AI and Rare Diseases ── Rare blood disorders ── 2026-11-04
Rare blood disorders: treatment, trials and AI
- Abstract
Objective: To develop, internally validate and compare machine learning models for classifying kidney dysfunction in children aged ≤18 years with sickle cell disease (SCD), where kidney dysfunction may occur during childhood and remain clinically unrecognised.Design Retrospective study.Setting Paediatric Sickle Cell Clinic at Mulago National Referral Hospital and the Uganda Sickle Pan Africa Research Consortium registry.Participants A total of 893 children aged ≤18 years with confirmed SCD and complete renal function records were included.Main outcome measures Kidney dysfunction was defined as an estimated glomerular filtration rate (eGFR)<90 mL/min/1.73 m², while eGFR≥90 mL/min/1.73 m² was classified as normal kidney function. Random forest (RF), support vector machine (SVM) and extreme gradient boosting (XGBoost) models were developed using clinical and laboratory variables. Model performance was assessed using stratified fivefold cross-validation, with balanced accuracy, macro F1-score and area under the receiver operating characteristic curve (AUROC) as the main evaluation metrics.Results Among 893 participants, 488 (54.6%) had normal kidney function and 405 (45.4%) had kidney dysfunction. RF and XGBoost demonstrated comparable classification performance. RF achieved a balanced accuracy of 0.80 and AUROC of 0.86 (95% CI 0.84 to 0.89), while XGBoost achieved a balanced accuracy of 0.79 and AUROC of 0.87 (95% CI 0.85 to 0.89). SVM showed lower performance, with a balanced accuracy of 0.62 and AUROC of 0.69 (95% CI 0.66 to 0.72). SHAP (SHapley Additive exPlanations) analysis identified urine albumin and creatinine ratio as the strongest contributors to classification.Conclusion Machine learning models using routinely collected clinical and laboratory data showed potential for classifying kidney dysfunction in children with SCD.
Journal IF-equivalent: 0.0 (OpenAlex 2-year mean citedness, value as of 2026-11-03, retrieved 2026-11-04; not the official Clarivate IF)Reference: Esutu NE, Agaba I, Nakku-Joloba E, Kiguli S, Namazzi R, Mukose AD, et al. Development and internal validation of machine learning models for classification of kidney dysfunction in children aged ≤18 years with sickle cell disease at Mulago National Referral Hospital, Kampala, Uganda: a retrospective study. bmj connections digital health & ai. 2026 Nov 2;1(1):e000027. doi:10.1136/bmjcdh-2026-000027.Checked: Abstract only