Standardization of Data for Pathology Research
Click here to view the slide deck.
TITLE: MEDICAL PATHOLOGY TERMINOLOGY DICTIONARY (MEDIPATED)
Author(s): Kingsley Ebare & Esther Abels
Addressed Parties: Industry, Academia, Biopharma, Regulatory Bodies who work with clinical studies and artificial intelligence in digital pathology
Background: Algorithms have shown to support drug development and decisions in diagnostics. Diagnostic Algorithms have been developed and are in clinical use in the clinical field for a few decades. Pathology is the one of the last disciplines in healthcare to become digitized. To develop and use algorithms to its full potential, the metadata is crucial input. One of the essential elements of training algorithms is the input using clinical data, such as clinical features and clinical outcome data. In the drug industry, data is standardized using Clinical Data Interchange Standards Consortium (CDISC) operational data model and MedDRA. However, in the medical device field and especially in the field of pathology, the standardized use of medical terminology is lacking.
Approach & Objectives: The objective is to develop a dictionary with standardized data so coding can be applied to improve data collection, safety information, retrieval, evaluation, statistical analyses, and presentation to improving scientific reporting and to reduce the regulatory burden relevant for submissions to Regulatory Authorities. It will also allow for sharing regulatory information globally for In Vitro Diagnostic devices used in Digital Pathology.
There are three main areas that require coding for algorithm development: 1. Training 2. Verification and 3. Validation.
Deliverables: Create a framework for standardization of data used in the design, execution, analysis, regulatory submission and archival of pathology research studies using CAP cancer protocols as input. Also develop a dictionary to code pathologic features as well as medical pathologic diagnoses.
Value proposition: Standardized pathology data is currently missing despite the availability of some tools like synoptic reporting for cancers. Providing this to the industry and regulatory authorities would be helpful in training, generating and presenting evidence for algorithms used in clinical diagnostic decision making. Also, with increasing demand for interoperability, pathology data standardization will allow for efficient data exchange, easier aggregation and analysis and in the long run decreased cost of conducting research.
Clinical – Standardization and removal of potential error sources will improve clinical interpretation in a harmonized way for pathology diagnosis.
Regulatory – Standardization of data facilitates statistical analyses and allows meta analyses to be used for showing devices are safe and performing to its intended use. In addition, it facilitates the analyses and cross analyses with data collected during drug trials, for example in the field of Companion Diagnostics
R&D – Tissue, imaging and metadata standardization will allow harmonized usage in training and as such allows for a general training approach of algorithms which will have a widespread impact into research and development.
Funding sources: None
Benefit to patients : technical advance, increased quality, outcome, access
References/Resources:
https://www.meddra.org/how-to-use/support-documentation/english (accessed Oct 27, 2019)
https://www.cdisc.org/standards (accessed Oct 28, 2019)