INNOVATION IN HEALTHCARE
AND HEALTH EDUCATION

ABSTRACTS

The PERMIT Project – introduction and rationale behind the project (10 min)
As personalised medicine (PM) continues to rapidly evolve, integrating new and innovative research methods, methodological research is required to ensure the robustness of evidence generated, to support funding decisions by research councils or peer-review by scientific journals, and to ensure the most effective solutions are reaching patients and citizens. The PERMIT project, funded by the European Union’s Research and Innovation Horizon 2020 program (Grant Agreement n° 874825) was designed to develop a set of methodological recommendations to support investigators, evaluators, peer-reviewers, regulators and other stakeholders involved in designing, funding, conducting, publishing and validating PM research. With a multinational consortium, composed of all key stakeholders of the PM research ecosystem, it strove to include all of the relevant perspectives in the joint development of the recommendations.
Methodology – general approach for developing the PERMIT recommendations (10 min)
A common methodological framework was followed for the development of the recommendations. It was designed to be thorough, but flexible, to accommodate the specificities of each of the four identified stages of the PM research pipeline. The approach consisted of four phases, beginning with the collection of background information through scoping reviews and a gap analysis. Findings of the gap analysis were then discussed in the development phase, where working sessions and workshops with field experts took place. The draft recommendations from this phase moved on to the finalization phase where consensus on the final recommendations was reached. The final phase focused on defining the most appropriate dissemination and implementation strategies and mobilizing efforts to achieve them.
Collection of patients and omics data – methods for data collection and quality, and design of stratification and validation cohorts, gaps and challenges, recommendations and their impact (15 min)
The first stage of the PM research pipeline involves the prospective or retrospective collection and consolidation of patient data. The collection of data including –omics, imaging, clinical, lifestyle, exposome data and more, provides the basis for the stratification of patients sharing a similar profiles into homogenous clusters. The recommendations for this stage were developed in collaboration with field experts, following the mapping of currently used methods used in stroke, cancer and Alzheimer’s PM research. They contribute to better defining the best approach to data generation, integration and curation as well as ensuring data quality. They also help to address the adequate planning for validation of clustering and provide guidance on the calculation of the sample size.
Computational analysis and modelling – machine learning methods for patient stratification, gaps and challenges, recommendations and examples (15 min)
Intricately tied to the first stage, the stratification of patients constitutes the second stage of the PM research pipeline. Discrimination and sub-group characterization of complex and heterogeneous disorders often requires advanced machine learning (ML) analyses to identify multi-variable biomarker signatures, as combinatorial signatures have the potential to provide more robust and accurate stratification models. Successful ML-based patient stratification requires both an adequate prior study design and a suitable choice of data pre-processing, model building and evaluation methods during the study implementation. The recommendations for ML stratification methods cover the planning phase, discovery and modelling phase, and the validation phase. They provide guidance to avoid common pitfalls and help to ensure a robust and reproducible approach is taken.
Translational development for personalised medicine – preclinical models for personalised medicine, gaps and challenges, recommendations and their potential impact (15 min)
Translational development is the third stage of the PM research pipeline. Although much progress has been made in the development of innovative and complex preclinical model systems, fundamental deficits in translational methods still exist, and they prevent the further development of PM. The recommendations for the translational step of the PM research pipeline focus on five areas: 1) clinically relevant translational research; 2) robust model development; 3) transparency and education; 4) revised regulation; and 5) interaction with clinical research and patient engagement. They help to address the identified challenges regarding the lack of standardised protocols, validation procedures and quality assessment practices, as well as the need for more accurate reporting, and reporting of negative results.
Clinical trial for personalised medicine – most commonly used designs for personalised medicine, gaps and challenges, recommendations and examples (15 min)
The evaluation of treatments through clinical trials is the final stage of the PM research pipeline. Innovative and complex designs continue to arise and to be applied to PM research. It is therefore key to ensure that the designs are the most adequate to address the research questions and that they have the capacity to produce robust and reproducible results that can be seamlessly used to make decisions for integration of therapies into care. Four types of recommendations for PM clinical trials were developed: 1) general recommendations on the elements that all PM trials should consider; 2)recommendations for the assessment of personalised vs non-personalised approaches; 3) design-specific recommendations for umbrella, basket and platform designs and 4) topic-specific recommendations to aid in selecting the most adequate trial design depending on the PM research typology.