Machine Learning and Multivariate Analysis of Hemorheological Dysfunction and Clinical Outcomes in HbSS Patients with Malaria Co-Infection

Background: Malaria co-infection may aggravate hemorheological abnormalities and clinical morbidity among individuals with homozygous sickle cell disease (HbSS). However, the relationships between malaria status, blood rheology, hematological characteristics, and clinical outcomes remain incompletely characterized. This study assessed these relationships and evaluated the predictive utility of statistical and machine-learning approaches. Methods: A retrospective cross-sectional analytical study included 45 laboratory-confirmed HbSS patients, comprising 37 malaria-positive and 8 malaria-negative individuals. Demographic, parasitological, hematological, hemorheological, and clinical variables were analysed using Welch’s t-test, correlation analysis, multivariable regression, K-means clustering, principal component analysis (PCA), and machine-learning models. The analytical workflow was implemented in Google Colaboratory using Python. Results: Malaria-positive patients had higher blood viscosity (4.80 ± 0.34 and 3.80 ± 0.43 mPa; p < 0.001) and lower erythrocyte deformability (0.88 ± 0.03 and 0.92 ± 0.02 mPa; p < 0.001) than malaria-negative patients. They also had lower haemoglobin and haematocrit, higher reticulocyte and white blood cell counts, more frequent pain crises, and longer hospitalization. Multiple regression identified pain-crisis frequency and blood viscosity as significant predictors of hospitalization duration. K-means clustering identified three patient groups with progressively higher viscosity and lower deformability. PCA showed that two components explained 72.4% of total variance. Blood viscosity discriminated malaria status with an AUC of 0.94 (95% CI: 0.86-0.99); at 4.25 mPa·s, sensitivity and specificity were 91.9% and 87.5%, respectively. Random Forest regression achieved R2 = 0.72 for hospitalization duration. Conclusion: Malaria co-infection was associated with greater hemorheological impairment and clinical burden among HbSS patients. These findings support further evaluation of hemorheological biomarkers and validated predictive models in larger longitudinal cohorts.