Gary Collins

Job losses at research-intensive universities double in two years - Research Professional News Exclusive: Scale of redundancies revealed by RPN branded a “disaster” for UK research capacity

Redundancies at research-intensive universities double in two years.

www.researchprofessionalnews.com/rr-news-uk-u...

when your dataset is definitely real

Evidence of Unreliable Data and Poor Data Provenance in Clinical Prediction Model Research and Clinical Practice Clinical prediction models are often created using large routinely collected datasets. It is essential that prediction models are developed with appropriate data and methods and transparently reported...

Much of my work in meta-research is on finding problems in research, so I've seen a lot of bad practices. However, even I was shocked by hundreds of researchers publishing papers using data that is faked and has no data provenance. www.medrxiv.org/content/10.6.... Amazing work by my student Alex.

Job losses at research-intensive universities double in two years - Research Professional News Exclusive: Scale of redundancies revealed branded a “disaster” for UK research capacity

Job losses at UK research-intensive universities double in two years.

Exclusive: Scale of redundancies branded a “disaster”.

www.researchprofessionalnews.com/rr-news-uk-u...

We are setting out to develop some new recommendations (TRIPOD-CODE) to provide guidance on reporting the availability and structure of code for predictive AI healthcare tools

Watch this space, and read the protocol here

link.springer.com/article/10.1...

#transparency #code #reproducibility

NEW #openaccess paper in @bmj.com led by the brilliant @matthewluney.bsky.social "Effectiveness of drug interventions to prevent delirium after surgery for older adults: systematic review and network meta-analysis of RCTs"

--> www.bmj.com/content/392/...

#NIHR #surgery #delirium

NEW #openaccess PAPER in BMJ Digital Health & AI

"Public perceptions of health data sharing for artificial intelligence research: a qualitative focus group study in the UK"

--> bmjdigitalhealth.bmj.com/content/2/1/...

#digitalhealth #AI #publicperceptions #datasharing

Two openings for PhD candidates:

1️⃣ Use causal inference methods for early evaluation of the downstream effect of algorithms on patient outcomes. 👉 www.lumc.nl/en/about-lum...

2️⃣ Develop methods for evaluating patient predictions under different treatment options. 👉 www.lumc.nl/en/about-lum...

Interested in learning about IPD meta-analysis projects? Join me for a gentle introduction to the topic, in the seminar below (Wednesday Jan 28th 12 to 1pm GMT)

www.ticketsource.co.uk/arcyorkshire...

Adherence to TRIPOD+AI guideline: An updated reporting assessment tool - Journal of Clinical Epidemiology www.jclinepi.com/article/S089...

Delighted to share our new Perspective, published in the inaugural issue of @natureportfolio.nature.com 𝗡𝗮𝘁𝘂𝗿𝗲 𝗛𝗲𝗮𝗹𝘁𝗵. We shared opportunities and challenges of using large language models (LLMs) in global health.

www.nature.com/articles/s44...

#AI #LLM #Global #Health #DukeNUS

Large language models in global health - Nature Health Large language models (LLMs) are emerging as powerful tools in healthcare, with a growing role in global health, particularly in low- and middle-income countries. This Perspective examines the current...

Large language models in global health.

Perspective from Nan Liu and colleagues

#AI #healthAI #LLM
www.nature.com/articles/s44...

I launched version 3.0 of my browser extension "Lazy Scholar", a free in-browser research assistant. It opens automatically when you load an academic article.

See: lazyscholar.org/2026/01/10/l...

Some people bring up (1) the cost of criticism and (2) that a lot of criticism has already been voiced but ignored. Both points are valid, so here are some suggestion for (1) reducing backlash and (2) increasing impact (from this talk of mine: juliarohrer.com/wp-content/u...

Doug Altman - an eternal inspiration for all medical statisticians and non-statistician medical researchers. #StatsSky #Statistics

If you’re working on clinical prediction models, whether regression or machine learning methods then TRIPOD+AI is for you

Better reporting → better science → safer, more trustworthy clinical AI.

Read the full guidance here
--> www.bmj.com/content/385/...

Who should be using TRIPOD+AI?

- Researchers developing or evaluating prediction models
- Clinical AI teams
- Journal editors + peer reviewers
- Regulators + guideline developers
- Anyone aiming to improve transparency in healthcare AI

If you publish prediction model research, this is you!

What TRIPOD+AI aims to do:

- Improve clarity and completeness of reporting
- Support replication and critical appraisal
- Reduce research waste
- Strengthen trust among clinicians, regulators, and patients

It’s about making prediction model research usable.

Why does TRIPOD+AI matter? Because prediction model studies are still plagued by:

- Poor reporting
- Missing methodological detail
- Unclear model specifications
- High risk of bias

Transparent reporting isn’t optional, it’s underpins trustworthy clinical AI.

End of year reminder: the TRIPOD+AI reporting guideline is reporting standard for all clinical prediction model studies, including those using machine learning and AI.

--> www.bmj.com/content/385/...

"The making of a statistician: Doug Altman" - just published in @bmj.com celebrating his 1 million citations and reflecting on his remarkable career and legacy. One of the most influential statisticians in modern medical research.

--> www.bmj.com/content/391/...

#BMJChristmas #methodologymatters

Sequential sample size calculations and learning curves safeguard the robust development of a clinical prediction model for individuals - Journal of Clinical Epidemiology www.jclinepi.com/article/S089...

Clinical prediction models using machine learning in oncology: challenges and recommendations Clinical prediction models are widely developed in the field of oncology, providing individualised risk estimates to aid diagnosis and prognosis. Machine learning methods are increasingly being used t...

ICYMI: "Clinical prediction models using machine learning in oncology: challenges and recommendations"

--> bmjoncology.bmj.com/content/4/1/...

#machinelearning #digitalhealth #predictionmodels #methodsmatter

Point of no returns: researchers are crossing a threshold in the fight for funding With so little money to go round, the costs of competing for grants can exceed what the grants are worth. When that happens, nobody wins.

"How can funders avoid crossing the Szilard point?"

The Szilard point is "the threshold at which the total cost of competing for a grant equals (or surpasses) the value of the available funding."

Doug Altman was an internationally renowned statistician who served as The BMJ’s chief statistical adviser.

Read about life and work that made this statistician a "citation millionaire"
#BMJChristmas
www.bmj.com/content/391/...

ICYMI: NEW PAPER "Evaluation of performance measures in predictive artificial intelligence models to support medical decisions: overview and guidance"

--> doi.org/10.1016/j.la...

#AI #Machinelearning #predictiveAI

NEW PAPER: A decomposition of Fisher’s information to inform sample size for developing or updating fair and precise clinical prediction models - part 2: time-to-event outcomes

* Implemented via pmstabilityss module, facilitates models with precise & fair individual-level predictions
rdcu.be/eUVam

Evaluation of performance measures in predictive artificial intelligence models to support medical decisions: overview and guidance Numerous measures have been proposed to illustrate the performance of predictive artificial intelligence (AI) models. Selecting appropriate performance measures is essential for predictive AI models i...

Our guidance regarding performance measures for medical AI models is finally out!

- Stop bashing AUROC, although it does not settle things
- Calibration and clinical utility are key
- Show risk distributions
- Classification statistics (e.g. F1) are improper

www.thelancet.com/journals/lan...

Nice to finally meet you in person this evening Peter.

Machine learning has developed remarkable new ideas about how to develop prediction algorithms. But always baffled me why the field had to reinvent how to evaluate models. Here we show F score should not be used to evaluate medical prediction models link.springer.com/epdf/10.1186...