El Dipòsit Digital ha actualitzat el programari. Contacteu amb dipositdigital@ub.edu per informar de qualsevol incidència.

 
Carregant...
Miniatura

Tipus de document

Treball de fi de grau

Data de publicació

Llicència de publicació

cc-by-nc-nd (c) autor, 202X
Si us plau utilitzeu sempre aquest identificador per citar o enllaçar aquest document: https://hdl.handle.net/2445/221864

Use of Machine Learning and SNOMED CT Encoded Health Problems to Predict Hospital Discharge Diagnoses

Títol de la revista

ISSN de la revista

Títol del volum

Resum

The accurate classification of discharge diagnoses is a critical step in clinical decision-making, as it has direct effect on patient care, hospital management, and administrative tasks. Traditionally, diagnostic coding has been a manual and time-consuming process, typically done after a patient is discharged, which could lead to delays for subsequent processes such as billing, reporting, and care optimization. Recently, the Hospital Clínic de Barcelona has integrated a structured list of health problems coded in SNOMED CT into the Electronic Health Record (EHR) from the beginning of the patient’s hospitalization. This development has enabled the reuse of structured clinical data throughout the care process and has opened the door for predictive tools using Machine Learning (ML). The goal of this research is to determine whether there’s a significant relationship between reported health problems and the final ICD-10 discharge diagnoses. To explore this, data obtained from the Hospital Clínic de Barcelona was analysed, incorporating information from various clinical sources, such as demographics, laboratory results, prescriptions, and admissions records. Feature engineering was also carried out and methods based on decision trees, along with ANOVA tests, were used to identify the most relevant input variables. Subsequently, several supervised ML models, including Decision Trees (DTs), Random Forest (RF), and XGBoost were trained and evaluated. The best performing model, a Decision Tree classifier, achieved an accuracy of 69.8%, with a recall and F1-score of 0.68, and an AUC of 0.83. While no single variable served as a dominant predictor, the results show that health problems coded in SNOMED CT, combined with other clinical and

Descripció

Treballs Finals de Grau d'Enginyeria Biomèdica. Facultat de Medicina i Ciències de la Salut. Universitat de Barcelona. Curs: 2024-2025. Tutor: Barrios Montenegro, J. C. ; Director: Santiago Frid

Citació

Citació

CHEN, Cindy. Use of Machine Learning and SNOMED CT  Encoded Health Problems to Predict Hospital  Discharge Diagnoses. [consulta: 27 de novembre de 2025]. [Disponible a: https://hdl.handle.net/2445/221864]

Exportar metadades

JSON - METS

Compartir registre