# Notes

## Attendees
- David Wagg
- Janine Cooomber
- Scott Hosking
- Sophie Arana
- Kalle Westerling
- Nahim Iqbal
- Karen de Cesare
- Chris Burr
- Cass Gould van Praag
- Kirstine Dale
- Steve Nieder
- Chris Charlton Mathews (remote)
- Katrina Payne (remote)

## Applogies
- Keith
- Eric
- Ben McCarther

## Table of Contents

[TOC]

-----

# Scientific Vision update
- Draw a line under the [vision and strategy](https://thealanturininstitute.sharepoint.com/:w:/r/sites/TRICDTLeadershipteam/_layouts/15/Doc.aspx?sourcedoc=%7B0A77D3DE-6435-4306-8B0D-8FA9EE24B7AD%7D&file=TRIC%20DT%20overview_v2.0.docx&action=default&mobileredirect=true&DefaultItemOpen=1) which has been developed in the TRIC to date, and move to Vision 2.0
- This should be shareable - we can point people to it
- No strong views have been voiced in the meeting about this document, so it is taken as approved
- Objectives - These have not changed in the new strategy
- Advance science and implementation
- build open and reproducible tools
- Build a CoP.
- Now move on the oporational phase and impact in the broadest sense.
- Impact includes outputs, influence... etc.
-
## Lead updates

### Health (Steve)

- ([slides](==get this link from Janine==))
- Want to make tools for DT of the heart **at scale** from patient data
- PDRA Marieka appointed to start in April 2024
- Other appointees were not able to commit.
- **AutoEmulate** - library of emulators replaceing a comlicated model with a low cost model, to make them more accessible to the wider community, with robust statistical inference. (Martin Stoffel working on this)
- CB: Hub are well placed to contribute to this. SN: Designed to be cross cutting over themes, and open source. Goal to make something easy to use for mechanical engineers.
- How to open up new tools to practictioners that don't necessarily have the technical knowhow (beyond running programs in matlab)
- Community of Practice would require shared computational language to make outputs sharable.
- CGVP: Never to early to build in community discussions. SN: Noted that this is a new way of working and will take some adjustment.
- Emulators as potential touchpoint across themes
- Submitted an EPSRC programme grant with ATI, Sheffiled, Nottingham (==timeline==)
- Sumitted £1million grant with Molly Malakar to Norwegan government on menopause - wiating to hear (NORA - norwegian artificial intelligence research cluster)
- Patient models to guide pediatric cardiac surgery, should hear ==waht?== in october
- PDRA starting 17th October for ==...==
- Ran Isaac Newton workshop Intersection on Uncertaintty quantification, AI
- Contributions to QQR
- Policy engagement - contributing to review ("Unlock the power?")
- 5 papers published this year with Turing affiliation.

==Note Trina askes if there is a publications section on the website. Cass to loop her into webite updates==

