The Importance of Real-World Data to Precision Medicine

Abstract

data analysis • real world data Personalizing medical treatment and healthcare delivery on the basis of a person's biology is not new.Administering blood transfusions and differentiating the treatment of sickle-cell disease from carrier status are very old examples.However, what is accelerating the momentum toward more personalized medicine is the deep characterization of individuals by genome, proteome, metabolome or microbiome allowing very precise specificity, alongside the development of (usually) genetically targeted therapies that allows this precise specification to deliver more effective and safer treatments [1].This widely heralded opportunity has triggered substantial governmental and industry investments, for example in the USA [2,3] and a new European Union Declaration of Co-operation toward access to at least 1 million sequenced genomes in the EU by 2022 [4].These commitments and investments will accelerate the research that helps to identify how new generation therapies can be more rapidly identified for relevant molecular targets.Cancers [5] and rare diseases [6] are widely perceived to be the early wins, primarily because therapies can be easily directed toward manifestations of a single gene or sequence with limited requirement for collateral clinical data and minimal dependence on lifestyle or wellness data.These first-line conditions are important areas for societal benefit and are to be welcomed.However, the promise of personalized medicine is greater than this, and making inroads into some of society's greatest healthcare burdens, such as long-term conditions and their primary/secondary prevention, requires much richer clinical and patient-generated data alongside the molecular data, in order to meaningfully characterize patients into subgroups for whom treatments, care pathways and prevention strategies can be better tailored.Artificial intelligence informed care decisions are another form of personalization that is receiving attention and investment, but is not often referred to as a part of personalized medicine -which it is.A wider range of health and care information is needed for machine learning, probably requiring the complete electronic health record (EHR) and increasingly requiring patient-generated monitoring and lifestyle data.Access to fine-grained health data is the next scalability challenge for personalized medicine.New evidence derived from large populations of data is needed to direct the development of new targeted drugs, to discover unmet needs and the value of novel therapies.An evidence-based reorganization of healthcare is also needed to deliver more personalized packages of care.Infrastructures providing large populations of detailed health records, alongside genotypic information and/or bio-samples, are a critical success factor to scaling up personalized medicine [7].Biobanks, as a high-fidelity resource of health data and samples, are not new (e.g., the Estonian BioBank was started in 1999 [8]).What is new and challenging is how to conduct research on large networks of routinely collected health data (real-world data) on the broad range of patients who reflect the true diversity of persons with a condition and not only those who volunteer for clinical trials or biobanks.Some countries have established multiple patient registries that collect and clean routine EHR data for use in healthcare quality improvement and in research, perhaps in Europe the most well-known being Sweden.However, single-disease registries, despite their value, are also no longer sufficient if we are to address the next European and global challenge of multi-morbidity.In recent years we have seen national

Ähnliche Arbeiten