Impact case study database

Background

Schlumberger is the world's leading provider of technology and digital solutions for reservoir characterization, drilling, production, and processing to the energy industry, with product sales and services in more than 120 countries and employing approximately 82,000 people in 2020.

Schlumberger’s global R&D activities related to drilling and cementing operations, which represent nearly a third of the overall company, generated a revenue of USD10,000,000,000 in 2019. The portfolio for this covers a wide range of technologies such as drill bits, directional drilling tools, formation evaluation while drilling tools, drilling fluids, pressure control equipment, and rig control systems. An integral part of this business is the services that Schlumberger provides to the wider industry where their technologies are deployed by highly trained field engineers and technicians. However, this model has become extremely challenging to sustain - the company considers that the key initiative for their future success and differentiation is to move their services up through levels of automation. [S1]

Involvement in the 2011-2016 industry-EPSRC Autonomous Intelligent Systems program [F1,F2] led to then Head of Schlumberger Research in Well Construction Automation meeting Professors Maria Fox and Derek Long, and discovering automated planning. Schlumberger’s Vice President for Technology Development (Well Construction) observes [S1]:

“Schlumberger Cambridge Research has always been at the forefront of innovation. Automation has been an area of research for us for many years. The real step change in our approach to automate highly dynamic and uncertain drilling process started when we engaged Derek and Maria first as consultants and eventually as full-time employees.”

Fox and Long joined Schlumberger, in 2016, to bring onboard their research knowledge and complete the deployment of plan-based automation in the DrillOps software, a commercial implementation of an automated drilling function.

Product development, testing, and initial rollout

The initial phase of this process involved product development, testing and initial deployments for field testing. The second phase, of larger scale commercial rollout, began in late 2019 and early 2020, but has been disrupted by COVID-19. Nevertheless, DrillOps has already been deployed on 28 rigs in North America, Saudi Arabia, Italy and also off-shore. As part of the second phase, all Schlumberger’s different businesses have started to deploy the relevant parts of DrillOps.

What this means so far “is that, [the] directional drilling service in the US is now deployed through a system called Connect BHA, a DrillOps product, while […] trajectory planning and optimization in the US is done using DrillPlan. They will be expanded to other countries in 2021 as [Schlumberger] builds the in-country cloud infrastructure. Additionally, [Schlumberger] has globally deployed [DrillOps] automation in 28 drilling rigs with plans to increase to over 100 rigs that operates in our integrated business”. [S1]

Initial benefits derived from Plan-Based Automation

Schlumberger’s Science and Technology Manager (Automation and Planning) explains [S2]:

*“The work has provided wide reaching business impact, […] through the DrillOps autonomous drilling service, which is not incremental but marks a fundamental shift in the way we approach automation both within our own operations and in solutions we are developing with major oil and gas operators (for example with Exxon Mobil as announced in Q2 2020). Automation, through plan execution enabled by the work of Fox and Long, **creates consistency of operations, which drives safety, reduces time-to-target, and the minimization of environmental footprint.*”

In their 2020 Q3 financial report [S4], Schlumberger reports:

“In Saudi Arabia, […] DrillOps* automation well delivery solutions surpassed 63,000 ft drilled, achieving a key milestone for our Integrated Well Construction LSTK operations. The on-bottom rate of penetration (ROP) with AutoROP was 17% higher than previous wells drilled by the same rigs' field average. Furthermore, DrillOps controlled the preconnection, reaming, and backreaming operations, significantly reducing nonproductive time, optimized well delivery time, and contributed to a 30% improvement in on-bottom ROP and shoe-to-shoe run in a recent section of a horizontal well.”

This highlights the incentives, both to Schlumberger and its clients, of the automated drilling function, in terms of improved efficiency, consistency and reliability. In Q1 2019, the typical well in the Permian took 27 days (an industry-wide average, not indicative of typical Schlumberger performance) and costs USD595/foot drilled [S6, figure 2]. A typical well is 6,000 to 7,000 feet and costs at least USD4,000,000 [S6]. With a 17% improvement in on-bottom ROP, as shown in the Saudi Arabia example above [S4] (although more recent figures in reference to Schlumberger's DrillOps solutions suggests about 20%-30% improvements in efficiency [S7]), each well would save at least USD500,000 (a low estimate).

