Using Machine Learning to Identify Risk Scores for Delirium Under Different Clinical Settings

Aug 4, 2026

Delirium is a common, acute neuropsychiatric syndrome frequently overlooked in hospitals because its presentation is diverse and fluctuates over time. Early identification is critical to enable timely interventions and reduce progression and related complications. The study develops a machine learning predictive algorithm for incident delirium using electronic health record (EHR) data, deriving an ICU delirium dataset from MIMIC, extracting important clinical features from AUBMC charts, and training multiple traditional and deep learning models to estimate delirium risk scores for hospitalized and ICU patients.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages82

Reading Time3 h 27 min

AccessFree to download

Acknowledgements

Abstract of the Thesis

Table of Contents

AMERICAN UNIVERSITY OF BEIRUT USING MACHINE LEARNING TO IDENTIFY RISK SCORES FOR DELIRIUM UNDER DIFFERENT CLINICAL SETTINGS by MARIAM AHMED SAFIELDIN A thesis submitted in partial fulﬁllment of the requirements for the degree of Master of Science to the Department of Computer Science of the Faculty of Arts and Sciences at the American University of Beirut Beirut, Lebanon April 2023 AMERICAN UNIVERSITY OF BEIRUT USING MACHINE LEARNING TO IDENTIFY RISK SCORES FOR DELIRIUM UNDER DIFFERENT CLINICAL SETTINGS by MARIAM AHMED SAFIELDIN Approved by: Dr. Wassim El-Hajj, Associate Professor Advisor Computer Science | Dr. Shady Elbassiouni, Associate Professor Computer Science | Member of Committee
| | --- | --- | Dr. Farid Talih, Associate Professor Member of Committee Psychiatry Date of thesis defense: April 28, 2023 AMERICAN UNIVERSITY OF BEIRUT THESIS RELEASE FORM Safieldin Mariam Ahmed Student Name: Last First Middle I authorize the American University of Beirut, to: (a) reproduce hard or electronic copies of my thesis; (b) include such copies in the archives and digital repositories of the University; and (c) make freely available such copies to third parties for research or educational purposes As of the date of submission of my thesis After 1 year from the date of submission of my thesis . After 2 years from the date of submission of my thesis . After 3 years from the date of submission of my thesis . May 5, 2023 Signature Date Acknowledgements First and foremost, I express my deepest gratitude to Allah for granting me the strength and determination to undertake and successfully complete my research. I extend my heartfelt appreciation to Dr. Wassim El-Hajj, my supervisor, for his unwavering support and guidance throughout the research project. Additionally, I am grateful to Dr. Farid Talih for his invaluable feedback and constructive criticism that signiﬁcantly improved the quality of my research. I would like to extend my appreciation to the psychiatry team for their collaboration and thought-provoking discussions that enriched my research experience. I would like to thank Dr. Shady El-Bassiouni for his insightful feedback and valuable suggestions that helped me reﬁne my work. I would like to extend my sincere gratitude to Mrs. Abeer Elfawal for her unwavering support and steadfast belief in me. I am also deeply grateful to Miriam Ayres and Dr. Robert Myers for their invaluable encouragement throughout this journey. I would also like to thank my dear friends, especially Godiolla, for their unwavering encouragement. Their friendship has been a constant source of motivation. Furthermore, I would like to express my sincere appreciation to the Mastercard Foundation for their generous funding and support of my Master’s degree program. I am grateful for the opportunities they have provided me. To my beloved siblings, Karim and Manar, I am grateful for the unwavering love and support you have shown me throughout my academic journey. I am overwhelmed with gratitude as I express my heartfelt appreciation to my dear mother, whose love, care, and support have been a constant source of inspiration and motivation for me. She has always been my guiding light, helping me to navigate the upsand downs of life and providing me with the strength and courage to pursue my dreams. I owe an immeasurable debt of gratitude to my dearest brother, Dodo, whose unwavering support and constant motivation have been instrumental in my academic achievements. I owe my ability to continue my educational pursuits to him. His belief in me has never wavered, and his constant encouragement and support have propelled me forward, even in the face of the greatest challenges. I am blessed beyond measure for every second I experienced life with him, and I will forever cherish the immeasurable impact he has had on my life. Abstract of the Thesis of Mariam Ahmed Saﬁeldin for Master of

FAQs

What problem does the thesis address?

The thesis addresses the frequent under-recognition of delirium in hospitals and the need for early identification to prevent progression and complications.

How is the predictive model built?

A delirium dataset is derived from MIMIC data, important clinical features are extracted from EHR charts from AUBMC, and multiple machine learning models are trained to predict delirium risk scores.

Which model performed best according to the results?

The CatBoost model shows the highest performance compared with other trained models.

Aug 4, 2026

Upload by Theodore