Impact case study database

The optimised scalable methods for learning probabilistic models are being used as part of the CRAN bnlearn package (S1, S2). The bnlearn package is used worldwide and has had 113,028 downloads since the optimised code was made available (S3). In particular, the optimised algorithms in this package are being used by the MHRA for modelling complex disease comorbidities from huge primary care datasets (Wang et al. 2019) to generate synthetic data (S4, S5).

The collaboration with the MHRA has used the scalable Bayesian network package to develop a synthetic data generation framework that enables the generation of high-fidelity synthetic datasets that can be used for benchmarking machine learning algorithms for regulatory purposes (Tucker 2020). The collaboration was highly successful, far exceeding the promised deliverables in the original project scope. Based on this work, the MHRA put in a business case for scaling up production of synthetic data by the MHRA’s specialist division, the Clinical Practice Research Datalink (CPRD), and awarded additional funds from NHSx to build on the previous work (S6).

** 1) Changing National Policy** – As a result of the success of the scalable synthetic data generator, the CPRD has revised its guidance for reviewing machine learning research applications submitted as part of data access requests based on the empirical evidence from this project (S4, S5). In addition, the CPRD is planning to launch a synthetic data generation service, with 12 staff already devoted to working on the project (S4). This new service will involve government analysts adopting innovative methodological approaches based on the work. The MHRA’s intention is that these research informed decisions, which include further work with Brunel, will ultimately lead to direct benefits for the health and wellbeing of people by bringing medical device products to markets more quickly. In addition, commerce and the economy will benefit through making the UK an attractive environment for supporting the development of software medical devices. This will be supported by better public policy with improved regulatory pathways that support innovation (S5). This is the first time such data has been made available and the project is currently shortlisted for a Civil Service award. The work has led to the MHRA releasing two synthetic datasets for the first time: a proof-of-concept synthetic cardiovascular risk dataset has been made available for access by the wider research community. In addition, a synthetic dataset has been generated to facilitate Covid-19 research, which was separately funded by NHSx and can also be accessed via CPRD (S7). The work has been the subject of a press release issued jointly by the MHRA, the Department for Health and Social Care (DHSC) and the Department for Business, Energy and Industrial Strategy (S8). Already, Sensyne Health ( https://www.sensynehealth.com) are developing an app for modelling diabetes using the service. The app aims to predict the need for intervention based upon blood glucose level monitoring and the development and validation of this app is made possible through using the synthetic data service.

2) Expanding the UK Economy – HealthTech is now the largest employer in the broader Life Sciences sector, employing 131,800 people in 4,060 companies, with a combined turnover of GBP25,600,000,000 ( according to the association of British HealthTech Industries - https://www.abhi.org.uk). A report by the market research firm Mordor Intelligence suggests that the global mobile health market will reach GBP46,370,000,000 in 2021. This market is expected to save the UK NHS GBP2,000,000,000 per year (IQVIA Institute S9). According to Deloitte, the biggest barriers to this sector in the UK are cultural and regulatory (S10). One stumbling block to expansion of this sector is the sharing of primary care data that is limited by data protection laws. The new synthetic datasets enable a rapid expansion of health-based apps that utilise historical primary care data, minimising risk of patient identification. This, in turn, will open up a new sector in machine-learning based health apps, which is currently not available, giving the UK a distinct competitive advantage by generating a new economy with the associated high-skill jobs, and offering considerable savings to the NHS.

3) Improving Public Health – The public will see the benefit of a whole new breed of apps that can learn from synthetic historical patient data, saving the NHS substantial costs through better monitoring and diagnosing. For example, diabetes costs the UK NHS GBP9,800,000,000 (diabetes.org.uk). The Sensyne app enables patients to reduce the need for appointments and to give them greater control over their care. The synthetic data generator has enabled Sensyne Health to generate new insights, develop new algorithms, and enable experiment repeatability whilst abiding by GDPR regulation. In addition, some clinically significant populations are under-represented in real-world data. The synthetic data generation methods enable them to produce arbitrarily large samples of such patients for analysis and algorithm development. (S11). The Covid-19 synthetic data holds extremely valuable information about GP visits and has the potential to unlock some of the early signals of the arrival of Covid-19 into the UK based upon recorded symptoms, geographic location and demographic information. The potential of releasing the synthetic data on cardiovascular disease (CVD) is huge. CVD affects 7,600 000 people in the UK and accounts for 27% of all annual deaths – costing the economy GBP19,000,000,000. The opportunity for the health tech sector to develop and validate new apps that can predict and monitor cardiovascular risk using historical data with state-of-the-art machine learning technology, will benefit patients, clinicians and the economy.

