Professional Documents
Culture Documents
The Role of The University Library in Creating Inclusive Healthcare Hackathons: A Case Study With Design-Thinking Processes
The Role of The University Library in Creating Inclusive Healthcare Hackathons: A Case Study With Design-Thinking Processes
Article
International Federation of
Library Associations and Institutions
The role of the university library 2019, Vol. 45(3) 246–253
ª The Author(s) 2019
in creating inclusive healthcare Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/0340035219854214
hackathons: A case study with journals.sagepub.com/home/ifl
design-thinking processes
Bethany McGowan
Purdue University, Indiana, USA
Abstract
Librarians can utilize design-thinking practices to develop instructional materials, in the development of new
products and services, and in prototyping novel solutions to problems. This paper will explore the role of
design thinking in teaching and learning via the use of the Blended Librarians Adapted Addie Model (BLAAM),
and will illustrate how well-designed learning approaches can be used to create inclusive learning
environments. It will present a case study showcasing how an academic health sciences librarian utilized a
design-thinking process to create a health data literacy instruction service that encourages diverse participation
in healthcare hackathons.
Keywords
Data literacy, design thinking, diversity, hackathons, health
environment for health sciences students interested in can overcome such challenges to design care
working with large biomedical datasets. responses based on data generated from the healthcare
systems in which they work and in which their patient
data was originally created, can provide more tailored
Healthcare hackathons as data literacy care, improve their understanding of the effectiveness
instruction opportunities of treatments, and improve the prediction of health
Hackathons offer a means to gather an interdisciplin- outcomes (Schneeweiss, 2014). This ability to trans-
ary group of participants to solve a specific real-world form data into actionable knowledge can supplement
problem and learn from one another. They encourage the information provided by research studies, where
participants to utilize design-thinking approaches to patient populations might be different or which might
innovate, require a successful pitch that convinces have been conducted in a controlled research environ-
stakeholders of the usability of a solution, and require ment (Schneeweiss, 2014). In his exploration of learn-
participants to quickly respond to feedback and adjust ing from big biomedical data, Professor of Medicine
their proposed solutions accordingly (Silver et. al., and Epidemiology Sebastian Schneeweiss suggests
2016). Though hackathons have roots in the technol- that health care professionals should understand how
ogy industry, they have been embraced by academia to analyze the data being collected in their local health
and by libraries (Burton et. al., 2018) as a vehicle for care systems to inform patient care, using the same
inquiry-based and student-centered learning, and have hospital information systems that already generate
been used in health-related disciplines as a means to that data. Schneeweiss offers a few considerations
encourage students to be both information seekers and to facilitate learning from such health data outputs:
knowledge creators as they find and analyze biome- 1. Methods should ensure that the patient groups
dical data to create an end-product (Camacho et. al., used in comparison groups are similar;
2018; Decker et. al., 2015; Duhring, 2014; Kienzler 2. Aspects of analyses should be automated with-
and Fontanesi, 2017). To make it easier to describe out losing validity;
hackathons to healthcare professionals – healthcare 3. Automated analyses should be repeated in
administrators, physicians, nurses, etc. – unfamiliar cycles synced to data refreshes;
with the concept but whose buy-in would be needed 4. Software solutions should be intuitive and
to support such events, Silver et. al. (2016: 177) require only minimal investment of time and
define the term healthcare hackathon as: training to set-up and use;
5. Analysis results should be easily digestible for
a competitive event (live or virtual) that has three spe-
both busy clinicians and patients.
cific goals – accelerating the innovation of medical solu-
tions, improving the design in the beginning stages, and These considerations are useful for hackathon partici-
supporting educational training for all participants – and pants designing software applications and for judges
aims to accomplish them by focusing on a specific prob-
evaluating hackathon participant submissions. Schnee-
lem (pain point), bringing together in an open innova-
tion format (internal and external resources) an
weiss suggests that learning to treat patients with the
interdisciplinary group of individuals (hackers) that data that they themselves generate and understanding
include, but are not limited to, physicians and other the aforementioned considerations will lead health pro-
healthcare professionals, data scientists, engineers, user fessionals to actionable practices that improve patient
interface designers, business professionals, students and treatment and outcomes (Schneeweiss, 2014).
