PSM in Biomedical and Health Informatics Program – Internship Presentations

Students in the Professional Science Master’s in Biomedical and Health Informatics Program complete an internship that synthesizes knowledge gleaned from the program curriculum. The purpose of the internship is to expand classroom learning to include “hands-on” experience in health IT.

In this occasion Roshan John, Jeff Stocker, Eric Cui, and Aaron Bohlmann will be presenting their internship projects via Zoom.

Tiltles and abstracts below:

Assessing telehealth modalities and their association with health disparities and health access

Presenter: Roshan John

Abstract: This inaugural internship project provided firsthand experience in health informatics research and involved examining telehealth usage patterns across modalities. Deliverables developed during the project included an introductory literature review on existing telehealth studies and data analysis of a telehealth dataset for a study paper. This was done at CHIP @ UNC in conjunction with Dr. Saif Khairat and Malvika Pillai, as part of the manuscript writing process. The research focused on comparing phone and video modalities of telehealth during March and April 2020 within the UNC Virtual Urgent Care system. It produces insight into associations between modality and patients with differing health access and demographics. The internship spanned from May through July 2020 as an all remote experience, which provided opportunities to develop and sustain independent work cycles throughout the project. The internship was a meaningful first exposure to working with programmatic tools in health informatics and a foundational experience in assessing health data to build on for future experiences in the field.

AnalyzeMed: A pharmacogenomics information system 

Presenter: Jeff Stocker

Abstract: Genetic makeup can predispose individuals to a variety of diseases, but it can also lead to adverse reactions with prescription medications. Pharmacogenomics examines the interaction between a person’s genetic variance and their response to drugs. Historically, these drug-gene studies have predominantly surveyed non-Hispanic white patients, leaving minority groups largely underserved. AnalyzeMed, a web application that allows end-users to query African-American pharmacogenomic research, aims to reduce this disparity. Utilization of this application and other pharmacogenomic information systems can help guide better patient-provider communication, facilitate research, and help create more personalized treatments.

Building an Online Healthcare Data Analytics Training Program in the COVID-19 Era

Presenter: Eric Cui

Abstract: Increasing and diversifying the health informatics workforce has been identified as a priority to meet the nation’s growing healthcare needs. In response, the Extensible Network-Accessible Biomedical & Health Informatics Lifelong Learning Environment (ENABLE) project was created by the Carolina Health Informatics Program (CHIP) to train students and professionals from diverse backgrounds in biomedical and health informatics. The traditional program consisted of in-person summer boot camps alongside supplemental online resources. However, due to COVID-19 restrictions, ENABLE transitioned into an online-only curriculum this summer. This presentation will provide an overview of building an online healthcare data analytics training program for ENABLE. Specifically, this presentation will detail the process of building the online environment, curriculum development, student evaluation, and discuss the program’s performance and future implications.

Using Keywords and Index Terms to Improve Clustering and Cluster Labels

Presenter: Aaron Bohlmann

Abstract: Recently I finished a rough draft of a scoping review centered around medication adherence and machine learning. As part of this review I categorized all of the included sources into natural categories. This corpus of categorized literature provided me with an excellent opportunity to test out possible ways of improving clustering and cluster labels for the PATTIE corpus analysis tool. Since the back end of the system is not readily accessible to me, I decided to approach this problem by making changes to both the search terms and the corpus. By including an exhaustive list of key words and indexing terms related to the main search terms I was able to get the system to produce longer and potentially more informative cluster labels. Adding keywords and indexing terms to the corpus resulted in a higher purity score for clustering by abstract but did not affect purity score of documents clustered using full text. Clustering by keywords and indexing terms alone was also attempted but resulted poor clustering and uninformative cluster labels.