Assuming that drilling a typical well takes 27 days, a rig would typically drill about 12 wells per year. As evidenced previously, the automated system has been in use on selected client rigs since 2016 and is now deployed on 28 commercial rigs. Based on these numbers it is possible to estimate that the automation system, built around the planner and plan-execution components, is currently worth at least USD150,000,000 per annum in savings (USD500,000/well savings x 28 rigs x 12 wells/rig/year) to Schlumberger.

A key part of the Schlumberger’s long-term strategy

As the industry recovers from the COVID-19 crisis, Schlumberger has positioned itself to move quickly with its new technology to offer efficient, consistent, and cost-effective drilling automation, in order to maximise benefits for its clients and shareholders. As part of the company’s long-term strategy and its commitment to the technology, Schlumberger’s Science and Technology Manager (Automation and Planning) observes [S2]:

“The approach to AI planning has subsequently been investigated in other aspects of Schlumberger’s services business. In particular the automation of measurement operations (known as Wireline) and well-site service delivery, we expect to embed the approach across the organisation as part of the digital transformation to further realise the benefits.”

Additionally, Schlumberger’s Vice President for Technology Development (Well Construction) states [S1]:

“[The] goal with this product is to make it available to the wider industry and become the industry standard for drilling automation. A good example of this is the recent agreement that [Schlumberger] signed with Honghua where every new rig that Honghua build will have the DrillOps system built in it.”

The importance of this work to Schlumberger is also highlighted in the 2020 Q2 financial report to shareholders [S3] in which it is observed:

“As announced last quarter, Schlumberger and ExxonMobil are jointly working on the deployment of digital drilling solutions around planning, execution, and continuous improvement through learning. As a next step, ExxonMobil and Schlumberger have finalized an enabling agreement for the deployment of DrillOps* on-target well delivery solution in ExxonMobil’s unconventional operations. The technology is expected to enable faster, lower-cost wells through drilling automation”.

The Q2 financial report goes on to say:

“Schlumberger and Honghua Electric Co., Ltd. entered into a memorandum of understanding (MOU) for the seamless integration of the DrillOps on-target well delivery solution with all new Honghua rigs. Under the MOU, Honghua will manufacture and sell rigs that have plug-and-play capability with the DrillOps solution, which integrates planning and operations while automating well construction tasks in order for the rig to operate at peak performance throughout the execution of the drilling plan.”

As the globally leading well services company (reporting revenues of over USD32,000,000,000 in 2019), Schlumberger has consistently been part of investment portfolios for a wide range of pension funds and other large institutional shareholders, so its public financial statements are widely analysed and thus a key indicator of the direction it is heading strategically.

The future of Drilling Automation

To conclude, it is worth quoting Schlumberger’s Digital Strategy Development Manager in her article about DrillOps and drilling automation [S3]:

“ Digital and automation technologies present the best opportunity for the drilling industry to innovate its way out of the current crisis. However, seizing the opportunity will require preparation, collaboration, participation and commitment to transform the way wells are drilled, to move away from working in isolation to working together to capitalize on each other’s strengths. Doing so will not only help the industry to survive this downturn but to thrive in the future. When it comes to automation, all players have a key role to play, including operators, service companies and equipment manufacturers; however, rig contractors, as the owners and operators of the rig, are at the center of it all.”

Two companies have been founded to exploit Prof. Dai’s underpinning research: Movendo and AiTreat. In each company, one of the co-founders was a former research student of Prof. Dai. Each company has developed its own therapy robot. While much of the underlying research is similar, they use two quite different applications of robot technology with Movendo concentrating on a rehabilitation platform whereas the AiTreat robot delivers massage and acupuncture.

