Healthcare and technology are transforming patient care by leveraging data-driven insights that enhance decision-making and improve outcomes. At the Department of Biomedical Informatics (DBMI), we are doing the same by bringing bytes to the bedside with our weekly seminar series. This month, we explored software labyrinths, the use of infographics in HIV care, and genomic sharing in understudied populations. Engaging discussions with leading experts revealed how navigating complex problems can lead to groundbreaking advancements in treatment.
Navigating software like a dungeon crawl
Dave Bunten, a research data engineer at DBMI and a self-professed fan of computer games, drew a vivid analogy between software development and navigating a Roguelike dungeon filled with "surprises, risks, and hidden traps." He explained that each piece of software is akin to an adventure, where "every room offers either treasure or peril."
Roguelike computer game from Citrocipia, Wikimedia
As developers delve into various components, the outcomes can vary significantly with any changes to data or code. He emphasized the importance of preparation, stating that tackling challenges without it can lead to "significant obstacles." Just as "lanterns are necessary to illuminate dark corners of a dungeon," Bunten highlighted that software testing is crucial for guiding developers through the complexities of the development process.
Image from Bunten highlighting production code and test code
Bunten noted that testing ensures software meets specific expectations, acting as a beacon to identify potential issues. He underscored the necessity of distinguishing between production and test code, where production code "accomplishes tasks" while test code "checks the functionality" of the production code.
Images from Bunten. The top image shows creating or moving test scripts to test frameworks, which are versioned with production code. The bottom image shows that you can think of this as if you were moving useful tools into your inventory as part of your software adventure.
For illustrative purposes, he introduced pytest, an external package that can be installed with pip install pytest. In Python development, organizing production code in a src directory and test code in a tests directory is common. By default, pytest searches for tests within the tests directory and for modules prefixed with test_. He explains that assert statements are pivotal in Python testing, allowing verification of output correctness.
As developers write test code, Bunten advised categorizing tests into different types:
- Unit Tests validate small, isolated parts of code, like individual functions or methods.
- Integration Tests ensure that various software components function together seamlessly, which is particularly vital in scientific software where different models, algorithms, and data structures interact.
- System Tests evaluate the software from a user's perspective, often involving complete workflows or simulations to ensure smooth operation from data input to output.
He also mentioned static analysis, which assesses the software's structure without execution. "Linters, as static analysis tools, are excellent for addressing specific practices and ensuring a baseline level of code quality," he stated, all without requiring additional tests.
The DBMI software engineering team publishes blog posts to help faculty and staff gain insights into software development, implementation, and management. The team aims to engage wider audiences in order to address questions related to the field of software development.
Visualizing health for everyone
Samantha Stonbraker, PhD, MPH, RN, FAAN, assistant professor of nursing, recently joined DBMI as a secondary faculty member. During the seminar she presented on Oct. 24, she shared insights on developing and evaluating digital interventions to improve HIV treatment and prevention for vulnerable populations. She emphasized that her work occurs in "diverse multicultural and multilingual settings with limited resources," particularly focusing on safety net hospital systems and the Dominican Republic.
In her presentation, Stonbraker highlighted that Latinos face significant HIV-related health disparities, including "higher prevalence, delays in diagnosis and treatment, faster disease progression, and higher mortality rates" compared to their white counterparts. In the Dominican Republic, these challenges are also pronounced due to low economic status and health literacy, which complicate navigation of the healthcare system and communication with providers. She stressed the importance of well-designed infographics to make information delivery "visually appealing and culturally relevant." By embedding her infographic intervention into digital technologies, her goal is to enhance accessibility and use in diverse settings locally and globally.
In the Dominican Republic, around 70,000 people are living with HIV, with Clínica de Familia La Romana providing care for approximately 17,000 patients. Stonbraker began her project by designing infographics tailored for people with HIV (PWH) and their healthcare providers. This process included several preliminary studies to identify the content of the infographics, followed by a "rigorous, iterative participatory design methodology" that incorporated feedback on infographic prototypes from PWH, clinicians, and visualization experts. Her research assessed the feasibility, acceptability, and clinical utility of the infographics through qualitative methods.
Stonbraker later recruited 50 PWH to the first pilot study of infographic use at Clínica de Familia, who were then exposed to the infographic intervention during three study visits over three months and completed questionnaires at each visit. The results indicated "statistically significant improvements" in various aspects, including HIV-related knowledge and self-reported health status. Participants reported positive responses to the infographics.
Following this, Stonbraker adapted her infographics for the New York City context, recruiting 30 Latino PWH who participated in a similar pilot study. In-depth interviews were held with participants who spoke either English or Spanish at the end of the study to explore their perspectives on infographic use.
Following positive results from the initial pilot studies, Stonbraker and her team conducted an implementation study of infographic use across the Dominican Republic. Feedback from participants suggested that the training method for using the infographics was effective and useful for widespread scale-up, while program-level data from over 4,500 PWH indicated there were higher viral loads and more visits attending following program-wide infographic use.
Example of an infographic titled "Tips to help with adherence"
Stonbraker also incorporated her infographics into an app called "Info Viz for Health®," powered by Cliexa, which features several interactive elements and language options. It’s designed to be provider-facing and adhere to SMART App standards for seamless healthcare system integration through FHIR. Stonebreaker is now working to integrate the app into various electronic health record (EHR) systems in the US.
For more information on her project and/or ideas for collaboration, please do not hesitate to contact Stonbraker at Samantha.Stonbraker@CUAnschutz.edu. Stonbraker's seminar marked the first DBMI seminar offering continuing medical education credits. Two additional CME seminars are available this semester.
Diving into genealogical data
During October, we also welcomed Jazlyn Mooney, PhD, as our speaker for the Precision Health & Genomics: Indigenous Mentoring and Ethics (PrIME) series in collaboration with our weekly Bytes to Bedside seminar. Mooney is an assistant professor of quantitative and computational biology at the University of Southern California (USC).
During her visit, Mooney spent several days on campus with Katrina Claw, PhD, associate professor of biomedical informatics, and her lab team. They participated in the Centers for American Indian and Alaska Native Health (CAIANH) Open House, featuring a traditional storyteller, a powwow, artwork, and a local vendor offering Indian tacos. After her seminar, Mooney hosted a meet-and-greet with our postdoctoral trainees, creating a space for questions and networking.
The Claw Lab with Jazlyn Mooney, PhD, at the CAIANH Open House. Image courtesy of Katrina Claw, PhD
Mooney’s research focuses on African American genealogical ancestors, a subject she finds personally meaningful as it allows her to explore her own family history. She aims to analyze genealogical data to ascertain "when and where individuals in a family tree originated."
Her study differentiates between genealogical ancestors—those documented in family trees—and genetic ancestors, who share genetic material with an individual. She seeks to determine if current ancestry proportions can be used to estimate the number of genealogical ancestors from various source populations contributing to the genome of an admixed individual.
To achieve this, Mooney developed a flexible model that simulates the admixture process over discrete generations. The model allows her to estimate admixture rates and produces outputs that reflect the present-day population’s ancestry distribution and genealogical ancestors.
Ultimately, her goal is to deepen our understanding of African American history, focusing on the period from 1619 to 1865. However, the model could be applied to any population to understand ancestry contributions from each parental source population in the past.
Visit our events page to attend an upcoming DBMI Bytes to Bedside Seminar.