other stakeholders who work in teams and follow a pro-
cess to develop initial prototypes, pitch them to a panel
of judges experienced in innovation and quickly alter Applying design thinking to promote
them according to the feedback (pivoting). inclusion in hackathons: The BLAAM model
Criticisms of hackathons include that they promote a
Participation in healthcare hackathons helps cur- culture unwelcoming to women and underrepresented
rent and future health professionals identify, consider, minorities by:
and develop solutions to challenges associated with
how big biomedical data influences patient well- 1. fostering an overly competitive environment;
being, and hackathons are particularly useful for bet- 2. using a format or following a schedule that
ter understanding challenges associated with the potential participants might find unappealing,
underuse of uniform data standards and challenges i.e. staying up for days on end; and
associated with navigating siloed data (Schneeweiss, 3. lacking in diversity, which leads to feelings of
2014; Silver et. al., 2016). Health professionals who otherness. (Decker, et. al., 2015)
248 IFLA Journal 45(3)
data challenge offerings, extracurricular activities that women and members of underrepresented
would allow students to practice data skills, high- minority groups, audiences often underrepre-
lighted an emphasis on engineering, technology, and sented in data-hacking challenges.
computer science disciplines. The assessment phase 2. Behavior: By the end of the service, learners
focused on better understanding learner needs by should be able to locate relevant open-source
discussing anticipated outcomes and goals with sta- datasets, clean and organize raw data, combine
keholders, including health sciences faculty, instruc- data from multiple sources and in multiple for-
tors, and students. An informal needs assessment via mats, and use their cleaned and formatted data
casual conversation and reference interviews with to build an application using the Javascript
students suggested that health sciences students programming language.
thought data literacy was important but were not 3. Condition: To execute the learned behavior,
participating in the many data-hacking challenges learners should be comfortable searching the
going on in other areas of campus for various rea- web for open datasets, finding datasets in
sons, including discomfort in a competitive hacking library-provided databases, be able to select
environment and time restraints. At the end of a the appropriate tools to clean and combine data
librarian-led class discussion on research data man- in various formats, and have a beginner’s
agement, a group of health sciences students were knowledge of the Javascript programming
asked if they had considered participating in a data language.
hacking challenge. These two quotes summarize the 4. Degree: Learners will be assessed by the qual-
group’s response: ity of their project deliverable – a functioning
Javascript application prototype using at least
I learned Javascript because I think that knowing at least one open data source. They will also be
one programming language and knowing how to work assessed by the improvement of their applica-
with data will become even more important throughout tion over the course of the data-hacking event
my career, but I don’t think my skills are good enough to (for individual hacks) or by peer evaluation of
compete with, for example, someone getting a degree in
their contribution to their data hacking team
computer science. That’s why I haven’t attended any of
the hackathon events on campus. (quote from an under- (for team hacks).
graduate student majoring in a health sciences
discipline). Service objectives.
I don’t know how to code, no programming lan- 1. The proposed service aims to motivate health
guages at all, but I think understanding how to manage sciences students to participate in health data-
raw data is important. I would be interested in a class or hacking challenges by providing opportunities
workshop or special event, something that would teach complementary to the interests, skills, and
me to work with raw data despite my not having pro- experience of students and professionals in
gramming knowledge. (quote from an undergraduate
health-related professions.
student majoring in a health sciences discipline).
2. Participants will demonstrate an ability to
Based on the needs assessment, a problem state- execute data literacy-related competencies,
ment was developed: including the management of large biomedi-
How can librarians encourage health sciences stu- cal datasets, to create an application, service,
dent participation in health data-hacking challenges? or tool.
