Standards and Ontology Driven Workflow Automation of Clinical Studies
Clinical trial costs have been rising to keep up with the complexities of drug development and the business need for better benefits to risk ratios. This 2014 paper, “Examination of Clinical Trial Cost and Barriers for Drug Development” written by the Assistant Secretary of Planning and Evaluation of U.S. Department of Health and Human Services (HHS) points out many of the cost drivers and barriers for drug development. At the time of writing this paper took into account various aspects including electronic health records, simplifying protocols and controlling amendments, wider use of EDC and other forms of testing including at-home or remote. A number of developments within the industry and regulatory environment now offer new opportunities to both reduce and streamline the clinical trial data and workflow process while improving the richness and value of the data being collected.
In December 2014, the FDA mandated that all submissions of studies, both nonclinical and clinical must be in machine consumable digital format. This was not just about requiring submissions to be in electronic format instead of paper, it was mandating that the data be published in a manner that is machine readable and usable so that the data can be instantly viewed, analyzed by programs and scripts, and otherwise re-purposed for review without the need for manual transcription or handling. The de jure standard of choice has been CDISC and its tabulation model SDTM for raw data and ADaM standards for analysis ready data extracted from the collected data. These were data “Exchange” standards meant for two systems to communicate data in a very coherent, clear and consistent manner.
The paper by HHS considers the possibilities of streamlining and reducing costs of clinical trials within the data and workflows through automation exploiting various standards and published controlled terminology lists to codify specifications for quantitative and qualitative data. Separately, the paper also covers the processes that touch site investigators, scientists, technicians, and regulatory affairs very well.
The lifecycle of a specific clinical study with regards to data begins with a protocol and ends up with a completed study report that is ultimately used to determine objectively by the drug developers and regulatory agencies if the drug is effective and safe. This lifecycle is punctuated by a series of stages of workflows to specify, collect, organize, analyze or package data. Most of these workflow stages require considerable manual effort involving data management, programmers, bio-statisticians and Study Data Tabulation Model (SDTM) savvy data wranglers. These stages are designed to address questions about how the study should be conducted such as: what kind of data will be collected and when it will be collected from subjects recruited into the trial; how the data will be analyzed to generate an objective unbiased view of specific cohorts in the trial to the drug candidate; and how the data shall be modeled and reported for submission to the regulatory agencies.