REaltime DAta Synthesis and Analysis (REDASA) INFODEMIC

Intelligent automation of data search, curation, interpretation and communication for clinical decision making during infodemics.

Dr James Kinross - Primary Investigator

In March 2020, Imperial College London and The Institute for Health Metrics and Evaluation (IHME) each produced COVID-19 epidemiological models that altered global public health policy. However, policymakers and clinicians were unable to quickly respond to these predictions. The slow response was due to the absence of accurately synthesised data guiding evidence-based clinical prevention and therapeutic strategies designed to protect staff and patients. A major reason for this has been the overwhelming pace and scale of data production; by August 2020, 37,362 peer-reviewed scientific papers, 1503 policy documents and 1097 data sets had been published. By March 2021, 14,576 pre-print articles were published without peer review. At this time, Clinicaltrials.gov reported over 5,185 trials relating to COVID-19 and these numbers are growing. The volume of data has come at the expense of its reliability; Over 86 COVID-19 papers have been retracted since March 2020, with high profile failures causing lasting patient harm. The complexity and ambiguity inherent in the distributed, COVID-19 knowledge corpus has hindered scientists searching for valuable therapeutic targets. Opaque data reporting has eroded public trust, exacerbated by misinformation spread through ongoing dialogue carried out by state actors, politicians, cybercriminals, and conspiracy theorists across media platforms(2, 3).

Our target audience is clinicians, policymakers, and the biomedical community. This platform is built by clinicians and our solutions provide field experts with relevant, accurate, and timely data on therapeutic interventions, preventions, and safety strategies for both patients and staff. Examples from the COVID-19 response illustrating the potential impact of REDASA include the failure to protect frontline healthcare workers due to inadequate evidence for effective PPE use, and the need to provide living decision support advice leveraging the latest evidence to ensure optimal patient outcomes. We have engaged with front line clinicians and with our current global user base of over 450 registered data curators from 9 different countries.  Our secondary audience is the biomedical community. Significant uncertainty persists about the fundamental biology of SARS-Cov-2, which is significantly impeding safe vaccine development and therapies. Biomedical scientists deserve a 21st-century peer-review methodology that’s fit for purpose, free from bias and less vulnerable to misuse. Our objective is to fundamentally transform how scientist’s contributions to the interpretation and quality assurance of evidence and data is acknowledged, incentivised and rewarded.

The core mission of REDASA is to improve knowledge and outcomes through the improved acquisition, analysis and communication of infodemic data. We aim to directly demonstrate this benefit through the prospective trialling of this technology with clinicians through our partners at NICE within the timeline of this award. This means we are aiming to create meaningful, clinical knowledge that can be shared with front line clinicians and policymakers. We will also work with the IHME to work specifically on the distribution of transparent and validated data to the public on future pandemic use cases. REDASA is committed to sharing our data through open data frameworks and to creating a community of ML developers who wish to use our ground truth data sets for social good. Our data will be publicly available through the AWS open data scheme, the CURADR platform, and through direct initiatives with our academic partners. CURADR is at its core a platform designed for positive social change and public health. It is open and free to access, it is designed to acknowledge and reward clinicians, biomedical experts and the public that give back to their community. Our ultimate objective is to create value through collaboration and partnership.

1. Healthcare workers and policymakers: NICE will assess the robustness and appropriateness of REDASA evidence and will develop new mechanisms for incorporating the evidence into guidance that will be developed and tested based on COVID-19 and secondary infodemic use cases. NICE will also assess if and how novel sources of data, such as social media, sourced through the projects data pipeline can contribute to signal detection and for informing future areas where specific updates to guidance may be needed across its entire portfolio. 

2. Biomedical experts: We will measure the impact of this technology through the growth of the community of CURADR researchers. Our goal is to have 10,000 active users by the end of this award. This target is based on our ability to recruit 450 curators in 6 months during our pilot phase, simply through social engagement. 

3. Public: We will work with IHME, our patient advisory panel, and the Imperial College Patient Experience Research Centre to create engaging and visual reports across multiple mediums and languages. Specifically, we will aim to reach communities that are more likely to be adversely affected by infodemics.

The primary goal is to have global impact by scaling through three domains:

Data and analysis
Year 1: We will rapidly increase the number and type of data sources that are ingested by building strategic partnerships with academic organisations, publishing houses and healthcare institutions. 

Year 3: Our open data sets will be used by a global population of data scientists as the standards for ML analysis of health data.

Curation and peer review
Year 1: The objective is to ensure the stability of the platform before increasing the number of curators above 10,000 people.

