Data Flow Analysis
Transforming complex clinical trial data systems into efficient, integrated workflows
Backstory
In early 2020 our team was engaged by a very large pharma company to conduct an in-depth analysis of how their clinical trial data flows internally.
As a small company, we were surprised they'd choose us instead of the traditional large consulting companies.
Our champion said he wanted to have "a fresh set of eyes" looking at their current challenges. In his words: "a point of view free from pre-existing vendor allegiances."
Our Process
We conducted over a dozen personal interviews (via Zoom) with individuals responsible for every facet of the data flow.
One of our strengths as a consulting company is that we're not pharma experts.
We are database and software experts. This allows us to bring to the table both process as well as implementation suggestions that our clients may not even be aware of.
What this means is that during our interviews:
- first, we questioned the answers ("why do you do things this way?")
- then we questioned the questions ("why do you even do this?")
- and finally we asked new questions ("if you could eliminate this step in the process, what would be the business value?")
Results
Our study identified many chokepoints, areas of potential data inconsistencies, as well as huge labor costs caused by many manual processes.
The current process, described in detail below, includes many manual processes and multiple data transformations that add significant delays to the submission process.
Areas of Challenge Identified
Standards Management
Some of the challenges we identified include:
- Standards stored in Excel, Word
- Difficult to track changes, versions
- Long, slow trial design process
- Metadata changes are not fully reflected in subsequent processes
Study Build
Some of the challenges we identified include:
- Takes up to 60 days to build study
- Expensive to make form changes
- Weak Protocol – build connection
- Process tied to specific EDC
- Manual integration of external data
Data Ingestion
Some of the challenges we identified include:
- Multiple transformations:
- extract data from EDC
- integrate non-CRF data (labs)
- Transformations use old SAS code
- Scalability, robustness of storage
Internal Data Consumers
Some of the challenges we identified include:
- More transformations:
- for each Data Consumer
- in multiple formats
- Transformations use old SAS code
- No feedback loop to metadata
Customized Dashboards
Some of the challenges we identified include:
- Manual creation of study-specific dashboards (Spotfire, Power BI)
- No Protocol – dashboard link
Data Conversion for Submission
Some of the challenges we identified include:
- Many transformation SAS macros
- Data issues identified very late
- Significant labor costs
- Potential for human errors
TrialTwin
It's based on this work that we started building our TrialTwin platform.
We do not want to provide just another "patch" on top of obsolete, paper-centric processes. Our vision is a metadata-driven, integrated system that replaces most current spreadsheet-based processes.
TrialTwin combines a metadata-driven repository of both content and data plus automated processing and generation of Synthetic Health Data.
The TrialTwin platform is:
- Metadata-centric –> single source of truth (see our Regulatory Repository)
- Automated –> reduce labor cost, increase quality
- Processing –> accelerate analysis speed
- Synthetic Data –> test and validate early and often (see Synthetic Health Data)