Machine learning algorithms to predict the 1 year unfavourable prognosis for advanced schistosomiasis

Area of Science:

• Medical Informatics
• Computational Biology
• Public Health

Background:

• Short-term prognosis for advanced schistosomiasis remains understudied.
• Routine data from government medical assistance programs can be leveraged for prognostic modeling.

Purpose of the Study:

• To develop machine learning models for predicting 1-year prognosis in advanced schistosomiasis.
• To identify key predictors of unfavorable outcomes using routinely available data.

Main Methods:

• Utilized a database of 9541 advanced schistosomiasis patients (2008-2018).
• Applied five machine learning algorithms: logistic regression (LR), decision tree (DT), random forest (RF), artificial neural network (ANN), and extreme gradient boosting (XGBoost).
• Evaluated model performance using the area under the receiver operating characteristic curve (AUC).

Main Results:

• 13.1% of patients experienced unfavorable prognoses within 1 year.
• XGBoost achieved the highest AUC (0.846), outperforming other models.
• Key predictors for unfavorable prognosis included ascitic fluid volume, hemoglobin (HB), total bilirubin (TB), albumin (ALB), and platelets (PT).

Conclusions:

• XGBoost is a highly effective algorithm for predicting 1-year unfavorable prognosis in advanced schistosomiasis.
• The developed model provides a simple, useful tool for clinical short-term risk assessment.
• Identifying key predictors can aid in early intervention and patient management.