An independent evaluation has found that research at King’s led directly, and via the careers of the two former students and the technology of the two firms, to high quality products which benefits thousands of patients globally per annum. The larger part of the underlying technology is directly attributable to King’s research, although both firms also have added substantially to that core, for example in diagnostics and supporting clinical workflows, as well as extending clinical applications. However, it is unlikely that either firm would ever have been started or developed its technologies without the research inputs from King’s. [S1, p.19]

Movendo: technology development, founding, investment and company growth

Research conducted by Prof. Dai and his then PhD student Dr Jody Saglia [R3] (based on [R1, R2]) led to a key patent. The work attracted investment of over EUR10,000,000 from the Italian National Institute for Insurance against Accidents at Work (INAIL), that was helping to develop the ankle rehabilitation device “Arbot” and ran a successful RCT (Randomised Control Trial) on orthopaedic post-traumatic patients with lower limb injuries. The clinical trial and the patent led to the development of a new device for total body rehabilitation, Hunova, based on the core technology of the ankle rehabilitation device “Arbot” and the Patent. Hunova was been developed and clinically validated at the Italian Institute of Technology by a team including Dr Saglia, who raised EUR10,000,000 in corporate venture funding from Dompé Pharmaceuticals, one of the main Italian biopharmaceutical groups [S7] to start a new company, Movendo Technology, in late 2016. [S2]

Movendo Technology acquired an exclusive license for the Arbot and Hunova technology and now operates globally to commercialise its solutions in the rehabilitation market. Hunova is Movendo Technology’s principal product. It targets rehabilitation of the lower limbs, from ankle to knee and hip, the pelvis, and trunk, by providing a large variety of training modalities. [S6]

Movendo Technology Group employs 34 people (as of 31 July 2020), and annual turnover rose from EUR1,292,000 in 2017 to EUR2,764,000 in 2019. Its market share is approximately 2% [S11]. To date, it has sold 69 units. In addition to its core staff, Movendo employs around 1,500 agents worldwide. [S10]

In total, more than 10,000 patients have been treated at over 100 clinics and hospitals in countries including Italy, Germany, Czech Republic, Netherlands, Spain, US, Dubai, Canada, Greece, Saudi Arabia, Puerto Rico, Switzerland and Ukraine. Over 800,000 physiotherapy treatments have been delivered. [S11]

Dr Saglia, Movendo Technology, co-founder and former CTO [S2]:

“Prof. Dai’s contribution has been significant along the entire journey - from the very beginning of the concept idea to the final product design passing through clinical validation. This is a clear example of world leading research being successfully transferred to the industrial world, making a real impact on society. [Today] Movendo is continuously fostering a research and technology development collaboration with Prof. Dai’s group for its future products.”

AiTreat: Technology development, founding, investment and company growth

AiTreat Pte Ltd. (Singapore) has developed EMMA, a massage robot based on the technology proposed in [R5, R6, R7] by Prof. Dai and his PhD student, Chen Qiu. Following the receipt of an ACE Grant from SPRING Singapore (2015), the company has now completed three rounds of fundraising with the last round of investments coming from Brain Robotics Capital LP (a prestigious US VC), Tasly Pharmaceutical Group Co Ltd (Shanghai) and Ogawa Smart Healthcare Technology Group Co Ltd (Shenzhen). [S8]

“EMMA is AiTreat’s primary development and the first successful one of its kind in the world. EMMA is a state-of-the-art soft tissue therapeutic treatment system that utilizes robotic technology and medical knowledge that is stored using Artificial Intelligence. […] EMMA is equipped with sensors to measure muscle stiffness and uses 3D vision technology to analyze the patient’s body.” [AiTreat website, quoted in [S1, p.6]]

AiTreat Pte Ltd. has produced several generations of the massaging robot EMMA in four clinics in Singapore, with over 1,000 customers and over 5,000 times of use [S9]. The company turnover is USD460,000, 60% of which is due to using Prof. Dai’s technology. [S9]

AiTreat Pte Ltd was the winner of the Microsoft Developer Day Start-up Competition (2016), winner of 'Best Product/Application Design Award' at Shanghai International Start-up Competition (2016), and recipient of a StartupSG Tech Grant from Enterprise Singapore (2017). [S8]

Benefits of the robotic systems to medical practitioners and patients

The key benefits cited by practitioners who use the robotic systems are [S1, p.9]:

• Treating more patients, with better health outcomes. “Combining several types of fingering including thumbing, index fingering and palm motion in one for massaging to provide versatile massage treatments using one end-effector, multiple types of massage, ... what this does is improve efficiency, enabling a large number of treatments.” [CTO of AiTreat, quoted in [S1, p.16]]

• Cost savings for clinics and patients

• Treatment equal to the best physiotherapist, but cheaper and easier to reproduce consistently. "Treatment from our robots is about equivalent to the best treatment from the best physiotherapist but the latter is unreliable, expensive and hard to reproduce. Some traditional treatments require two to three persons to implement." [Director, Sales and Marketing, Movendo, quoted in [S1, p.16]]

• Consistent standard of treatment. "For example, [… with traditional treatment …] we give them exercises, [some] people don't do it, or we prescribe some random massage, it helps for a few days. But maybe it [results in] a misalignment of ligaments. We prescribe sessions on Emma for the back, and then acupuncture, with very good results." [CTO of AiTreat, quoted in [S1, p.16]]

• The flexibility of the robotic systems. “Adaptability - we can change from simulating rubber bands to swimming in a pool with the flick of a switch” [Director, Sales and Marketing, Movendo, quoted in [S1, p.9]]

The clinical efficacy of the Movendo devices is documented in published research as at least equivalent and sometimes better than conventional treatment, in particular for:

• Diagnosing susceptibility to falls in older people [S3]

• Rehabilitation for patients with Parkinson’s disease, after spinal injury and stroke [S4]

• Rehabilitation after lower limb trauma [S5]

Clinicians report that patients respond favourably to the robot treatment. “Patients are intrigued…They like the consistency and the [lack of ] worry about who [they] will get. We get a lot of word of mouth; they bring their friends.” [AiTreat clinician, quoted in [S1, p.15]]

The independent evaluation concludes:

“The two firms provide robotic technology, based substantially on research from KCL, which is part of a wave of clinical robotics which is likely to massively influence or even revolutionise the way clinical massage and acupuncture are delivered, and may extend beyond those core applications.” [S1, p.19]

King’s research has enabled a departure from legacy protocol and architecture design principles with significant impact on global telecoms industries and standards:

4.1 Impact on Ericsson’s [text removed for publication]:

Ericsson is the world’s second largest vendor in telco services, software and infrastructure, and 5G market leader with an annual revenue of USD28,000,000,000. “ 2.5bn subscribers and 40% of the world’s mobile traffic is carried over Ericsson’s networks” [A, page 1], thus powering half of the world’s mobile commercial and societal services. Ericsson and King’s have engaged in close joint research, innovation and deployment/experimentation over the past seven years. The underpinning research 2.1 and 2.5 achieved impact in four areas:

• **[text removed for publication] use-case developments: “ King’s played an instrumental role in gauging early 5G architecture capabilities and translating them into viable societal use cases. As a result, Ericsson has spearheaded many use cases [text removed for publication] . These have been documented on Ericsson’s website with specific mention of King’s inception; notably in health, education and gaming, and the arts and culture” [A, p.2]. Important world’s first were the output of King’s work: “ the world’s first robotic surgery over 5G […] which impacted the industry vision in the use of 5G for medical interventions […]” and “ UK’s first 5G 3.5GHz deployment with Vodafone, 5G sliced drone using EE’s and Verizon’s networks, or the world’s first 5G music lesson with Jamie Cullum”. [A p.3, B, C]

• Impact on [text removed for publication] : “Based on the pioneering use case developments and 5G prototyping, the work of King’s College London has* [text removed for publication] to address novel 5G use cases” [A, p.3] and “it […] is [text removed for publication] ” [A, p.2]. As a result: “ In confidence, [text removed for publication] ” [A, p.3].

• Skills development [text removed for publication] : “Ericsson was able to [text removed for publication] ” [A, p.2].

• Marketing, branding and societal outreach: “The work conducted by King’s […] has led to significant press and media coverage* [C] . Notably outlets, such as CNN, BBC, Wired, FT, have continuously reported on the technical advances of this work. [text removed for publication] ” [A, p.3].

