# HDR UK Early Career Researcher Forum


Thank you for being part of this vibrant community. Let’s make the most of this opportunity to learn, share, and grow together!

We are delighted to welcome you to the Early Career Research Forum! As your co-hosts, it is an honor to bring together such a dynamic and passionate group of emerging researchers and technologists from diverse fields in health data science.

This forum is designed to foster collaboration, exchange fresh ideas, and provide a supportive space to navigate the challenges and opportunities unique to early-stage researchers. Whether you're here to present your work, engage with fellow attendees, or learn from the insightful sessions.

After the forum today, please join us in building a community of early career researchers, so we can support one another and take the objectives of this platform forward to inspire and empower you into the future.

Warm regards,
Fatemeh Torabi & Elliot McClenaghan
Early Career Research Forum Co-Hosts

Share access to this notebook with your peers
[Access link](https://hackmd.io/@1WiKcWdxRSiQet-oTOioHw/SylLzHjhhA)

# Agenda
To access the full agenda use [this link](https://www.hdruk.ac.uk/events/inaugural-hdr-uk-early-career-researcher-forum/)

## Day in a glimpse

# Industry partners

Make sure to visit the exhibitions presented to you by our industry partners
* Genomic England
* HDR UK PPIE
* GSK
* AstraZeneca
* AWS
* Alumni Network

you can locate them usign the below floor map

## Vote for questions via mentimeter polls or submit your questions in note sections provided under each talk

go to: www.menti.com
Ender the code: 4442 2957

or scan this QR code


# [10:00-10:20] - Opening Address

Research can be both thrilling and daunting, especially at the early stages of your career. Our goal with this forum is to connect you with peers, mentors, and resources that will help you advance your work and broaden your horizons. Together, we will explore new perspectives, discuss cutting-edge developments, and build networks that will last long beyond this event.

## Talk Highlights
**Do not give up**
* Reach out for any advice
* Fate, luck, sheer grit and determination
* No two career pathways are the same, and it's important to do what you enjoy so it doesn't feel like hard work all the time, and not worry that things are not necesssarily following the same path as other people.
* Experience through placements and keep self reflection and re-evaluation of path forward.
* PhD is hard work/life balance but you will survive it as we did
* Having support network, suround yourself with great people

**Top tips**:
* Find your focus and soround yourself with supportive people
* Fail and allow yourself to fail - it does make you stronger
* Don't give up
* Be nice to people : you never know who you are gonna meet on the way up or on the way down
* Be different : do what you want to do and find a way to do it
*
## Vote for the questions
go to: www.menti.com
Ender the code: 4442 2957

or scan this QR code


## Submit your own questions
Please use this space to add your own questions

# [10:20-11:05] - Fellows & Alumnis

## Rona Strawbridge

### Talk Highlights
* Be visible
* Be strategic
* Right place, right time*
* Luck= awareness of opportunities
* Consider different type of research dissemination such as the conversation
* Keep a list of your activities in hand
* Mentors are so important

## Honghan Wu


### Talk Highlights
* From gamer to professor
* Non-traditional career route to academia; advantages & disadvantages?
* Keep yourself interested.
* Trust yourself and get support when necessary.
* Find mentors
* Seek suggestions
* Growing your network

## Kluivert Boakye Duah

### Talk Highlights
* Importance of mentorship
* Finding mentors among busy senior academics can be challenging; how best to overcome that?
* Work life balance in more developed stage of career
* Academia or Research, which one is the right way?
* Talking opportunities, being visulant for availabel opportunities + luck
* There are examplary candidates in different networks
* Bring the change

## Vote for the questions
go to: www.menti.com
Ender the code: 4442 2957

or scan this QR code


## Submit your own questions
Please use this space to add your own questions

### Suggested questions:
* What advice specifically on preparing for and securing a leadership role can you give?
* Ronna mentioned an up-to-date CV as a good way to progress, are there any resources you use to construct and update your CV that others could make use of?
* On trusting yourself, how can you decide whether to accept or reject a piece of advice from a trusted mentor?
* What is a practical first step for finding a mentor?
* What do you think of "reverse" mentoring schemes where a more senior professional is the mentee?

