Secure data exchange within complex privacy legislation.
To complement the minimal registry dataset using machine learning/artificial intelligence.
While patient registries contain carefully curated and collected datasets, these datasets are necessarily limited in scope. These registries focus on key disease outcomes that are considered relevant to clinical investigators and that are feasible to collect long-term with busy healthcare professionals in mind. Methodology will be developed and tested to complement routinely collected registry data with additional data sources. First, a minimal data set will be determined that is needed to answer specific regulatory and/or HTA questions, e.g., long-term efficacy, safety adherence or appropriate use. This minimal dataset contains valid data which is collected without additional registration burden for healthcare providers. Data mining / language processing techniques will be probed to unlock unstructured data elements. Second, standards for safely exchanging data between data sources/centres will be tested and implement ability of such standard procedure evaluated. Studies will be performed where the basic DICA data and the total data set including the data obtained through data mining techniques will be leveraged for their usefulness as external control data. These external control data can serve regulators and HTA bodies in their reassessment procedures of drugs conditionally approved or reimbursed based on uncontrolled single arm trials. Additionally, these data will be leveraged to assess these drugs’ effects in real life versus those observed in the clinical trial. We will evaluate the impact the impact of the fast-changing treatment paradigm, due to availability of new therapies. These data impact the interpretation of clinical trials, as the standard of care given in the trial may no longer be relevant for current day clinical practice. The data mining approach will add to the methodological framework on how registry core datasets can be expanded.
Dutch Institute for Clinical Auditing (DICA) (https://dica.nl/)
