Gary Collins 🇪🇺
@GSCollins
Prof of Med Stats @UniofOxford | Director @CSMOxford & UK @EQUATORNetwork | @TRIPODStatement +AI | BMJ Stats Ed | 🚴
ID:299115067
https://scholar.google.co.uk/citations?user=cVKF81gAAAAJ 15-05-2011 14:48:36
5,8K Tweets
10,4K Followers
995 Following
Excited that our new #openaccess #systematicreview has just been published npj Digital Medicine looking at diagnostic accuracy of #AI in #digitalpathology 💻 🔬
rdcu.be/dGTOD
We pleased to our announce our new partnership between the Centre for Open Science (Center for Open Science) and the EQUATOR Network to improve transparency and quality of reporting of research.
equator-network.org/2024/05/02/new…
cos.io/blog/cos-and-t…
ICYMI: International consensus based recommendations for #MachineLearning studies in #healthcare (TRIPOD+AI) in The BMJ
bmj.com/content/385/bm…
Extended guidance
bmj.com/content/suppl/…
#ArtificialIntelligence #AI #statsX #scicomm #Stats #statstwitter #PleaseShare
🔥🔥🔥 New guidelines for reporting prediction models including AI models
⭐️TRIPOD-AI⭐️
Check out 👇 AASLD INASL EASLnews EASL Education BASL education feed APASL News Mamatha Bhat Sumeet Asrani Akash Roy Arun Valsan Gary Collins 🇪🇺
bmj.com/content/bmj/38…
ICYMI. Our latest #openaccess recommendations for reporting #AI / #machinelearning prediction model studies in healthcare are out in the The BMJ
bmj.com/content/385/bm…
Don't forget there is further brief guidance for each recommendation in the supplementary material
Some red flags when reading a prediction model paper
- no sample size calc
- no mention of missing data
- dichotomania
- any mention of class imbalance/smote/classifier
- no assessment of calibration
- AUC > 0.95
- no internal validation
- no model or link to code
#statstwitter
.Brennan Kahan is publishing some really helpful papers on the estimands approach eg he and others wrote a very helpful explainer on the E9 addendum. Follow Brennan to stay up to date with developments 8/8
Addendum explainer:
bmj.com/content/384/bm…
A word we hear a lot about currently in the trials world is #Estimands . But what are they and why are they useful? 1/8
#MethodologyMonday #85
Do you work in #ClinicalTrials ? Estimands could help make sure your study answers the right research question.
Our latest Trial Talk podcast episode features Brennan Kahan discussing his work on the estimands framework.
🎧Tune in to the full episode👉 bit.ly/3xPnYaS
We looked at dynamic random forest prediction models for central line-associated bloodstream infections on EHR from UZ Leuven.
We compared binary, multiclass, time-to-event, and competing risk outcome definitions. With elena albu Laure Wynants @laurewynants.bsky.social
arxiv.org/abs/2404.16127
Pre-conference courses on Tuesday 16th
An Introduction to Diagnostic Tests and Studies of Diagnostic Tests - Jon Deeks FMedSci, Alice Sitch, and Yemisi Takwoingi
An Introduction to Risk Prediction Models & Sample Size Calculations for Development & Validation Richard Riley (R²)
TRIPOD+AI is a new reporting guideline led by @UniofOxord researcher Gary Collins 🇪🇺 for #artificialintelligence studies developing or validating prediction models in healthcare powered by #machinelearning methods
bmj.com/content/385/bm…
#OxfordAI #ethicalAI