This research has had impact in many areas of manufacturing and logistics in terms of enabling better decisions through the use of faster experimentation ( THEME1) and larger simulations ( THEME2). It has therefore enabled (i) many commercial cloud-based high-performance simulation (HPS) systems, (ii) a major high-performance simulation industrial application of large scale simulation and associated innovations, and (iii) contributions to standardisation. Overall, based on substantial industrial experience, the major impact of this work has been to create a foundation for future industrial high performance simulation systems by combining both themes.

(THEME1) The impact of cloud-based HPS has been to enable vendors to create new products/services with a knock-on impact to their clients through more detailed analyses within project timescales. A diverse range of applications include business process simulation, footwear design, emissions reduction, inventory management and freight transportation. Overall, the impact so far has been evidenced by 30 European SMEs from 12 countries that reported the following estimated economic impact as a result of this work: a cumulative turnover increase of approximately GBP13,000,000, leading to the development of around 150 new products or services in manufacturing & logistics and contributing to the creation of around 100 new jobs. Additionally, these companies reduced the time to market for new products and improved business processes, customer satisfaction and business practices. The above numbers are evidenced in official project reports/deliverables of the CloudSME and COLA projects submitted to the European Commission and provided by executives of the involved companies ( E1, E2). The CSSP is also being used in other large projects (e.g. the H2020 Cloudifacturing project involving around 200 SMEs). These figures can be considered to be a lower bound.

The FP7 CloudSME project also led to the founding of CloudSME UG in 2015, a new start up to continue to develop new cloud-based HPS ( E3). With Taylor as a scientific advisor, the company now has 4 employees and an annual turnover of approximately GBP133,000 since 2017 (primarily through contracts in manufacturing) and made its first profit in 2019. The impact of Brunel’s work is further supported by two videos created for the European Commission that demonstrates the results of their ICT for Manufacturing SMEs (I4MS) programme (CloudSME was funded under this). The video by Hobsons Brewery, a UK-based SME craft-brewer, shows how cloud-based simulation made their business more efficient ( E4). The video by Podoactiva, a Spanish SME manufacturer of tailored foot insoles, shows how the technology enabled them to increase their market by enabling new insole design applications for podiatrists. These projects also enabled Saker Solutions, Hobsons Brewery and Taylor/Anagnostou to successfully bid for an InnovateUK project (CraftBrew) that facilitated the development of a low-cost enterprise management system for small brewers.

( THEME1 and THEME2) Taylor/Anagnostou have worked with Saker Solutions (SME, UK) and Sellafield PLC to develop many discrete-event simulations of nuclear waste reprocessing. The impact of THEME2 (large models) comes from training Saker to create large distributed simulations. Taylor/Anagnostou worked with Saker Solutions and Sellafield PLC to develop a distributed simulation of the Magnox Swarf Storage Silo (MSSS) system. This is the only distributed simulation used in the nuclear industry ( E6). This reuses and links six previously developed simulations to form a large-scale simulation of nuclear waste recycling in this area. SAKERGRID, a high-performance simulation system previously developed with Taylor to manage and deploy simulation experiments on a computing cluster (and now cloud) has been significantly extended through the above research and deployed at both Saker and Sellafield ( THEME1). This has led to an increased turnover at Saker of GBP1,400,000 since 2014 with some GBP2,000,000 – GBP3,000,000 of benefit mainly to the nuclear industry (Sellafield) but also manufacturing, defence and retail sectors. In addition, there are unquantified savings arising from the reduced turnaround time of experimentation enabled by the use of SAKERGRID which both reduces project lead time and increases the number of scenarios which can be examined. The upgrading of SAKERGRID to cloud also enabled Saker to continue working effectively during the COVID pandemic ( E7).