4.2 Impact on Konica Minolta [text removed for publication]:

Konica Minolta (KM) offers a large variety of products in the form of office equipment, medical imaging, graphic imaging, optical devices and measuring instruments at a GBP7,400,000,000 annual revenue. It is market leader in enterprise printing products. King’s has engaged with KM over a years-long engagement via [text removed for publication]:

• Establishment [text removed for publication] : “The roadmap developed by Prof M Dohler in 2014-2015 outlined a set of opportunities for Konica Minolta around [text removed for publication] capabilities. Furthermore, King’s College London […] advocated for a [text removed for publication] . […] Following the KCL report, Konica Minolta focused on [text removed for publication] known as Distributed Cloud Intelligence (DCI)” [D, E]. Further, “the ongoing collaboration has been specifically useful for DCI and has helped [text removed for publication] ” [D].

• *[text removed for publication] : “An important outcome [of the joint projects] has been Konica Minolta’s [text removed for publication] [D, F]. The [text removed for publication] benefits are reported [text removed for publication] [D]; thus, giving an increased competitive [text removed for publication] [D].

4.3 Impact [text removed for publication] via 3GPP standards contributions:

The 3rd Generation Partnership Project (3GPP) is the world’s largest umbrella of seven standards organizations which develop cellular telecommunications technologies, including radio access, core network and service capabilities. More than 700 industry partners contribute to 3GPP, making it the largest standards body shaping the £-trillion telco market. King’s research on converged 5G telecommunications architectures was integrated into 3GPP with the support of BT. The standards contributions underpinned by the research have been fed directly into Release 16 and 17 Technical specifications [G], i.e. 3GPP TS 23.501 Sections 4.2.10. (Architecture Reference Model for ATSSS Support) and 5.32 (Support for ATSSS) as well as 3GPP TS 23.502 Section 4.22 (ATSSS procedures). The underpinning research 2.2 achieved impact in two areas:

• *[text removed for publication] 5G technical specifications: “The work that King’s and BT carried out on 5G convergence has been [text removed for publication] to support both fixed and mobile access technologies, and in particular the ability for operators to make dynamic decisions to route traffic to customers to optimise quality of experience in the most efficient manner” [H]. The contributions are specification standards, and essential to be implemented, hence confirmed to impact all broadband providers who have both fixed and mobile network infrastructure [H].

• **[text removed for publication] 5G feature rollouts: “This [convergence] feature is expected to be incorporated into vendor roadmaps and operator network deployments across the globe as 5G solutions evolve and mature towards a fully converged architecture, delivering significant benefits to customer experience” [G].

**4.4 Development of cost-efficient slicing template for largest telco association:**The Global System for Mobile Communications Association (GSMA) is the world’s largest telco association representing and influencing the business practice of more than 750 mobile network operators worldwide.

King’s underpinning research 2.4 impacted the pioneering concept of a Generic Slice Template that allows multiple operators to exchange the basic information of their network slice without revealing their technical details that could expose their competitive advantage. Hence, for the first time, user services can go across multiple operator domains with the exchange of slice templates between the involved operators. “These insights have provided independent thought-leadership for the GSMA to push for the creation of the Generic Slice Template (GST)” [H].

King’s work was thus included in the “The ‘5G Guide: a reference for operators’ [which] is the de-facto industry guidebook on the commercial, policy and technical considerations for the 5G business case. Following its completion, it was distributed to the GSMA Board for adoption at the February 2019 Board meeting and subsequently distributed to the CEOs of the 750 mobile operators in the world in April 2019” [H]. Importantly, “all the GSMA Board member companies, i.e. the [text removed for publication] largest telcos in the world, have now incorporated or are considering incorporating slicing into their 5G network plans, following a similar strategy as outlined in the document” [H].

King’s work “has impacted the industry’s position on how to optimise the creation and management of slices. Two specific recommendations from KCL stand out: i) optimise granularity of network slices so as to balance the need for more differentiation versus the costs of provisioning slices; ii) and [the creation of] standardised slice templates with predefined and optional fields that could bring down the cost and time to deploy of network slices significantly, guaranteeing interoperability and enabling automation of slice management on global scale” [H].

