Therapy management for patients with congestive heart failure (CHF) requires individualized adjustments to drug dosages based on key physiological indicators (body weight, heart rate, and blood pressure) that reflect a patient’s evolving clinical state. This study presents a fuzzy logicbased decision support system which acts as an offline supervisory controller for weekly medication tuning in home-based CHF care. The system encodes cardiologist expertise into 80 fuzzy rules, enabling interpretable mapping from clinical inputs to therapy set-points for diuretics, betablockers (β-blockers), and angiotensin-converting enzyme (ACE) inhibitors. Operating on discrete-time cycles analogous to batch chemical processes, the system bridges intervals between specialist consultations while ensuring clinical guideline adherence and responding to physiological deviations. Initial validation against independent cardiologist prescriptions demonstrated over 95% concordance, confirming expert therapeutic reasoning replication. To enhance personalization, the architecture integrates an adaptive Radial Basis Function (RBF) layer that stores historical patient responses and computes Gaussian-weighted corrections to fuzzy outputs, enabling alignment with individual treatment trajectories while preserving safety and transparency. Unlike previous CHF decision-support tools (which either run continuously or rely on opaque black-box predictors), this approach introduces the first offline supervisory controller coupling an interpretable fuzzy-rule core with a transparent adaptive layer, delivering patientspecific yet fully auditable dosage adjustments.
Offline supervisory fuzzy control for weekly set-point adjustment in home-based congestive heart failure therapy
Bartolomeo Cosenza
;Gabriele Pannocchia
2026-01-01
Abstract
Therapy management for patients with congestive heart failure (CHF) requires individualized adjustments to drug dosages based on key physiological indicators (body weight, heart rate, and blood pressure) that reflect a patient’s evolving clinical state. This study presents a fuzzy logicbased decision support system which acts as an offline supervisory controller for weekly medication tuning in home-based CHF care. The system encodes cardiologist expertise into 80 fuzzy rules, enabling interpretable mapping from clinical inputs to therapy set-points for diuretics, betablockers (β-blockers), and angiotensin-converting enzyme (ACE) inhibitors. Operating on discrete-time cycles analogous to batch chemical processes, the system bridges intervals between specialist consultations while ensuring clinical guideline adherence and responding to physiological deviations. Initial validation against independent cardiologist prescriptions demonstrated over 95% concordance, confirming expert therapeutic reasoning replication. To enhance personalization, the architecture integrates an adaptive Radial Basis Function (RBF) layer that stores historical patient responses and computes Gaussian-weighted corrections to fuzzy outputs, enabling alignment with individual treatment trajectories while preserving safety and transparency. Unlike previous CHF decision-support tools (which either run continuously or rely on opaque black-box predictors), this approach introduces the first offline supervisory controller coupling an interpretable fuzzy-rule core with a transparent adaptive layer, delivering patientspecific yet fully auditable dosage adjustments.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
