Prediction medium and long-term risks in asplenic patients: a precision medicine approach

The spleen plays a fundamental role in immune response and blood filtration. It is responsible removing bacteria, particles, and damaged or aged blood cells, as well as the production of antibodies and the regulation of B and T lymphocyte activity. The loss of splenic function, whether due to surgical or functional asplenia, compromises the body's ability to respond to bacterial infections and increases the risk of venous and arterial thrombosis. Due to the variety of underlying diseases and conditions, predictive factors and tools for associated long-term risks are lacking. This study aims to develop and validate Machine Learning models for the prediction of infectious and thrombotic events in asplenic patients. To this extent, clinical data from 1,800 patients followed at 42 centers of the “The Italian Network for Asplenia (INA)” were collected. Before training the predictive model, an intensive pre-processing phase of data curation was performed, including trasformation, missing data handling, statistical distribution analyses and association test. During the model training, feature importance was calculated to improve the model’s performance and interpretability. This preliminary study showed promising results in getting new insights for predicting the risk of infectious and thrombotic events in asplenic patients. Implementing predictive models based on relevant clinical features could provide physicians with a useful tool in the preventive management of asplenic patients. This would allow for timely interventions and improve prognosis.