### Natural Environment (Scott)

- Interest in bringing solutions to remote regions especially Global South (currently antarctica)
- Need for digital twin to get updates in real time about icemelt.
- Calving event January 2023, ice flows into ocean faster now (4m per ..?) ==took most of the year to identify this== we need better ways of identifying this type of event, faster. (DT tech = good response)
- 4 million grant accepted through EPSRC.. (on data fusion, filling data gaps, increasing resolution, decision making)
- Also building Emulators for ice sheet which can work across multiple resolutions
- These models are carbon intensive, so need to make sure qwe're running the right simulations to benefit the uncertainty quantification (UQ)
- Open source tool, in Julia.
- Seasonal forcasting of Sea ice
- AI model outperforms the leading physics model state by 6 months (with monthly resolution)
- If you can predict forward and the prediction is incorrect, you gain feedback which can be used to adapt the pysical sensing etc.
- Plans for combining all this information into DT
- 3 year plan/endgoals: sea ice predictions and sea level rise.
- The AI models and components can be far more beneficial when they all come together in a larger national ecosystem
- Met Office :==mised the updates on this, lets get Scotts slides==
- 2nd aim about building a community of stakeholders - Kirstine leading this,
- Joining the dots, and getting people to talk!
- work includes IMFe (runs out of funding in October)
- Workshop 18th October which brings together some key stakeholders ==get invite to this if possible - Chris is going?==
- key outcomes
- asset register
- convienging
- way forward for IMFe after funding runs out (what happens to their CoP, which parts can TRIC take over?)
- Groups who might take activity forward
- TWINE (NERC + Met Office)
- DT Hub
- National DT programme (name = Alex Luck?)
- NERC has set out a couple of calls - expect TRIC to be sumbitting (==Cass: I thought we talked about supporting but not submitting to the TWINE calls?? apart from TWINE, which other calls?==)
- SN: Would the library of emulators be useful to your emulator project?
- Look for where we can share code and not duplicate effort. 80% of your effort is building the frameworks. Should conenct on APIs etc.
- KW: There is a fornightly community call for Deep Sensor which this could plug into.
- [name=Kalle] **TODO**: invite Martin to community call
- DW: Would it make sense to have more specific leadership about building CoPs? Do we need to identify team members to take leadership on this.
- SH: Would be useful to pencil some names against our objectives. Would be useful to see where we do have effort.
- CB: There are a number of routes into CoP work, which we have started developing in the Hub.
- KD: IMFe recommendations guides everything in TWINE, and how the recomendation map into the scope of activities in our partners. Can work on which recomendations should be carried forward by the Hub. There are several groups who are working on this and they are not fantastically connected.
- suggestion/**TODO**: revisit those recommendations. ==share slides that maps out which recommendations TRIC will take on== [name=Kalle] Does Chris have access?
- SN: There are a lot of communities in health, so haven't felt the need to build new ones. From a Turing branding there is value, but not sure there is need. What it the drive in environment?
- KD: There are a lot of unconencted groups, so the push for us is to push together the unconnected groups, we just nede some leadership and oversite.
- CB: Agree, we need to bridge links.
- KD: Can we let the health communities know what we're doing?
- SN: Need to know what we're "selling". More inclined to wait until you have results so you done loose credibility.
- KD: Two tracks: building prototypes and building community. They are parrallel
- SN: the pitch for community would be about connecting groups.
- DW: Let's pick this up after other updates
- ==Cass: inverted pyramid of community vs infrastructure deveoplement==