# [11:05-11:15] Comfort Break

# [11:15-12:15] Doctoral Science Showcase

## Fabian Falck

### Talk Highlights
* Generative AI models
* Learning data distribution and using the generative model in different conditions
* Clustring high dimentional data: lots of different assets presented in the dataset - how to go about it where there are multiple and diverse characteristics on various facets?
* Exploratory data analysis (EDA) will feed into your approach onto this

Steps of EDA
* Visualising each facet in the data
* Finding homogenous examples vs hetrogeneous facets
* No relationship or inferencial activities happening at this stage
* Hierachical
* Looking at existing Units in data
* Unit is an approximation of a data piece
* Units use average of 2 coefficients to approximate but there is an alternative approach to this...hierachical VAEs
* Fusion models current state-of-the-art

## Emma Whitfield

### Talk Highlights
* Diagnosis is a challenging task compounded by various factors
* Compare frequency of a set of characteristics based on their occurance during the process of diagnosis
* Why emergency presentation matters: there is a well-established corrolation between ED presence for different types of cancer
* It is important to depict similaries in ED presence for cancer vs other disease

## Haoting Zhang

### Talk Highlights
* drug combination synergies
* AZ public dataset including more than 11000 drug combinations
* A combined therapy is more effective than single therapy
* There is no gold standard for synergy score and it is framework dependent
* Ideal drug: maximum efficacy, minimum potency so a small dosage will provide required therapeutic target
*

## Vote for the questions
go to: www.menti.com
Ender the code: 4442 2957

or scan this QR code


## Submit your own questions
Please use this space to add your own questions

For Fabian:
* How much uptake (in industry & academia) is there of the state-of-the-art fusion models you described?
*

For Emma:
* How can the diagnostic process algorithms be translated across different primary care systems/countries?
* Is there any risk of time-related biases in these analyses?
* How distinct is the ED presence for cancer vs other disease?

For Haoting:
* Has these findings made their way to clinical Trials or drugs?
* What are the limitations of using Bayesian distributions versus point estimates ?

# [12:15-13:15] Lunch break

# [13:15-13:45] Programme Fellows

## Amal Rammah

### Talk Highlights
* Birth cohort from 2006 onwards
* within TRE and open to collaboration
* Question: any effort or possibility focused on reducing the cost of data access? the PI of project has this in mind

## Cosmika Goswami

### Talk Highlights
* PhD in mathematical Statistics
* Current work focused on medicines in Acute and chronic care programme
* Prevelance of high dose antipsychotic medications
* started July 2023-Data of all prescribed items from Scotland

## James Mitchell-White


### Talk Highlights
* LLettuce: mapping the common Data Model easier
* OMOP CDM
* Process and build queries based on standardised concepts

## Naomi Launders

### Talk Highlights
* Health enequalities in cardiovascular disease in people with mental health problem, CVD and cancer
* Accounting for deprivation
* Primary care + Cohort/Census data for CVD and cancer screening
* Multi-state model for disease progression

## Thomas Starkey

### Talk Highlights
* DOAC use in patients
* Evaluation of risk of bleeding
* CHADSVASC or other CVD risk schemes and how are you going to evaluate them?

# [13:45-14:45] Workshops

## Tutors profiles

### Matthew Forshaw
**Developing a resilient Artificial Intelligence profession: observations from the UK’s skills landscape** 

Location: Main space, 6th floor

#### Notes from workshop
Please use this space to add notes from this workshop

AfDSP: alliance for data science professionals.

AI Skills for Business Competency Framework -> section for AI professionals.

AI upskill government fund in progress.

Turing data science and AI educators' programme -> 12 week course in evidence-based pedagogy.

Message from Matt: Thank you for having me at the event. If anyone would like to chat further, please do get in touch with me at [e-mail], any time!

