Predictive and Prescriptive Analytics based on Big Data for the Management of High Blood Pressure (#902)

Authors

Alvarado Diaz, Witman Joel

Meneses Claudio, Brian Andreé

Abstract

High Blood Pressure (HBP) is one of the leading causes of cardiovascular disease, premature mortality, and catastrophic health expenditure in Latin America, especially when it coexists with diabetes mellitus and progresses to chronic kidney disease (CKD). Despite the widespread availability of effective and low-cost antihypertensive treatments, traditional management models, reactive and focused on isolated clinical episodes, have shown a limited capacity to contain the lack of control and its clinical and financial consequences. That is why our objective is to develop a predictive model of risk of hypertensive decontrol and nephrological progression, and to propose a prescriptive analytics approach to optimize the allocation of financial resources in health. A retrospective study is carried out analyzing 24.2 million transactional records of medical care, pharmacy and diagnoses corresponding to 1.02 million patients treated between 2022 and 2024, using Big Data architecture based on Apache Spark. Different Machine Learning models were trained, and the Gradient Boosted Trees (GBT) model was chosen for the prediction of complications. As a result, we were able to identify a critical data quality gap. The predictive model reached a recall of 89.74%. Prescriptive analysis demonstrated that preventive intervention reduces cost compared to delayed renal treatment. We conclude that the sustainability of health systems requires a transition from reactive models to prescriptive nephroprotection strategies, supported by advanced analytics and systematic auditing of clinical data.