Research on provenance at King’s, along with the influence of standardisation of provenance at the World Wide Web Consortium (W3C) (Miles, PROV-DM), technology transfer to business (Curcin), toolkits and services to promote take-up (Moreau), influence of guidelines for data protection (Moreau), and, later significant world-wide adoption without King’s direct involvement, have led to the global recognition that provenance is a critical facet of good data governance and an essential function of an IT system.

Given the openness of W3C standard PROV, it is impossible to track all the usages of PROV. Furthermore, as PROV is used in the background, generally, as part of a data management function, it is challenging to identify its impact in isolation to the rest of the functionality. Thus, below, we explore key strands of impact, including explicitly listing publicly documented usages of PROV or PROV concepts [G].

4.1. Impacts on public policy and services

Many governmental organisations have a duty to make data publicly available: open government data is regarded as creating value (worth billions of pounds world-wide) in many areas, including transparency, democracy, participation, innovation and efficiency. We focus on two illustrations of open government. In the UK, the Gazettes are the official journals of public record, whereas, in the US, the U.S. Global Change Research Program (USGCRP, a cooperation between 13 federal agencies) publishes a quadrennial National Climate Assessment. Both have independently adopted PROV as a mechanism to improve public navigation and access to information through the use of linked knowledge, thereby addressing an overarching aim of information transparency.

“The credibility, trust, and integrity of the [Gazette] data has been strengthened because of PROV” [D, p.11]. PROV “works in the background to provide a clear, ethical and transparent data source” [D, p.1], and “ensures that the official public record is credible and accurate” [D, p.2].

In the US, the use of PROV had an impact on the environment, as the “policy debate on climate change has been influenced through the work of USGCRP: PROV-linked research meant that all information was meticulously documented which led to less controversy” [D, p.16]. “There have been a lot of scandals and difficult thinking in the scientific community about reproducibility and about accuracy and integrity, and […] PROV is a structure that can really support advances in general scientific practice to address those concerns” (Climate Adaptation Lead, USGCRP) [D, p.5]. Overall, PROV helps increase trust in data and processes, and one of its tangible impacts is the reduction of FOI requests, “as the accuracy of PROV renders many FOI requests pointless” [D, p. 13].

4.2. Impacts on practitioners

4.2.1. Impact on standards and standardisation bodies

The impact on standards reported for the REF 2014 now goes well beyond the original standardisation of PROV at W3C, which was specified in 2013 as a domain-agnostic ontology for provenance, building on research at King's by Miles on the requirements for provenance. Since then, we can report secondary adoption of PROV, which has been referenced and key concepts directly imported in specifications published by other standardisation bodies (HL7, Allatrope Foundation, RDA, IVOA), reaching out to 500+ member organisations. All these specifications have an impact on a range of practitioners, including software engineers, working on scientific data management systems and health care systems. These examples of secondary standardisation build on King’s research but without direct involvement from King’s, which is evidence of broad and sustained adoption of provenance. Specifically, Health Level Seven (HL7) is the international body for healthcare standards (500+ corporate members), representing healthcare providers, government stakeholders, payers, and pharmaceutical companies. HL7’s Fast Healthcare Interoperability Resources (FHIR) model is the standard for programmatic communication of health data, and it includes a mapping to the core concepts of PROV [E]. The Allatrope Foundation (a consortium of 40 pharmaceutical organisations) defined a universal data format that standardizes laboratory experimental parameters in order to remove human error and enhance scientific reproducibility; its Audit Trail ontology is building on PROV [G].

The Research Data Alliance (RDA, 50+ research organisations, 11,000 individual members) aims to build the social and technical bridges to enable open sharing and re-use of data. The RDA Recommendation [C, p10-11] incorporates PROV as metadata for the management of research objects in data centres to support reproducible research. The International Virtual Observatory Alliance (IVOA), comprising 20 Virtual Observatory (inter-)national programs, has published its recommendation ProvenanceDM applying PROV to astronomical data [H, p.13].