### Infrastructure (David)

- UK assets are getting old (railway, some victorian bridges made with 100 year lifespan, but still going.)
- 2018 proejcted cost of maintaining infrasticture ver next 10y = £600 billion
- Early 2000s commitments ot new energy sources. These have been on life extensions for the last 20y - current policy is unclear
- recent Offshore wind contract no one bid for (==[name=Cass] I understand they were under costsed by gov.==)
- Don't plan for policy and funding to come in from Gov.
- vision is for DTs to help accelerate decarbonisation, enhance resilience, reduce operational costs
- Aim to reduce operational costs
- UK has high building and operational costs compared to other countries
- Facilitating new transport systems - DT could be importnant here.
- Envisage that TRIC would provide focal point for academics, be a place for collaboration and acceleration
- How can be signal something different from DT Hub?
- Research staff appointed
- Chris Lindley - MAchine transfer learning
- Ziad Ghauch - learning and reasoning methods
- Jack Pool (DfT finded on short straights) Traffic forcasting
- Xiaoxue Shen - knowledge models for decisions support
- Keen to work collaboratively across themes
- Concrete ideas?
- Events:
- Oden Institute visited
- AIUK 2023
- Newton Institute 1-day meeting in Cambridge in June
- Planned event at Georgia Tech in early 2024 (Jan/Feb)
- Bristol airport claims to be operating at net 0 - would like to bring them to this meeting.
- Partnerships
- Fujitsu = interest in DT and beyond (Trina working on this)
- Oden + Georgia Tech (visit in early 2024)
- David + Adam Sobey going + others?
- Need to think long term about maximising value form research visits
- knowledge models for digital twins
- include traditional ML, more "silicone valley" tech stack, LLMs etc
- nice images here - ==get hold of slides==
- Ontological knowledge models for DTs
- Network+ - Turing leading
- Appointed manager and interviewing for coordinator
- Need to work out how this will be integrated with TRIC.
- Objectives
- Knowledge exchange
- accelerate research
- DT research agensa
- pilot research and feasability
- Outreach, skills and sustainability
- diagram borrowed from cyber physical infrastructural consultation doc ([name=Cass]: I have a screen shot of this one - note the "consultation" part of this - this is why commuity should come first)
- Large focus on bringing communities together, which we need to align with TRIC
- Case study development likely coming up here
- Ex: Wessex (?) basin
- biodiversity + affect by polution, economy = crossover between infrastructure and other themes here
- [name=Sophie] Not sure where this case study originated from.
- [name=Kalle] I think they came in from Network+ somehow.
- SH: How to engage with biodiversity and built environment, as distinct from built environment. There is a gap on conencting our work - maybe another co-director or theme. Thinking with KD about brining in resource and what specific skill sets/expertise would be needed
- [name=Kalle] Are there communities across Turing where we can connect? Alden? + People working in Scivision + connections across ai4environment Slack team?
- [name=Sophie] many of BAS researchers emphasize how they want their work to also be relevant for biodiversity. Building trust in their community for the type of AI solutions, seems to be one challenge
- CB: re alignment effort; **where** should we embed groups? We don't just want to duplicate.
- [Doc from British Standards Institute](https://alanturinginstitute.enterprise.slack.com/files/U03KXHCSEHH/F05NS5HG4UU/iot-1_n105_jtc1-sc41-335-dtr__iso-iec_tr_30172_ed1-_digital_twin_-_use_cases___1_.pdf). Also, [thread here](https://alan-turing-institute.slack.com/archives/C04RTSJKV4M/p1692609401706549). Creates agreed upon scheme for case studies, which means there can be more formal analysis of the cases - generating a systematic repo
- KP: Need to make sure we're focusing on outputs and deliverables as much as inputs.
- SN: A lot of the goals align with the DT Hub. Do we have an alignment (David is on the panel). Is it worth having a representative from DT Hub in TRIC?
- DW: We don't want to replicate. Discussion to be complimentary
- KD: Inviting them to sit on our Advisory pannel - JC: This is on pause

### Hub (Chris)
- Engaging with researchers - Kalle and Sophie
- Now a clear need for support for new PDRAs - developing and testing [onboarding doc](https://hackmd.io/@tric-dt-hub/HJ2jbArkp). this will inform part of the broader ways of working (WOW)
- Need to make sure we have the infrastructure in place and that we can communicate it. Clearly artilate where there is touch points for engagement.
- Our objectives:
- advance research to address technical + societal challenges
- open and reproducible computaitonal tools
- CoP
- Ensuring impact in line with
- TEA funding application submitting in October
- If we don't get the funding, there is still good opportunities - DT Hub has understood value and need.
- Stakeholder engagement - at the moment our work has been focused on the Natural Environment Theme, so need to broaden and extend.

KP: Need to think about how we can highlight DT partnerships ==Cass shedule an in-person meeting with Trina==

