RESUMEN
Applying evidence-based medicine prevents medical errors highlighting the need for applying Clinical Guidelines (CGs) to improve patient care by nurses. However, nurses often face challenges in utilizing CGs due to patient-specific needs. Developing a Clinical Decision Support System (CDSS) can provide real-time context-sensitive CG-based recommendations. Therefore, there is a need to acquire and represent CGs in a machine-applicable manner. Also, there is a need to be able to provide recommendations episodically, only when requested, and not continuously, and to assess previous partial performance of evidence-based actions on a continuous scale. This study evaluated the feasibility of acquiring and representing major nursing CGs, in a machine-applicable manner for episodic use. Using data from an Israeli geriatric center, the results suggest that an episodic CDSS effectively supports the application of formalized nursing knowledge.
Asunto(s)
Sistemas de Apoyo a Decisiones Clínicas , Enfermería Basada en la Evidencia , Guías de Práctica Clínica como Asunto , Israel , Humanos , Medicina Basada en la EvidenciaRESUMEN
Medical errors contribute significantly to morbidity and mortality, emphasizing the critical role of Clinical Guidelines (GLs) in patient care. Automating GL application can enhance GL adherence, improve patient outcomes, and reduce costs. However, several barriers exist to GL implementation and real-time automated support. Challenges include creating a formalized, machine-comprehensible GL representation, and an episodic decision-support system for sporadic treatment advice. This system must accommodate the non-continuous nature of care delivery, including partial actions or partially met treatment goals. We describe the design and implementation of an episodic GL-based clinical decision support system and its retrospective technical evaluation using patient records from a geriatric center. Initial evaluation scores of the e-Picard system were promising, with a mean 94% correctness and 90% completeness based on 50 random pressure ulcer patients. Errors were mainly due to knowledge specification, algorithmic issues, and missing data. Post-corrections, scores improved to 100% correctness and a mean 97% completeness, with missing data still affecting completeness. The results validate the system's capability to assess guideline adherence and provide quality recommendations. Despite initial limitations, we have demonstrated the feasibility of providing, through the e-Picard episodic algorithm, realistic medical decision-making support for noncontinuous, intermittent consultations.