Equitable access to education and research is vital to development in Africa and is a major priority for African Nations. For example, the African Union Agenda 2063’s Call to Action commits to speeding up actions to (c) “Catalyse education and skills revolution and actively promote science, technology, research and innovation…” This aligns with Sustainable Development Goal 4 (Ensure inclusive and equitable quality education and promote lifelong learning opportunities for all). As noted by the World Bank, National Research and Education Networks (NRENs) are a major factor in enabling this – as demonstrated in over 120 countries, NRENs organize centralised public university Internet access/connectivity, offer additional software, systems and services through common standards to enable worldwide scientific collaboration and digital educational resources, and, through economies of scale, drive down prices of these. In Africa emerging NRENs are having a major impact on public Universities (e.g. cutting Internet costs from GBP3,000 to GBP4,400 per megabit per second (Mbps) per month to under GBP73 per Mbps). However, African universities cannot realise the full benefit of these reduced costs without networked services supporting education and research. Taylor and Anagnostou have worked with emerging African NRENs to develop these services and to accelerate significantly their maturity. This has had a major knock-on effect to their user community, the staff and students of African universities. Taylor and Anagnostou have identified two main examples of impact.

(i) African Network Service Development. Network service deployment is extremely complex. For example, services supporting access and collaboration need a portfolio of services Authentication and Authorization Infrastructures (AAIs). These allow NRENs and their universities to establish trusted user identities and to authorise home institution access to internal networks and digital resources. Importantly, it allows home users to be authenticated at other partner NRENs and universities across the world (e.g. via eduroam). There are multiple complex routes to deployment (federated single sign-on, e.g. Shiboleth, eduGAIN, etc.) which required a highly skilled technical team. Taylor led consortia and worked with the RENs Ubuntunet Alliance and WACREN to investigate of the most effective route to service deployment of this and other associated services (e.g. Science Gateways) based on a novel e-Infrastructure Reference Architecture ( E1). From this, Taylor led the development of training materials delivered by the RENs to their NRENs. This work was adopted by both these associations as well as at least 6 NRENs (Kenya, Malawi, Nigeria, South Africa, Tanzania and Zambia) supporting approximately 160 Universities and 1,800,000 students ( E2). Our work was instrumental in initiating several NRENs including the Lagos State-based Eko-Konnect cluster of the Nigerian NREN (NgREN) that serves 176 Universities and around 2,000,000 students. ( E3). Taylor/Anagnostou also worked in the H2020 TANDEM project with WACREN to create a NREN Service Roadmap used to guide the development of NRENs in 11 African Nations in West Africa ( E4). These NRENs form a formal part of education policy in these countries – this work has therefore directly contributed to the educational policies of at least 17 African Nations ( E5).

(ii) The OECD, UNESCO and the World Bank have indicated the value and benefits of Open Science in developed economies (e.g. over a billion USD in 10 years in the USA, wider innovation in SMEs, citizen science, etc.) In developing economies further benefits are possible (e.g. access to African research outputs, access to research by research institutions, significantly reduced costs of data access/use, etc.) ( E6) With the Ubuntunet Alliance and WACREN, Anagnostou/Taylor used the FAIR Open Science Platform (OSP) (co-developed by Brunel and UNICT (Italy)) to demonstrate how Open Science could be adopted by NRENs and used to support their Universities. Again with UNICT, Anagnostou/Taylor ran several hackfests (2016/7) to jump start several African Open Science projects. This resulted in 35 scientists from 6 African countries developing 49 applications servicing 24 scientific communities (ranging from bioinformatics to healthcare). It also enabled several NRENs to begin to create Open Access Repositories (Ethiopia, Nigeria, Somalia, Tanzania and Uganda) that are servicing their scientific communities (e.g. this has enabled the Africa Centre of Excellence in Phytomedicine and Research Development (ACEPRD) at University of Jos (Nigeria) to create a repository of over 200 local plant species) ( E7). The most successful of these formed the basis of the National Academic Digital Repository of Ethiopia (NADRE - https://nadre.ethernet.edu.et//) with EthERNet (Ethiopia’s NREN). 47 public Universities are able to use this service to openly publish their work, data and theses. Academic end users benefit from this national NREN service (including many now using ORCIDs for the first time). This overcame the problems with maintaining institutional ones (around 10 repositories with only three working) ( E8, E9). In 2019 this led to the Ethiopian Ministry of Science and Higher Education mandating Open Science and the free access to all outputs from publicly-funded research.