Year 3: Established social impact metrics, incentivisation methodologies and information veracity measures that can be globally adopted and which radically transform how health data is communicated and interpreted.  Our objective is to be the new standard in peer review.

Evidence based medicine:
Year 1: Our objective is to have REDASA living systematic reviews methodologies validated by NICE and IHME.

Year 3: We will demonstrate improved patient outcomes in infodemic use cases that extend beyond COVID-19 by translating our infodemic methodologies into evidence that has clinical utility.

Data repository: Our objective is to host the largest publicly available collection of clinical and biomedical data relevant to COVID-19 globally. This archive will be contemporaneous and entirely unique.

CURADR network: Our conservative aim is to reach 10,000 data curators. We have reached 450 within 6 months and we believe this to be achievable. Our platform will be scalable and our process will be able to host a much larger number if needed.

Open data: >100,000 annotated papers / data sets. We have curated 7,000 papers to date, and we believe this to be achievable. The ML-based decision support will improve the speed and accuracy of the process, rapidly surpassing this figure. Our objective is to scale beyond COVID-19 into other acute and chronic diseases, and emerging global health challenges.

Living evidence: Our objective is to transform lives through the improved access and interpretation of data. Through our partners at NICE, we will measure the impact of clinical evidence through the assessment of national audits and change of clinical practice.  At the end of year one, we expect to see a 25% decrease in turn-around-time for implementing new therapies into clinical practice compared to traditionally published journal literature.

Our strategy is to build the REDASA engine iteratively into a global platform, demonstrating both economic and social value as we scale. Initially, the UK and the USA, where our lead partners are located. We will leverage the global reach of IHME to access low- and middle-income countries where pandemics, misinformation campaigns have had a deleterious impact on healthcare provision. 

Barriers and solutions: 

1) Access to local data sources. 

We will work with our academic collaborators and AWS to access local data sources in EMEA regions, Asia, South America and Australasia. Governmental regulation: Specific challenges in accessing the Chinese market where the government may prevent access to data or our products reach. We will engage with Chinese hospitals (via existing links to Wuhan), WHO (via ICL and AWS) and Chinese academic institutions. 

2) Data security and public trust. 

We will adopt a transparent data governance strategy, open data sharing policy and stringent data protection protocols. Legal: Our technology is based on web archiving, therefore it is dependent on our data partners providing open access. 

Economic: 

Our objective is to identify a scalable business model through which we can provide fee for service for commercial partners while providing free services for public bodies.

ICL, NICE, AWS, IHME, MirrorWeb, Cloudwick Technologies

A major driver behind our application has been the creation of a multi-disciplinary collaboration between academia, global cloud service provider (AWS) and computing SMEs. This is critical for supporting the REDASA innovation and the wider disruption to the living systematic review necessary for delivering a high impact infodemic engine. We will leverage this award to create sustainable partnerships with the Institute for Health Metrics and Evaluation and NICE allowing us to exploit unique COVID-19 data sets and to access world-leading policy and data science expertise. More critically, the challenge is closely aligned with values of social good, knowledge sharing, and scientific excellence. It offers a unique opportunity for working across academic, commercial and technical boundaries and for performing high risk, high return translational work that will change our ability to respond to any future global healthcare crisis. Trinity funding will support researchers at ICL, IHME and NICE and it will mean the industry can commit to providing full-time involvement of talented computer scientists to solving the infodemic use case challenge. It will also allow us to build a sustainable business model and a global network of clinicians and academics, that will support the natural growth of REDASA beyond this award.

We have assembled a global consortium of leading academic, institutions, industry and SMEs with a credible track record in infodemic use case management. The Institute for Health Metrics and Evaluation (IHME), University of Washington: IHME provides a rigorous and comparable measurement of the world's most important health data sets. This highly influential organisation has produced evidence that directly impacts healthcare policy. They are compelling partners that will ensure this work has a global impact. National Institute for Clinical Excellence: The NICE 5-year strategic plan has two key strategy pillars: Development of living guidelines and leadership in research and data science(7). As a partner NICE will ensure that REDASA produces guidelines that have a clinical impact. Amazon Web Services (AWS): AWS have expertise in large-scale distributed computing and operations that have a global impact. Most critically AWS is already working as the lead digital transformation partner with the NHS and other global healthcare providers. MirrorWeb provides the digital history of the UK Government by archiving all government websites and social media feeds and it will serve as the engine of the REDASA project. Cloudwick will be responsible for our native data lake that powers our agile cloud analytics strategy.