[융합과학부] A Dimensionality Reduction Model to increase the efficiency and accuracy of clinical trial feasibility assessment using electronic medical records

Speaker: 이형기 교수(융합과학기술대학원)

Date & Time: 2018. 12. 4.(화), 17:00

Where: 융대원 D-122호

Abstract

Identifying eligible patients is arguably the most crucial part of the clinical trial process, yet is the most challenging. A Forte’s 2017 recent national survey1 found that recruiting potential patients to a trial was ranked in the top three pain tasks, due to an increase in the number and complexity of the eligibility criteria (EC) as well as screening procedures. Hence, it is important to assess clinical trials feasibility and distinguish the EC that are likely to hinder patient recruitment. Afterward, the EC can be modified or removed to meet the target accrual rate. Given ongoing challenges in trial recruitment, we propose a dimensionality reduction model to identify the common sensitive eligibility features (cSEFs) that could increase the efficiency of clinical trials feasibility in case where a certain EC could not be mapped out against electronic medical records (EMR). Using real-world patient data from EMR, our model allows us to calculate the relative importance of feasibility (RIF) for an individual eligibility criterion to determine cSEFs in a chosen disease area (i.e., essential hypertension). RIF, defined as ‘one minus the ratio of the number of patients meeting the total set of EC to the number of patients meeting a subset of the EC after removing one’, indicates how influential an eligibility criterion can be if not considered for clinical trials feasibility. Feature selection algorithms (i.e., embedded methods) in machine learning will be implemented to select the best subset of cSEFs for their correlation with the RIF. Selected cSEFs will be tested with EMR data to validate if they could adequately return a sufficient number of eligible patients.

Keyword

eligibility criteria, electronic medical records, dimensionality reduction model, clinical trial, clinical trials feasibility, feature selection, machine learning

Reference

1. “State of the Clinical Research Industry Report.” Forte; 2017.

Biography

Dr. Howard Lee is the Founder and Director of the Center for Convergence Approaches in Drug Development (CCADD). Dr. Lee serves as a Professor at the Department of Transdisciplinary Studies, Graduate School of Convergence Science and Technology, Seoul National University. Dr. Lee is also appointed at Seoul National University College of Medicine and Hospital, affiliated with the Department of Clinical Pharmacology and Therapeutics. Dr. Lee previously served as Head of Global Strategy and Planning, Clinical Trials Center, SNUH. As of August 2017, Dr. Lee was appointed Chair, the Interdisciplinary Graduate Program in Clinical Pharmacology, Seoul National University.

Dr. Lee is a board-certified physician in Family Medicine and Clinical Pharmacology and holds a PhD in Epidemiology. Dr. Lee completed a fellowship in Clinical Pharmacology at the Center for Drug Development Science (CDDS), Department of Pharmacology, Georgetown University School of Medicine in Washington DC, USA, where he subsequently served as an Assistant Professor of Medicine and eventually became Director after CDDS has joined the University of California San Francisco in association with the School of Pharmacy. Dr. Lee also had served as faculty in two other US-based universities including the University of California San Francisco and the University of Pittsburgh.

Dr. Lee’s professional career consists of an interesting mix of academia, industry, and government. Before he went to the US, Dr. Lee had a 5-year real-world experience in clinical drug development at two pharmaceutical companies as Medical Director. Dr. Lee also was a Guest Researcher at the Office of Clinical Pharmacology, Center for Drug Evaluation and Research, US Food and Drug Administration.

 

초청자 : 융합과학부 지능형융합시스템전공 박재흥 교수 (park73@snu.ac.kr)

2018-11-29|