## Ways of Working

- Now is the time to mobilise and delvier the objectives of the TRIC
- [WOW document](https://thealanturininstitute.sharepoint.com/:w:/r/sites/TRICDTLeadershipteam/_layouts/15/Doc.aspx?sourcedoc=%7B2BFD020A-EEAD-4F27-A7F8-5E789D4F2BFE%7D&file=WaysOfWorkingV1.0.docx&action=default&mobileredirect=true)
- Good questions and comments already
- All up for discussion
- Can be both internal and outward looking
- NEed to think about timescales and urgency/importance
- Urgency is about the internal stuff as we have. team starting and need to have things established so they can beenfit from them
- Outward focus
- CB: crucial is pathway for incoming PDRAs with [onboarding](https://hackmd.io/03LOiDziTu6dZi-tmyZiVA?view)
- Don't want to overwhelm
- First connection should be to Cass (via Janine) as welcome point, then we will think about the best route for their interest/expertise and funnel them into the relevant activities which .
- E.g. reading groups; special interest groups
- SH: A lot of our activity sits between themes. Need to make sure we're managing resources appropriately across the TRIC and the GC
- Could hire in people with explicit "2 hats" - design it in - "you are sitting in the space between - 50% of time on TRIC, 50% on E&S"
- This could be a model for connecting with wider Turing
- What are thoughts from other teams?
- DW: What is a "Cluster" - Wolfson have a definition?
- Researcher need and identity, often this is a Centre, need to be mindful of uis/them.
- clustering is bringing people together across boundaries = could be an external comms piece.
- building more agility than existed in programs and doesn't exist in grand challenges either
- SN: Risk that we fall into Turing ambiguity between GC/TRIC theme.
- CR: Valid risk but internally, our reporting can happen in both: How something was important for the TRIC, and not.
- DW: Important that individual researchers don't feel confused. Don't feel inhibited when talking to others.
- CB: Permeable boundaries
- SH: DT work happening at Turing but outside TRIC (been approached for TWINE call) = Matt Fry for example. How do we bring them in and make them feel part of TRIC
- SN: Horizontal v. vertical expansion. Getting all co-directors in the room... do we need another tier? Are they funded levels? Volunteer? where does the funding come from? Research fellows coming in might be interested in leading on cancer research, neurology... What's the funding here?
- SH: We need wide, horizontal view. But do we also need "project owners"? Free up some capacity from having a project level focus vs programme overview.
- JC: Do our PDRAs have appropriate seniority / experience?
- SN: Need someone who has done a couple of postdocs and leadership experience (eg lecturer) / Advocate for bringing in people full time
- SH: Challenge that people have dual affiliation - need people 100% Turing
- SN: Challenge at home insitutions about naming your time on grants. Could be more flexible with a 100% Turing employee
- CB: To what extent does Network+ model enable an infrastructure (of levels) like that? (Co-Is there already in chare of specific WPs)
- DW: from UKRI point of view, we promised we'd coalesce the academic community around DTs, the other networks and hubs don't have that focus, neither a focus on AI specifically.
- DW: What Scott was talking about is more like a fellowship
- CB: what about combining them?
- SN: Six co-Is, selected from knowledge, not likelyhood to collaborate with TRIC. Those people will be running themes.
- JC: We're custodians of Network+, can't be seen as funnelling into TRIC. Needs to be distinct.
- DW: We got additional funds we can leverage for TRIC project. We don't have to reapportion TRIC budget. Meetings, events, community building, outreach.
- SN: Build tools what have a common language across the themes (eg knowledge graph). We wrote independant proposals, and unsiprosingly have 3 indpendant workstremes. Lets ==get together and develop stuff together
- Cross-theme output (large paper with 34 authors)
- Most urgent things: PDRAs are starting. Giving them ownership of things. Asking one of them to take the lead on a seminar program around DT and a journal/book club activity.
- SH: Are there researchers with DT experience? We should look at their strengths.
- DW: Overlaps with engineering, science... Idea is KG will be interesting and emulators. Already exists as ideas.
- CGP: Ziad has interest in reproducibility. So as a RCM, that's interesting/ Look at what the expertise is, an let them come up with stuff. We can also help with raising leadership qualities.
- SH: We should look at passion within the team. Seminar series on someone might not get them out of bed in the morning.
- SN: What's the culture at the Turing? 3 months at Imperial.
- CGP: TTW/TPS practices. Energy that it takes to build this, we have it. It's tiring for one person to run. But we have infrastructure that we can plug into.
- SH: We should bring researchers together.
- CGP shared a slide of Brian Nosek's idea of behaviour change ==?== into reproducibility. Build infrastructure first, then along the pathway, you build communities. placing community at bottom, they can contribute rather than "build it and they will come" - they will help build the thing that they want and need. Ex: TTW project and community growth slide.
- KP: "What Cass is presenting is a REAL attractor for partners to work with us!"

