predefined thresholds, the model demonstrated a negative predictive value of 90% for ruling out CSPH and a positive predictive value of 96% for confirming the condition.
Perhaps most notably, the model dramatically reduced the proportion of patients falling into the diagnostic gray zone. Only 12.3% of patients remained indeterminate, compared with 47.9% under the Baveno VII criteria.
The model also outperformed alternative spleen stiffness-based approaches and, among patients assessed using VCTE, reduced gray-zone cases to 12% compared with more than 41% using the ANTICIPATE and NICER scoring systems.
Overall, the ELM Score offers a more accurate and broadly applicable method for identifying patients with HVPG-defined CSPH across different elastography platforms.
Source:
Giuffrè, M., Kresevic, S., Ravaioli, F., Zykus, R., Rautou, P.-E., Elkrief, L., Colecchia, L., Kukic, S., Barisic-Jaman, M., Grgurevic, I., Stefanescu, H., Hirooka, M., Fraquelli, M., Rosselli, M., Chang, P. E. J., Crocè, L., Ajcevic, M., Piscaglia, F., Reiberger, T., … Study group. (2026). Validation of a pan-ELastography Machine-learning (ELM) score to predict clinically significant portal hypertension in compensated advanced chronic liver disease. Journal of Hepatology. https://doi.org/10.1016/j.jhep.2026.06.032
Disclaimer: This website is primarily for healthcare professionals. The content here does not replace medical advice and should not be used as medical, diagnostic, endorsement, treatment, or prescription advice. Medical science evolves rapidly, and we strive to keep our information current. If you find any discrepancies, please contact us at corrections@medicaldialogues.in. Read our Correction Policy here. Nothing here should be used as a substitute for medical advice, diagnosis, or treatment. We do not endorse any healthcare advice that contradicts a physician's guidance. Use of this site is subject to our Terms of Use, Privacy Policy, and Advertisement Policy. For more details, read our Full Disclaimer here.
NOTE: Join us in combating medical misinformation. If you encounter a questionable health, medical, or medical education claim, email us at factcheck@medicaldialogues.in for evaluation.