The only report on the economic benefits of a NREN to its host nation (Canada) estimated an annual 5% increase in GDP through enhanced digital service provision and reduced Internet bandwidth costs for universities ( E10). These facilitated increased research productivity and higher numbers of highly qualified graduates entering the national economy. There is no recent data on graduates and their employment. However, as evidenced above, 3,800,000 students have received digitally enhanced education provided through digital educational platforms and services (access made possible through Brunel’s research) and have saved annually at least GBP81,700,000 in broadband costs ( E11). Finally, given that NRENs that Brunel has worked with are some of the most mature in Africa, if we estimate their effect on national economies to be 0.01% rather than 5% (limited by other factors such as high graduate unemployment ( E12)), then these NRENs and the services that Brunel researchers have enabled will have in 2019 contributed to an additional approximate GDP of GBP98,000,000 annually across Kenya, Nigeria, South Africa, Tanzania and Zambia (Malawi is still developing) ( E13).

NHS planning is typically driven by a mix of historical data and trend analysis. COVID-19 (a

novel virus), lacking such data, required rapid evidence analysis, synthesis, and simulation to

provide actionable local insight. Senior NHS managers (CIOs and direct reports) from West

London with Bell planned a collaboration strategy in March 2020. Brunel was then able to rapidly assemble academic teams deploying their research on evidence sifting (for model inputs), COVID-19 transmission and bed utilisation modelling. Research outputs were subsequently disseminated to the NHS through the delivery, presentation and discussion of 15 written forecast reports between April and December 2020, directly feeding into operational planning decisions of 2 NHS trusts, the London North West University Healthcare (LNWH) Trust and the Hillingdon Hospitals (HH) Trust, which together are responsible for the healthcare provision for 1,290,000 citizens. Scenarios were co-designed with the NHS as research was developed and validated. Forecasts were then disseminated to NHS analytics groups to further support short-term, elective and social care planning and transformation.

The epidemiological simulation toolkit has been used repeatedly from early in the first wave of

the pandemic [E8 and E9] to provide long term forecasts of COVID-19 spread to LNWH, and

was used during the second wave and after to provide forecasts for HH. This supported

healthcare delivery across West London in three different ways:

First, it informed the trusts’ operational response during subsequent waves of the pandemic

[E1], which led to LNWH arranging capacity in advance that benefited approximately 4,000

COVID-19 hospital patients during this period [E3]. Planning the zoning and repurposing of

wards was undertaken using a range of forecasted scenarios which included key decision or

trigger points. This was required as patient admission ranged from several per day to

approximately 1 per 10 minutes. As the pandemic progressed, models were refined and

adapted to better support general and acute (G&A) and intensive care unit (ICU) bed planning

(including surge capacity and stand-down points). This enabled the flexing of bed capacity

throughout 2020 from no COVID-19 beds up to 54 additional ICU beds and 44 acute respiratory

beds. The capacity increase also was aligned with the wider planning with local councils, such

that sufficient care home support and intermediate recovery beds were prepared for the surge

capacity scenario.

Second, it influenced the assumptions underlying the scenarios the trusts developed to help

them plan their non-COVID elective service recovery [E1]. Brunel’s forecasts provided early

warnings for new peaks which subsequently occurred, and quieter periods that occurred in

between. Both COVID-19 and elective service recovery planning affects the work pressure level

and well-being of approximately 2,600 relevant hospital staff (healthcare assistants plus

medical) and affects the care provisioning for approximately 1,600 elective inpatient and day

case admissions per week [E4]. In addition, the research helped LNWH to understand the

potential interplay of different factors in driving COVID19 admissions demand variations [E1].

Third, forecasts indicated the scale and pace at which capacity needs to be expanded and

whether it can be returned towards other health and care needs [E1]. The preparatory efforts,

informed by Brunel’s research, of both NHS trusts enabled them to facilitate more COVID-19

patients in their hospitals (629 beds occupied) during the peak of the third wave as compared to

the peak of the first wave (268 beds occupied) [E3].

The Brunel team was also requested by the Rapid Assistance In Modelling the Pandemic

(RAMP) initiative, which is coordinated by the Royal Society, to analyse the sensitivities and

limitations of the CovidSim code [REF5] using the VECMA sensitivity analysis and ensemble

simulation toolkit [REF6]. The results from Arabnejad, Groen, and others in the VECMA consortium were fed back to RAMP in a report [E7] and contributed towards a shift in SAGE away from sole reliance on the CovidSim model [E10]. The results are also published on the Science Museum web page for educational purposes to the general public [E6].