A healthcare hackathon service was conceptua-
Define objectives lized, and a sample hackathon challenge was created:
Bell’s A-B-C-D technique, as previously explained
under the “Define Objectives” phase of the BLAAM Scenario. Diseases and viruses are common in the US
model, was used to define service objectives: media landscape and public mindset; public aware-
ness, and knowledge tend to be easily affected and
1. Audience: The intended audience was directed by news stories. This often shapes a discourse
undergraduate students in health sciences dis- of fear. Imagine you are hearing the news about the
ciplines. Many of these targeted disciplines – return of a disease or virus (you can choose a specific
nursing, nutrition, and psychology, and virus) and you want to figure out the probability of it
speech, language, & hearing sciences – pro- spreading to your area. This could be based on geo-
vided an opportunity to recruit from a diverse graphy, proximity to cities, mountains, lakes, plains,
pool that included significant numbers of traffic connectivity, climate, airports, etc.
250 IFLA Journal 45(3)
Task. Using open data from the CDC and mapping it formed the heart of the hackathon service’s diversity
to environmental and travel conditions, build an commitment. The service, then, consisted of two inte-
application that will make the public more educated grated parts:
about the current state and impact of viruses so they
can make better-informed choices as they travel 1. pre-hakathon workshops that introduced the
inside the US. data literacy competencies needed to compete
in a health data hacking challenge; and
Mandatory requirements. 2. hacking events with iterative feedback that
1. Your application must use the mandatory data- allowed participants to practice and execute
set: Climate Data Online; the data literacy competencies learned in the
2. Your application must use at least one or more workshops.
datasets from Data.gov; Pre-event training opportunities included a Data
3. Your application must be written in the Java- Hacking Workshop Series that included research data
Script language; management workshops, an OpenRefine workshop
4. Your website must integrate a mashup; for cleaning and combining datasets, a Gephi work-
5. The code submission must be complemented shop for data visualization, and Javascript training.
with a readme file; The workshops encouraged participants to research
6. The readme file must have metadata: key- and develop the methods, software, and tools needed
words, a brief description, a unique identifier to analyze biomedical big data in advance of the hack-
of the author, a browser within which it was ing challenge. Participants were recruited from a
tested (must be Chrome); range of backgrounds through campus and commu-
7. Your application must use one of the suggested nity promotional activities – via campus activity fairs,
libraries: arbor.js, D3.Js, or sigmajs. through student organizations, and via recruitment in
This sample challenge was shared with col- entry-level computer science and health sciences
leagues for review, including fellow librarians with courses. To encourage diverse participation, outreach
data science interests, faculty in health sciences efforts were tailored to organizations like the regional
disciplines, and faculty with interest in open digital Clinical and Translational Science Institute (CTSI)
innovation. Their feedback was used to help visua- and campus cultural centers, as well as local civic
lize a challenge model. hacking organizations and makerspaces. A budget
was created and funding was sought, resulting in the
receipt of a National Institutes of Health (NIH) Big
Develop Data to Knowledge (BD2K) Hackathon Award, for
which the Librarian was the primary investigator.
After the needs assessment was conducted and a clear
idea of service objectives was defined, the librarian
used the BLAMM model to develop a plan for a new Deliver
health data-hacking service. A collaboration was After receiving the NIH BD2K Hackathon Award, the
forged between the Libraries and the Research Center proposed library hacking service was formally titled
for Open Digital Innovation, where a research team the Biomedical Big Data Hacking for Civic Health
had designed and tested a new hacking competition Awareness Project. To better meet the award require-
design appropriate for beginner hackers called Iron- ments, the target audience for the hackathons was
Hacks, three-week competitions where each partici- expanded to include graduate students, faculty, staff,
pant worked independently with integrated feedback local health professionals, and members of the public.
from technical and user experience experts (Bruns- The BD2K Hackathon Award funded two projects –
wicker et. al., 2017). A plan to collaborate to host a Black Ironhack in Fall 2016 and the Libraries Health
healthcare hackathon began with a recruitment strat- Data Hackathon in Spring 2017. Black Ironhack was
egy that encouraged the participation of novice devel- co-hosted by the Libraries and the Research Center
opers and data scientists by providing pre-hackathon for Open Digital Innovation and utilized a pre-
training that would allow all participants to meet the existing hackathon structure. The IronHacks model
basic level of coding and data science skills needed to is a 21-day, 4-stage self-directed hacking initiative
actively participate in the hack. This provision of that allows hackers the time and flexibility to create
introductory data literacy training to attract novice well-developed applications. At the end of each stage,
developers and data scientists combined with a self- participants receive feedback from experts and future
paced hacking model that provided iterative feedback app users. Black Ironhack asked participants to use
McGowan: The role of the university library in creating inclusive healthcare hackathons 251
open health and travel data to address health chal- between consumers and health information sources.