## TRIC-DT Theory of Change

- whatever we do, we'll make impact happen (Jane Goddall quote).
- the point: we get to decide what the difference we wannt to make.
- Structure
- Identifying your end goal
- Theory of change
- Next steps
- Identifying your end goal
- Activity 1
- It's 2043, what difference did the TRIC make in the world? They aren't researchers, don't know the details... Economic, social, environmental, health-related improvements in the world? Assume the best case scenario in the project
- Cass, Chris, Karen (hub and pmu)
- unified standards to enable sharing of code and data
- procedural standards, agreed upon
- culture change from quick and dirty, competative => collaborative, which is slow and difficult!
- Open standards, rather than closed, signing up to a working group. Not free to download. This is because there is a liability which they carry. Only industry can access them so there is a big s,kills gap. If someone sues you, need to show you've done your due diligence and show that you've followed a named standard. Do we need something to carry liability?
- CCM: Standards are a big part of the conversations at Turing. Without the standards, some people would do things less well. What practice changes as a result of putting these practices in place?
- Speed, transfer of information, collaboration. People able to contibute who previously weren't (not wasting time on working with their own inefficiencies)
- Scott, Nahim, Kalle, Sophie
- built infrastructure. It has become part of the world.. digital twins just "are". We see changes in a system, and know what they mean. That would be success for us.
- you don't work on the DT, you work on feeding the DT.
- if it would take a matter of weeks, it'd be great = we wouldn't start from scratch.
- business as usual = upskilling has been significant.
- DT for monitoring or controlling the assets?
- sustainability is at the forefront = our training and predicting is operationalised in a carbon-efficient way. We ensure we're doing this work for a reason. = You can lose precision where you don't need it.
- CCM: New way of doing things. Making this technology a normal part of the landscape - why would
- Janine, DW
- Travel and infrastructure, without trouble. Accessing rail and road efficiently. Electric cars charging easily. All of these systems are coordinated on an individual level. You know where your system is and its status at all points.
- CCM: Cost savings, efficicencies, environmental benefit, democratisation of travel
- KD, SN
- Normalised the use of DTs. Everyone has a DT to inform their personal health and lifestyle decisions. Use DTs to interact with the envrionment and mitigate our impact of the env. Robots moving among us which are linked to an internet of things+ - richer sources of information than current.
- Clearly TRIC is a program with transformative impact. Exploring this means we explore the diveristy of this impact. But also not getting lost in the weeds of unusable frameworks.
- Theory of change
- articulating the difference you make -- focused on the outcomes (can be small = After you've talked to someone, they've changed.)
- ToC: putting it into a model. Good for articulating for stakeholders (and how they will be affected). Good for building shared understanding, how we can improve, communicating our story to stakeholders. Some funders are also asking for them. ==If funders ask, CCM can help us with them==
- Example - clinical trial
- Example - Turing's ToC
- Cross-fertilisation focus in many programmes across Turing = improving quality of research more widely. Or accessibility, sustainability...
- Q from Janine: Is there an ask for Turing ToC being translated into programme level? Bonus points?
- CCM: No. You do what you do.
- Q: Why is "human health" and "all life to thrive" separate?
- CCM: They are. Related to Grand Challenges. Have already lead ToC workshops with E&S, and this is the outcome they landed on. Workshops to come with other GCs, and that may update the other outcomes.
- Q: SH: is "data science and AI" sufficient to cover DTs? KD: If it doesn't cover ==digital solutions/infrastructure?==, then no.
- KW: RAMs are tasked with impact. So much of our role is at the level above the "research and innovation" - What can we do to translate our objectives into this, build alignment over the 4y of the TRIC.
- CMM: Agree. This ins't meant to be a project plan, the level of detail isn't there. Need to unpack it into a version which works for us and directly reflects the work we are aiming to do. It should then be possible to align what we're doing into the Turing ToC.
- you don't have to think about this muhc-- it will come down from the Turing...
- [name=Kalle] I think differently about this, where I think it's a METHOD for us to apply internally.

- Next steps
- CMM: Do TRIC see value in this?
- DW: Would very mich like us to do something of this type, with CMM facilitating. Whether PDRAs should be included? DW thinks yes.
- [name=Kalle] Yes, PDRAs should be included, argument lies in Cass's shared pyramid (Brian Nosek) = we need buy-in from them — and _their input_.
- CMM: small group working on a high-level ToC?
- ...
- CB: Where does stakeholder mapping fit in wtih this?
- Part of ToC workshop would be stakeholder mapping... how infle=uential are stakeholders on us and how important is it for us to influence them?
-