Novel Endpoints for Osteoarthritis (OA) by applying Big Data Analytics

Leave a Comment / Uncategorized / By

HORIZON-JU-IHI-2024-08-02-two-stage

General information

Programme

Horizon Europe Framework Programme (HORIZON)Budget overview

Call

Innovative Health Initiative JU Call 8 (HORIZON-JU-IHI-2024-08-two-stage)

Type of action

HORIZON-JU-RIA HORIZON JU Research and Innovation Actions

Type of MGA

HORIZON Action Grant Budget-Based [HORIZON-AG]

Open For Submission

Deadline model

two-stage

Opening date

25 June 2024

Deadline dates

10 October 2024 17:00:00 Brussels time

23 April 2025 17:00:00 Brussels timeTopic description

ExpectedOutcome:

The action under this topic must contribute to all the outcomes listed below, by integrating existing data sets (clinical registries, prospective observational trials and real-world evidence data, for example from medical claims and biobanks as well as genotypic and epigenetic information), and data collections from historical and ongoing clinical trials (provided by industry partners).

  • Algorithms and models, including Artificial Intelligence (AI)-based models, that are adaptable to differences in data availability have been developed and validated in different datasets to allow for the identification of osteoarthritis (OA) patient subpopulations (phenotypes/endotypes) that will benefit from specific, targeted treatment approaches. The identification of subpopulations will be based on:
  1. the patient-specific burden of osteoarthritis with focus on underlying drivers (e.g. metabolic disease) and multi-morbidity/holistic patient profiles;
  2. the evaluation of underlying pathways driving local vs. centralised pain in joint disease and the correlation of symptoms to joint tissue pathology;
  3. the identification of key risk factors for pain in joint disease that can be linked to structural disease progression providing insights into the symptom–structure discordance in OA;
  4. the detection of joint areas at risk of progression and quantification of structural progression to a more advanced stage;
  5. the measures from existing innovative tools such as functional assessments with mobility and activity assessing devices (including algorithms) to reflect independence, gait measures, and assessments of muscular strength and function, as well as balance and coordination to subtly measure functional changes;
  6. evaluating the differences and commonalities of osteoarthritis (OA) and inflammation-driven joint diseases such as psoriatic arthritis (PsA), rheumatoid arthritis (RA), erosive hand osteoarthritis (eHOA).
  • A validation strategy is provided for a selected set of novel endpoints to measure and predict OA disease progression that enables planning of regulatory implementation pathways. This validation strategy supports innovative outcome-based and patient-centred development approaches for medicines and other therapeutic options to be discussed by regulatory authorities, health technology assessment (HTA) bodies, healthcare providers, patients, scientists and industry, shaping new approaches to the development of efficient treatments in OA and respective regulatory frameworks;
  • A decision tool is developed – based on the predictive models – that supports shared decision-making for patients, their caregivers and healthcare providers according to the predicted disease progression, the most likely associated OA disease drivers and the current disease burden;
  • A robust, trustworthy, and interpretable AI framework is established, that enables the development of guidelines or determines any boundaries for predictive modelling at various stages of value generation e.g. biological discovery, patient subgrouping, and clinical trials enrichment. Measures to mitigate the risk of bias and discrimination are implemented including, but not limited, to:
  1. careful consideration of data sets to ensure diversity and inclusion (or account for the lack thereof);
  2. the running of bias-unaware AI models and provision of fairness metrics;
  3. applying AI models within frameworks mitigating bias and promoting fairness during the pre-processing, in-processing and post-processing phases.
  • Data platform(s) are designed and implemented to allow a workable and efficient collaboration across the participating organisations in their respective geographies, respecting each data contributor’s access, privacy and consent approaches, which can be facilitated by federated data sharing. This outcome may serve as a blueprint for other data collaborations under the umbrella of the EU’s newly implemented AI act and data policies 1, 2.

It is expected that certain existing assets like clinical data, algorithms, and data storage infrastructure will be used as background in this action. Therefore, beneficiaries intending to participate in this data-driven action need to be comfortable with the principle that ownership of specific deliverables / project results which would be considered direct improvements to a beneficiary’s background asset, will need to be transferred back to the beneficiary who contributed the background asset to the project. Provision for, and conditions relating to such transfers should be specified in the project’s consortium agreement.

1 Proposal for a Regulation of the European Parliament and of the Council laying down harmonised rules on artificial intelligence (Artificial Intelligence Act) and amending certain Union legislative acts (2021/0106(COD), 26 Jan. 2024, pdf (europa.eu), last accessed 04.04.2024

2 Proposal for a regulation – The European Health Data Space Proposal for a regulation – The European Health Data Space – European Commission (europa.eu), last accessed 04.04.2024Scope:

Osteoarthritis (OA) has no cure and affects the lives of more than 500 million people worldwide with widespread individual, societal and economic consequences. Economic consequences pertain on one hand to health care utilisation and health care spending, OA is however also associated with relevant economic impact on the individual due to missed days at work, early retirement, and substantial out-of-pocket expenditures. Since OA primarily affects the elderly, females, patients with lower levels of education and socio-economic status and certain ethnicities, the associated economic risk hits already vulnerable populations. OA has long been underestimated in its impact; the disease negatively affects social functioning and ranks 7th for years lived with disability in people over 70 years. With its impact on activities of daily living, OA is a major risk factor for loss of independence. Additionally, OA is associated with increased mortality.