lenges, specifically to create a tracker for travelers Each team was assigned an early-career healthcare
interested in evaluating their risk of virus exposure. professional to advise on current challenges related
Participants were required to create a web application to working with biomedical data. There were 25 par-
and mashup (or combine) at least two open datasets. ticipants registered and 7 participants submitted final
Guidelines for the IronHacks are: projects, with the winning team including a nurse who
blended clinical experience, patient communication
1. Submit at least one application at each of the 4 knowledge, and computing expertise to develop a
hacking phases. pharmaceutical drug distribution app. The winning
2. Discuss general questions on the IronHack project codes and presentations are available on
forum only. Do not communicate solutions Github: https://github.com/anajaved/Heath-Data-
with competitors or others. Hackathon https://github.com/will1397/Health-Data/
3. Do not work in teams. IronHack is an individ- tree/master/HealthDataApp/app/src/main
ual competition.
4. The code you submit in the first phase does not
have to be a fully working application, it can Measure
be a non-interactive prototype (even an In his discussion on learning from big health care
image). data, Schneeweiss (2014) identified navigating siloed
5. You must be the originator of the code. Be data and consistent lack of use of uniform data stan-
ethical! dards as two major obstacles for health professional’s
utilization of big health data. These are both issues in
The event timeline was as follows: which librarian expertise and instruction can be
August–September 2016: Data Hacking Workshop instrumental. Though Schneeweiss focused specifi-
Series cally on data from electronic health records, these
September 14, 2016: IronHack Training Session issues are also reflected in open datasets from govern-
September 21–23, 2016: Self-Directed Hacking ment agencies and private organizations, propriertary
Phase data in subscription databases, and NGO datasets.
September 24–26, 2016: Expert Evaluation Phase 1 Though the study findings for this service represent
September 27–29, 2016: Self-Directed Hacking a single site case study and are limited as such, both
Phase 2 issues identified by Schneeweiss emerged as the
September 30–October 2, 2016: Evaluation most popularly reported challenge for hackathon
Phase 2 participants. Qualitative feedback from post-event
October 3–5, 2016: Self-Directed Hacking Phase 3 interviews with participants who submitted final
October 6–8, 2016: Evaluation Phase 3 projects revealed the desire for further training in
October 9–11, 2016: Self-Directed Hacking data management, particularly data organization,
Phase 4 data collection, and data analysis. For example, one
October 12–18, 2016: Evaluation Phase 4 winning participant expressed such concerns during
October 19, 2016: Awards Ceremony and his post-event interview, recorded at the Black
Announcement of Final Winners. IronHack awards ceremony: https://vimeo.com/
There were 89 participants registered for the hacka- 188211462. Participants also reported wanting to
thon and 43 participants attended at least one of the improve the efficiency of their data collection and
pre-hackathon training sessions. In the first phase 36 data analysis practices with either pre-hackathon
projects were submitted, 13 submitted in the second training or a guided data management component
phases, nine submitted in the third phase, and six during the first day of hacking.
projects were submitted for final review at the end Participants noted a few logistical concerns. Down-
of the 21-day hackathon. Expert evaluations and feed- loading multiple large datasets caused the wi-fi to be
back were based on usability, technological perfor- slower than expected during the first day of the
mance, and user experience. Libraries Health Data Hackathon, which was espe-
The Purdue University Libraries Health Data cially problematic and frustrating for hackers working
Hackathon was a 2-day hackathon hosted by Purdue with a short timeline. And, a sponsor provided snacks
University Libraries in Spring 2017. Participants were for the Libraries Health Data Hackathon, but partici-
asked to use open-source data to identify health trends pants reported that lunch, even just pizza or sand-
and opportunities, then work as a team to design an wiches, would have been more useful as they would
app to track behavior and enhance communication not have needed to break for lunch.
252 IFLA Journal 45(3)