4.2.2. Impact on guidance for AI practitioners

The General Data Protection Regulation (GDPR) is the European wide regulatory framework that codifies some rights for data subjects (the users who have provided data in return for services) and obligations on data controllers (the organisations that are providing these services). A key challenge is that regulatory frameworks remain high-level and do not specify practical ways for organisations to become compliant; this is a particularly salient problem for companies adopting AI in their new products. The Information Commissioner Office (ICO), the data protection regulator in the UK, published guidance around “Explaining decisions made with AI”, explicitly referring to provenance: “Such provenance information provides the foundations to generate explanations for an AI decision, as well as for making the processes that surround an AI decision model more transparent and accountable.” [F, p.59] The ICO report [F, p.59] includes a link (https://explain.openprovenance.org/loan/\) to King’s demonstrator for provenance-based explanations [4], which in collaboration with ICO was tailored to address seven key requirements of the GDPR. The software system, the first of this kind, provides a tangible artefact to which implementors can refer when implementing high-level guidance. The mark-ups and associated queries over provenance identified in the research were instrumental in delivering fit-for-purpose explanations and present a significant advance for AI practitioners. Appendix 4 [F, p.124] further refers to the King’s report [4].

4.2.3. Widespread adoption of PROV

Two fundamental characteristics of PROV is the simplicity of its core and its design based on a wide consensus, promoting interoperability, as exposed by King's research [2] and provenance toolkits and services hosted at openprovenance.org. For these reasons, several communities have coalesced around PROV (or PROV concepts) with a view to facilitating the interoperable exchange of provenance. The key driver for these communities stems from their work with complex workflows, involving multiple stakeholders, typically each with their respective IT systems, across which flows of data and decisions need to be documented to ensure their auditability or reproducibility. A growing list of adaptors (in excess of 50, counted November 1, 2020) is maintained at [G]. For instance, the environmental science community is adopting PROV across the world, as illustrated by a range of projects, in the USA (NASA, JPL, PNNL), Germany (DLR), Austria (EEA), and Australia (Geoscience Australia, CSIRO). There is adoption of FHIR provenance in the HealthCare community, including companies such as Astrazeneca, Smile CDR and Perspecta. In the Astronomy community, Applause, a collection of photographic plates with full provenance, is operationally deployed by the Leibniz-Institut für Astrophysik Potsdam (AIP). Finally, innovative commercial products are exploiting provenance capabilities, Surround Australia, Smile CDR, Perspecta, and Imosphere, which we specifically discuss in Section 4.3.

4.3. Impact on Imosphere Ltd’s commerce and practice

Imosphere Ltd is a company of 50 employees commercialising solutions for healthcare organisations. Their flagship Atmolytics product is a tool that produces interactive reports from patient cohort data. Thanks to an Innovate UK grant (KTP-509790), Atmolytics incorporated elements of Curcin's provenance template technology, to enable a range of new functionality in the product, such as a full audit trail of data sets, versioning, analytics and reports providing transparency and increasing the users' trust in the data findings. Provenance templates were deployed to model key Atmolytics behaviours, such as patient cohort management and decision making, and ensure a faithful provenance record in a standardised PROV format. The provenance module differentiates Atmolytics from the competition, by its functionality to maintain the integrity of data, supporting informed and trusted decision-making [A].

The new, provenance-enabled version was launched in the summer of 2018, and has been installed at customer sites in the UK, Europe and USA. There are in excess of 4,000 users in the UK and USA, managing data of over 1,000,000 patients (by November 2020). Among them, 60 councils in the UK are benefiting of provenance functionality to manage their individual care budgets. In the US, the National Cancer Institute “City of Hope” tracks over 100,000,000 medical events, whereas the medical school Meharry Medical College, Tennessee, tracks 250,000 patient’s data. For the end-users of the system the “network graph view of provenance data is a far more natural way of visualising and querying historical relationships between patient cohorts and analytical tasks”. [B]

Overall, due to PROV, Imosphere has benefitted significantly: “Using the W3C PROV standard and provenance templates for this task, saved us years in design and development time and ensured we are standard compliant for any further extensions, reducing time to market by approximately one year.” Furthermore, “the introduction of data provenance capabilities in the software, has also improved the software engineering aspect of our data analytics portals, as it promotes good practice in reusing and documenting analytical components across reports, avoiding duplication”. (Imosphere CEO, [B])