Despite major research efforts and increasing insights into the mechanism, epidemiology, risk factors and natural history of OA, various development efforts over the years have failed to provide a disease-modifying treatment. The epidemiology as well as clinical and biological insights strongly suggest the existence of several pheno- and endotypes of osteoarthritis; failure to account for those differences critically hampers progress in the field. The implementation of innovative approaches to stratify the patient population, predict the course of disease and define patient-relevant endpoints is specifically relevant in an ageing society with a high prevalence of obesity, metabolic syndrome, and multi-morbidity. Furthermore, there is an increasing prevalence of post-traumatic secondary OA in relatively young individuals affected at the prime of their lives. First studies towards the clustering of patient groups and development of predictive models have been published suggesting the feasibility of these approaches. Bringing all those insights together requires the collaboration of experts from various fields and can only be achieved in the concerted action of a public-private partnership, including existing initiatives.

The overall aim of this topic is to build a public-private partnership that is able to integrate and leverage the plethora of existing and currently collected data on OA, as well as the increasing insights and expertise gathered over decades of research. Further, the goal is to use a data driven approach to significantly progress the field by leveraging the novel opportunities that have emerged thanks to increased computing power and innovative methodologies in big data analysis, in order to:

  1. integrate different perspectives to improve the understanding of osteoarthritis as a complex disease;
  2. foster progress towards regulatory validation of patient-relevant endpoints to measure and predict OA disease progression as well as alternative endpoints to measure response to treatment;
  3. allow predictive modelling while actively seeking feedback to incorporate the perception of patients, care givers, primary care physicians and regulators.

The action generated by this topic should pave the way towards transforming the current isolated research efforts and static late-stage development approaches into a more patient-centred and simplified (more inclusive/enriched patient population, shorter study duration, potential enablement of the evaluation of preventive or early therapeutic strategies based on predicted outcomes, cost-effectiveness etc.) as well as sustainable part of clinical research and development. This aim is supported by increasing the insights into OA as an heterogenous disease with various underlying patient risk profiles, patho-mechanistic pathways and underlying genotypic/epigenetic/ metabolomic/transcriptomic phenomena based on big data. Such insights will allow for the creation of integrated risk profiles combining clinical and multi-omic approaches (e.g. clinical characteristics, transcriptomics, proteomics, genetic markers, and in-depth multimodal imaging data).

These advances are needed to support the development of patient-relevant and cost-efficient integrated health care solutions including focused, individualised treatments for specific patient segments. The use of AI-based approaches is crucial for the integration of the totality of existing patient datasets and mechanistic disease insights to better understand disease drivers in various tissues of joints thereby upscaling, broadening and/or sharpening current methodology.

The proposed action must:

