Sign in Subscribe
Machine learning

Personalized Medicine Breakthrough: Predicting SLE Activity through Hierarchical Machine Learning

Strategic Insights

— 1 min read

Researchers from Fondazione Policlinico Universitario Agostino Gemelli IRCCS have developed a hierarchical machine learning model to predict 12-month SLE activity in patients with systemic lupus erythematosus (SLE). This innovative approach considers patient demographics, laboratory, clinical features, treatments, and pathways, resulting in a reliable tool for predicting SLE activity. The model demonstrated enhanced performance, increasing the area under the receiver operating characteristic curve from 0.696 to 0.743, and identified key features that significantly influence the model's predictions.

Key Takeaways:

  • The study cohort consisted of 262 patients with at least an outpatient visit and an SLE admission from 2012 to 2020, at the Italian Gemelli Hospital.
  • The data included demographics, laboratory, clinical features, treatments, and pathways, covering 5962 contacts over the 8-year period.
  • The hierarchical machine learning model demonstrated enhanced performance, with an area under the receiver operating characteristic curve of 0.743 (95% CI 0.717-0.769).
  • The model identified 15 key features that significantly influence the predictions, including age at contact, response to therapy modifications, abnormal laboratory tests, and clinical manifestations.
  • The study proposes a hierarchical machine learning model to predict a 12-month SLE activity, defined as the occurrence of at least one event among SLE hospitalization, new organ-involved domain, and neurological, renal, or vascular manifestation within the following year.
  • The research concludes that this methodology can be generalized for predictive analytics in other chronic autoimmune diseases.

Statistics:

  • 262 patients with SLE were included in the study cohort.
  • 5962 contacts were analyzed over the 8-year period from 2012 to 2020.
  • The model's performance increased the area under the receiver operating characteristic curve from 0.696 to 0.743.
  • 15 key features were identified as significantly influencing the model's predictions.
  • 2012-2020: the time frame during which the study cohort was followed.

Sources:

  • Prediction of 1-Year Activity in Systemic Lupus Erythematosus: Hierarchical Machine Learning Approach. JMIR Formative Research, 2025;9.
  • NewsRx. Studies in the Area of Personalized Medicine Reported from Fondazione Policlinico Universitario Agostino Gemelli IRCCS (Prediction of 1-Year Activity in Systemic Lupus Erythematosus: Hierarchical Machine Learning Approach). Drug Week. September 19, 2025; p 7928.

Read more