insightScope 2021 – Data Abstraction Redesign

Posted on Friday, November 26th, 2021

Client	Dr. Dayre McNally MD, Katie O’Hearn MSc, Misty Pratt MES
Professor(s)	Kevin Holmes,
Program	Mobile Application Design and Development, Computer Engineering Technology – Computing Science
Students	Dhruv Patel (Research Assistant – Full Stack Developer – Computer Engineering Technology – Computing Science) Manan Patel (Research Assistant – Full Stack Developer – Computer Ravi Chandra Rachamalla (Research Associate – Full Stack Developer – Mobile Application Design and Development)

Project Description:

insightScope web application allows researchers to conduct and manage crowdsourced systematic and scoping reviews. It was started in 2018 by a team of Algonquin students from SLiDE led by Kevin Holmes and Sanket Patel, and researchers from CHEO led by Dr. Dayre McNally. Since its inception, insightScope strived to revolutionize the systematic review process. By using insightScope to build a large vetted team, the systematic review process is significantly accelerated from months and years to days and weeks.

During the last term, the insightScope team developed, refactored, tested, and deployed various features of the application. One major highlight is the significant enhancement of Data Abstraction. Data Abstraction occurs after full-text screening whereby the study team identifies and records relevant data, such as research design, type and number of participants, and patient outcome from the articles in the review. This involves collecting duplicate user responses to questions logically classified into study-type forms. Once reviewers have finished extracting data for an article, results are compared to find conflicts. The study team then resolves the conflicts, finalizing data elements for analysis and presentation of findings.

Early research from the team suggested that Data Abstraction by insightScope reduces the conflict resolution work by 50 percent. It increases the potential of running even more systematic reviews with better efficiency on our platform, increasing our prospects for scholarly output that drives evidence-based patient care.

The previous version of insightScope was limiting the types of data points that researchers could abstract as they relied on preconfigured forms. To overcome this, the team refactored the Data Abstraction to be configured by the principal investigator. Study type forms can now be created with multiple question types that accept answers in single or multiple choices, text, number, or date formats. One unique feature is the assignment of dynamic sub-questions that are toggled based on user selections. Further additions include help texts, comment enforcement, custom ordering of forms and questions, marking questions as not applicable, shared database of forms for reuse across multiple projects, assessments auto-save, etc.

The design and development phases of these features came with their own set of struggles and learning opportunities. An increase in customization surged logic complexity and edge cases which can potentially introduce bugs. One arduous thing was to devise a simple yet efficient conflict detection algorithm supporting all question types. Dramatic database changes could introduce anomalies. Through research, design sessions, demos, and close collaboration with the CHEO team, we delivered a robust solution. The team tackled a significant codebase refactor focusing on business logic abstraction, reusability, testability, maintainability and reducing technical debt. The team was provided with exciting opportunities to work with technologies like Amazon Web Services, Python data structures, HTML 5, CSS3, JavaScript, and open-source libraries like Draggable from Shopify.

We are proud of the accessible, user-friendly, and intuitive user interfaces that we developed. We were able to manage ourselves effectively using an agile development mindset. Throughout this process, we acquired many invaluable skills and experiences that aided us to accomplish our goals.