EUVIMEDScopri EUVIMED
CrossrefScheda informativa

System Architecture and Operational Principles for a Configurable Clinical Decision Support System for Model-Informed Precision Dosing: Development, Verification, and Validation Study

Roberto Rigamonti, Yuan James Pétermann, Melvyn Herzig, Annie Ester Cathignol, Bruno Da Rocha Carvalho, Stellah George Mpagama, Bibie Said, Margaretha Laurent Sariko, Chantal Csajka, Monia Guidi, Yann Thoma

JMIR Medical Informatics · 2026-11-12 · 14 · e95203-e95203

DOI: 10.2196/95203 ↗
AbstractDati bibliografici

Abstract

Abstract Background Therapeutic drug monitoring (TDM) supports individualized pharmacotherapy, yet its implementation in routine care remains limited by operational constraints, data quality issues, and the need for expert interpretation. Model-informed precision dosing (MIPD) addresses several of these limitations through population pharmacokinetic modeling and Bayesian forecasting, but its effective deployment in clinical practice still requires human TDM experts. Objective This paper presents Tucuxi-CDSS, a configurable and open-source clinical decision support system (CDSS) designed to enable safe, standardized, and automated MIPD-guided dose individualization, with a particular focus on safeguarding against pre-analytical and data entry errors. Methods We detail the Tucuxi-CDSS system architecture, its data validation workflow, decision logic, and configuration mechanisms. The system is built around extensive configurability through XML-based drug files, language files, and reporting templates. Verification was performed through unit testing, system-level tests, and random input fuzzing, while validation was conducted by comparing CDSS outputs against those of a human TDM expert in an in silico study. Results The resulting system supports end-to-end MIPD workflows, from data import and automated error detection through dosage adaptation and clinical report generation. An implementation study for rifampicin dosing optimization in routine tuberculosis care is currently underway in Tanzania, and an illustrative use case with imatinib demonstrates the system’s covariate handling and reporting capabilities. We also briefly summarize the in silico validation against a human TDM expert across 60 virtual patients: The CDSS matched the expert in error detection, with numerically higher sensitivity (53% versus 45%) but lower specificity (83% versus 100%), and it achieved higher therapeutic target attainment (55% versus 37%; P =.046). These gains came at the cost of 4 detrimental interventions in the CDSS arm, against none for the expert. Conclusions Tucuxi-CDSS provides a modular, extensible, and open-source platform intended to extend structured, reproducible TDM oversight to settings where trained pharmacologists are scarce. Its full performance evaluation against a TDM expert is presented in a companion paper and supports future prospective studies on CDSS adoption in clinical practice.

Conoscenze adatte al tuo lavoro.

Dall’inizio gratuito alla ricerca bibliografica personale, fino alla Research Suite completa.

Visualizza tariffe →

Sono inclusi tutti i testi integrali degli editori?

No. Citazioni, abstract e testi integrali sono contenuti diversi. La disponibilità dipende dalla fonte, dall’accesso aperto e dalle licenze esistenti. Un piano EUVIMED non sostituisce diritti aggiuntivi degli editori.

Visualizza tariffe →