REINFORCEMENT LEARNING-DRIVEN TREATMENT OPTIMIZATION FOR HYPERTENSION MANAGEMENT IN PRIMARY HEALTHCARE

Hypertension remains one of the leading contributors to cardiovascular morbidity and mortality worldwide, with treatment effectiveness highly dependent on personalised and continuous adjustment of therapeutic strategies. Traditional rule-based clinical guidelines often fail to account for patient-specific variability, behavioural patterns, lifestyle factors, and real-time physiological responses. This study proposes a Reinforcement Learning (RL)-driven framework for optimising hypertension management in primary healthcare settings. The model conceptualises treatment planning as a sequential decision-making process, where an RL agent learns optimal medication adjustments, lifestyle recommendations, and monitoring intervals through continuous interaction with patient profiles and clinical outcomes. Using data derived from electronic health records, wearable blood pressure monitors, and behavioural logs, the RL agent is trained to minimise systolic/diastolic blood pressure while reducing adverse drug events and improving long-term treatment adherence. The proposed system integrates an Explainable RL layer to enhance transparency and clinical acceptance by providing human-interpretable justification for treatment recommendations. Preliminary simulation results demonstrate that RL-based policies outperform standard clinical protocols in reducing time-to-control, improving stability of blood pressure levels, and personalising interventions based on individual patient trajectories. This work contributes a scalable, data-driven, patient-centric approach to hypertension management, demonstrating the transformative potential of RL in primary care. Future work will focus on real-world validation, integration with IoT-enabled monitoring devices, and usability evaluations with healthcare practitioners.