### Emily Ball
**Leading on funding applications and stimulating your future research opportunities** 
Location: Darwin, 7th Floor

### Ben Yarnall
**Leading on funding applications and stimulating your future research opportunities**
Location: Darwin, 7th Floor

#### Notes from workshop
Please use this space to add notes from this workshop

### Venexia Walker
**Producing analysis-ready datasets in TRE/SDE setting**
Location: Lacks2, 7th Floor

[**Presentation slides.**](https://docs.google.com/presentation/d/131jYLhFmzp7SOYotKY4LG_61e-poCQyF/edit?usp=sharing&ouid=117656171233849619714&rtpof=true&sd=true)

Stages of making an analysis ready data asset:

* Create a spine, recommended approach would be to take all those who are registered in a primary care in England - in establishing spine make sure there is not an overlap with the study cohort i.e. for studying patients with covid-19 if the vaccination data is a mandatory element on defining the spine the population might be biased by those who are not presented in vaccine dataset due to them being unvaccinated.
* Make an event list and decide what to do with recurrent events
* Be detailed about range of covariates, exposure and outcome variables in the study protocols
* Careful consideration of variable names: you can encode
* Agree on naming conventions
* Data quality checks and quality assurance on the curated table: people with a record post date of death, born on future dates etc.
* Sample size is less of a limit due to huge pool of national records
* The hospital dataset might hold people who are visitors or migrants and require emergency care, therefore a linkage to the primary care is a step to ensure people are actual residents of England or Wales.

#### Notes from workshop
Please use this space to add notes from this workshop

### Ruth Milne
**Telling your story: Communicating health data research**
Location: Franklin, 7th Floor

### Anna Woolman
**Telling your story: Communicating health data research**
Location: Franklin, 7th Floor

#### Notes from workshop
Please use this space to add notes from this workshop

# [14:45-16:00] Poster session

# [16:00-16:45] Technology showcase

## Branwen Snelling

### Talk Highlights
* Health metadata catalouge on HDR Gateway
* NLP on descriptive fields of metadata
* Gateway API -> term extraction -> medical vocabulary mapping -> impelementing the search engine
* New Entity Recognition (NER): Kaggle healthcare dataset
* Each record contained
* Elastic search is enabled on search engine: synonyms & stemming lamitisation
* Fuzzy matching : Matching terms with miss-spelling but more resource intensive to impelement e.g. "SONMED" will be matched to correct term "SNOMED"

## Erum Masood

### Talk Highlights
* FAIR data principles breaking data silos
* Data standardisation : OMOP CMD
* Will get in touch with Erum to provide QR codes for accessing her referenced research papers
* Open to collaboration
* non-clinical terms might not have a direct map to OMOP model
* Evaluation of usefuleness of OMOP transformation is key
* OMOP retains the source information
* Cohort discovery tool provides a way to facilitate federated analytics and a 95% of the overall cohort to the OMOP

## Fiona Chalmers

### Talk Highlights
* The CVD COVID UK, working across 3 national TREs
* Creating and sharing research project resources
* How do we define disease? a harmonised approach across research projects
* HDR UK Phenotype library: filled with diagnostic code lists
* There is a code list comparison tool and identify the codes used in common
* Harmonised definition of disease across projects

## Vote for the questions
go to: www.menti.com
Ender the code: 4442 2957

or scan this QR code


## Submit your own questions
Please use this space to add your own questions

# [16:45-17:20]Keynote Address

## Marta Milo

## Notes
* Determination at any stages of your careers
* Integrity and following your gut feeling
* Through the noise you see the gems and can shape the way forward
* Bravery is a disticnt character fueling the journey
* Make your voice heard: Rosaline Franklin a role model for computational biologists
* Team: teamwork, give generously
* Networks: an essential element, you have to give to the network and use it
* Mentors are essential: be impactful with questioning, rigor
* **Never stops questioning**
* **You need the skills to take the moment and make it yours**
* Answer the `so what?!` question: what is the impact of the work?
* What matters in the industrial environment is the pure impact of the science
* Connect your science to **Impact**
* Use curiousity and questioning
* Never Evenr comprimise in robustness of data that you stand 100% with

## Vote for the questions
go to: www.menti.com
Ender the code: 4442 2957

or scan this QR code


## Submit your own questions
Please use this space to add your own questions

# [17:20-17:30] Closing Remarks

**Get ready for poster prizes and group photo!**


# Keep in touch

* MailBox: [e-mail]
* Slack group: https://healthdataresearchuk.slack.com/archives/C07MGSM6VU7

# Contribute
Your testimony today will help shape our upcoming blog post, capturing the powerful moments and insights shared during this event. Your voice matters—let's make a meaningful impact together!

Please provide your contact details with your statement.

# End of the document