  • gather and provide access to high quality data – including clinical data from trials (mainly data from placebo arms from studies run outside the project) provided by the pre-identified industry consortium and by applicants as well as prospective observational data, registry data and cohort data including genetic, imaging, soluble biomarker, and data from wearables among others;
  • provide a flexible federated data lake house with appropriate tools for access, management and governance, data curation, integration, and augmentation for consequent high-performance analytics using for example new or contributed AI (foundation) models and modelling workflows. This infrastructure will deploy existing or newly developed approaches or implementations to host and analyse disparate data assets ranging from public, commercial, and not-for-profit observational and trial clinical data to -omics, images, or data from wearables. In their proposal applicants should address key challenges around federated data collection, data privacy, data transfer, data storage, data processing, curation, and harmonisation of data, etc. to achieve a comprehensive understanding of OA by upscaled, big data analytics from:
    1. genetic analyses (GWAS);
    2. AI-driven big data analyses for identification of clinical patterns in phenotypes and endotypes;
    3. algorithm-based imaging analyses of whole joints and peri-articular tissues;
    4. the evaluation of performance assessments using novel technologies and devices.
  • generate and provide a validation strategy for a risk model of disease progression by evaluating whether and to which extent risk factors and predictive models identified in the literature and the above-mentioned data sets are reliably predictive for the progression of structural joint changes as evidenced by imaging, pain and functional decline documented by patients and ultimately leading to joint replacement surgery. The combination of surrogate markers such as imaging [1] with medical history and medication, as well as with predictive markers (plasma-based multi-omics, polygenic risk scores) [2][3], patient reported outcome data and data from wearables or performance tests [4], will generate a more refined predictive engine in analogy to, for example, established fracture risk prediction algorithms in osteoporosis;
  • work towards a broad consensus between all stakeholders especially linking patients, caregivers and healthcare providers’ perspectives to regulatory and health technology assessment (HTA) bodies. This will enable the elaboration of a set of endpoints relevant to these groups depending on the phase of development of treatments (i.e. early phase trials for medication or device efficacy, while late-stage development needs to prove effectiveness, which may necessitate different sets of outcomes), incorporating the various domains of assessments, and taking into account the predominant effect (structural or symptomatic) of the evaluated treatment. This will help to shape new regulatory frameworks for accelerated targeted OA treatment development based on big data analyses, in-silico trials, digital twin approaches and similar innovative trial designs;
  • use data analysis and modelling to provide evidence and knowledge that could enable the evaluation of existing innovative tools (such as functional assessments, imaging approaches etc.) and innovative treatment solutions for OA, based on their scientific validity and feasibility as a prerequisite. Design a strategy to progress them towards regulatory validation and implementation. The action should provide an exploratory and interactive platform to evaluate the validity and user-preference of novel methods of evidence generation, such as the use of data from wearable devices, innovative imaging, and surrogate markers for joint replacement surgery;
  • model short- and long-term economic and public health impact from OA including morbidity and mortality. These new risk models should support benefit/risk assessment as well as quality and efficacy assessments of therapeutic interventions in patients diagnosed with OA to prevent or delay the onset of disease progression, but also avoid overtreatment and thereby optimise the use of health care resources;
  • develop a decision tool based on predictive models that can support shared decision-making between physicians, patients and their caregivers to select the intervention best suited to address the various stages and symptoms of OA in an individual patient, integrating also patient reported outcome and experience measure (PROMs and PREMs) data as well as patient preferences. The diversity of patients at risk or affected by the disease must be considered when discussing patient-relevant outcomes to enable the focused development of treatments and healthcare solutions specific to the needs of individual patients;
  • leverage real-world evidence (RWE) data to address the diversity of patients including sex and gender, ethnicity, and race disparities to develop patient engagement strategies. This should enable engagement with specific groups for the design of OA outcome trials and better promotion of OA management.

The action should contribute to addressing the research needs outlined in the Regulatory Science Research Needs initiative1, launched by the European Medicines Agency (EMA), assessing the utility of real-world healthcare data to improve the quality of randomised controlled trial simulations and patient and public involvement and engagement.

Therefore, applicants are expected to consider the potential regulatory impact of the results and – as relevant – develop a regulatory strategy and interaction plan for generating appropriate evidence as well as engaging with regulators in a timely manner (e.g. national competent authorities, EMA Innovation Task Force, qualification advice).

Consideration should be specifically given to patient and public involvement and engagement in the implementation of all of the above activities. The applicants are expected to leverage prior learnings, for example, previous experiences that have demonstrated the importance of transparent and accessible structures to receive input from patients, caregivers and health care providers as key stakeholders and integrate expertise from various fields relevant in this context [5]. The continuous and active engagement of all groups is indispensable to meet patients’ and providers’ needs and leverage synergies between practitioners and scientists, especially to ensure the sustainability of potential outputs.

Applicants should provide in their proposal evidence that they have in place all permissions (legal, ethical) needed for accessing the data necessary to implement the action.

Note that the implementation of prospective clinical studies is not supported by this topic.

1 https://www.ema.europa.eu/en/documents/other/regulatory-science-research-needs_en.pdf, last accessed March 19th 2024Expected Impact:

The project should contribute to all of the following impacts:

  • the federated integration of big data from disparate data sources including the use of digital twin and similar methodological approaches will lay the foundation for advanced clinical trial designs that allow for more efficient and smaller trials, as well as the reduction of patients’ burden and exposure to placebo;
  • the development of predictive models for disease progression and joint replacement, which are crucial to efficiently discuss treatment strategies, support assessments of quality in health care and equitably plan and allocate health care resources. In addition, such predictive models can revolutionise outcome trial designs, shortening the trial duration and patient burden as well as reducing development costs. The aspired modular flexibility to data availability allows for their sustained use in various settings and economic circumstances;
  • the stratification of different patient groups and targeting of treatments to patients’ needs and preferences, which enables the development of successful therapies, informs development strategies, improves patient and caregiver engagement and optimises trial designs. This stratification also supports data-based shared decision making for health care solutions in clinical practice;
  • availability of tools that enable specific functional measurements and reflect the real-life treatment benefit for patients. These tools have been positively evaluated for practicality and scientific validity and could be used for systematic assessments complementing clinical and patient reported information. All of the above will allow for better trial designs that can demonstrate the treatment benefits of medicines and health care solutions in early development programmes with limited numbers of patients.

for more information:

Leave a Comment

Your email address will not be published. Required fields are marked *