Intelligent Sensor-array for the Electrochemical Analysis of Hydroxyurea in Blood Samples of Children Affected by Sickle Cell Anemia

Sickle cell anemia (SCA) is an inherited blood disorder that leads to morbidity and early mortality. Hydroxyurea is an effective oral medication to treat SCA, and optimal dosing benefits from pharmacokinetic (PK)-based methods that require accurate analysis of hydroxyurea levels in timed patient blood samples. Current gold standard assay methods such as liquid chromatography-mass spectroscopy (LC-MS) require sophisticated instrumentation, trained personnel, and laborious sample pretreatment steps that may alter target molecule levels. Additionally, LC-MS is time-consuming and costly, leading to delays in treatment decisions, and is not feasible for use in low-resource settings. Herein, we report a novel approach for chemical analysis of serum hydroxyurea levels by using an array of intelligent electrochemical microsensors modified with thin films of nanomaterials to record the electrochemical signature of blood samples from 50 children treated with hydroxyurea. To analyze this complex dataset, multiple machine learning models were trained and optimized to predict hydroxyurea levels from microliter sample volumes. We evaluated three regression algorithms: elastic net, random forest (RF), and partial least squares regression (PLSR) across 11 different electrochemical feature matrices. Among these, PLSR demonstrated the best performance, achieving a root mean square error of 41.85 µM (3.18 μg/mL) and average prediction error of 7.24 µM (0.55 µg/mL), thereby enabling accurate analysis of hydroxyurea levels within the therapeutic range using microliter sample volumes from pediatric patients. With further miniaturization of such sensor arrays and integration into point-of-care testing devices, hydroxyurea PK-based dosing can be simplified for use in low-resource settings to improve SCA treatment outcomes worldwide.