- 14
- 35 259
Richard_D_Riley
เข้าร่วมเมื่อ 22 ก.ย. 2020
Videos about prediction models, prognosis research, IPD meta-analysis, and other methods for healthcare research
How to get your article rejected by the BMJ: 12 common statistical issues
Medical research is full of statistical issues and pitfalls, and many articles are rejected due to statistical concerns. As a Prof of Biostatistics & also a Statistics Editor for the BMJ, I see this every day. Here, I outline 12 common problems in the hope that flagging these will raise awareness and improve research quality in the future. Intended for a broad audience, the talk is based on our BMJ article "On the 12th day of Christmas, a statistician sent to me" published in 2022 (www.bmj.com/content/379/bmj-2022-072883) - for a one-page summary see image here Richard_D_Riley/status/1612396496443301889
Enjoy! Hope this helps you in your research.
Enjoy! Hope this helps you in your research.
มุมมอง: 3 643
วีดีโอ
Key Steps and Common Pitfalls in Clinical Prediction Model Research
มุมมอง 3.1Kปีที่แล้ว
Clinical prediction models estimate an individual’s risk of a particular health outcome. Thousands of prediction models are published each year, yet few are reliable or fit for purpose. In this talk, I outline key steps and common pitfalls in prediction model research, and outline ways to produce more reliable and clinically useful models - including protocols, better handling of continuous pre...
Power Calculations for Individual Participant Data (IPD) Meta-Analysis Projects
มุมมอง 3322 ปีที่แล้ว
Individual participant data (IPD) meta-analysis projects herald much promise (www.ipdma.co.uk), but are potentially time-consuming, often taking upwards of two years to engage with trial investigators; obtain, clean, harmonise and meta-analyse the IPD; and publish and disseminate results. Therefore, before embarking on an IPD meta-analysis project, researchers and funders need to be reassured t...
Sample size calculations for external validation of a clinical prediction model
มุมมอง 2.2K2 ปีที่แล้ว
In prediction model research, external validation is needed to examine an existing model’s performance using data independent to that used for model development. Current external validation studies often suffer from small sample sizes and consequently imprecise model performance estimates. In this talk, I propose how to determine the sample size needed for a new external validation study, where...
Stability of Clinical Prediction Models Developed Using Statistical or Machine Learning Approaches
มุมมอง 2.3K2 ปีที่แล้ว
Clinical prediction models are a hot topic. They use statistical or machine learning methods to estimate an individual’s risk of a particular health outcome, conditional on their values of multiple predictors (features). In this talk, I present research undertaken (paper forthcoming) with Prof Gary Collins (University of Oxford), to raise the concern that many models are developed using small d...
Pitfalls for Clinical Decision Support Based on Artificial Intelligence (AI)
มุมมอง 9613 ปีที่แล้ว
In this 26 minute keynote talk presented at the World Congress of Ultrasound in Obstetrics and Gynecology (16 Oct 2021), Prof Ben Van Calster discusses 5 central topics when examining the use of AI to inform clinical decisions. These topics cover the importance of methodology, the need for evidence of performance, expected heterogeneity in performance, model and data availability, and the diffi...
Sample size calculations for clinical prediction model research (aka "goodbye rules of thumb")
มุมมอง 4.9K3 ปีที่แล้ว
Intended for a broad audience, this short talk describes how to calculate the sample size required for developing or validating prediction models in healthcare (also known as prognostic models or predictive algorithms). Currently many researchers use 'rules of thumb' such as 10 events per variable - but here I propose more tailored approaches and illustrate them using applied examples. Referenc...
An introduction to risk prediction and prognostic models
มุมมอง 9K3 ปีที่แล้ว
This talk provides a gentle introduction to risk prediction and prognostic models for healthcare research. They are introduced in the context of the PROGRESS framework, with examples given of their role, impact, and statistical basis. Phases of prediction model research are outlined, and current problems and limitations discussed. Signposts are then provided for better practice, including key a...
Individual Participant Data (IPD) Meta-Analysis: introduction, rationale, & key steps
มุมมอง 3K3 ปีที่แล้ว
Intended for a broad audience, this talk provides an introduction to IPD meta-analysis projects for healthcare research. It discusses their rationale, advantages and key steps, and provides various examples including the examination of treatment effects, treatment-covariate interactions, test accuracy, prognostic factors and risk prediction models. For further details please see our website (ww...
Individual Participant Data (IPD) meta-analysis: advanced topics & recent debates
มุมมอง 1.2K3 ปีที่แล้ว
Recorded for the Japanese region of the International Biometrics Society. In this talk, I discuss pressing statistical issues for IPD meta-analysis of randomised trials, including: - estimation of treatment-covariate interactions - differences between one-stage and two-stage approaches - coding of the treatment variable to improve estimation of one-stage models
Development of prediction models using competing risk models in big healthcare databases
มุมมอง 7274 ปีที่แล้ว
Lucinda Archer from Keele University discusses issues in the development of a prediction model in an applied example with competing risks and big data. Recorded for the MEMTAB 2020 conference.
Minimum sample size for external validation of a clinical prediction model with a continuous outcome
มุมมอง 6014 ปีที่แล้ว
Talk by Lucinda Archer at Keele University, recorded for the MEMTAB 2020 conference, which introduces her work on sample size requirements for externally validating a clinical prediction model with a continuous outcome (e.g. blood pressure, fat mass). The corresponding paper is published here: onlinelibrary.wiley.com/doi/10.1002/sim.8766
Individual participant data (IPD) meta-analysis to examine treatment-covariate interactions
มุมมอง 8574 ปีที่แล้ว
In this 10 minute talk (recorded for the MEMTAB 2020 conference www.epi-centre.be/memtab2020) I discuss statistical recommendations for conduct and planning of IPD meta-analyses that examine to examine treatment effect modifiers (treatment covariate interactions). In particular, I identify common pitfalls and problems to avoid.
PROGRESS in Prognosis Research
มุมมอง 2.3K4 ปีที่แล้ว
An introduction to prognosis research and the PROGRESS framework, covering the four key topics of (i) overall prognosis, (ii) prognostic factors, (iii) prognostic models, and (iv) predictors of treatment effect. For more details visit: www.prognosisresearch.com
Thank you Prof. Riley. I have just read your papers, and then found you on TH-cam! It is really helpful!
two things come to mind , one have you checked what is the distribution of your target outcome(s) that best fit using basic metrics , secondly are you using symptoms from medical records as your X,s Or have you gone in to deeper (genetic-level) features to better predict your conditional distribution ?
Excelente video, lo compartiré
do you think we have to use random slopes and random intercept toget good individual predicted risk probabilities
thanks for share, great presentation!
Thank you for this "Christmas delight". A "must listen" for all researchers. Happy New Year 2024!🎉
Great video
very useful
Thanks a lot!
Thank you so much
Interesting
Thank you!
Very informative, thank you
Awesome... Thanks a lot
So, does that mean we have to do clinical trial especially Randomized Control Trial (RCT) to prove AI in Medicine efficacy?
Great talk. Kindly upload more videos
thank you for your nice explanation; I have a question. i need to do ipd meta analysis for 3 datasets to see association of a continuous variable with a binary outcome(hypertension). I don't have a treatment group. studies are cohort obserational study and the variable is subject AHI (index of apnea-hypopnea); would you please let me know how to do one stage MA in R?
Very interesting! Is there code in R or Python available for this?
Excellent!
Thank you!
Great!!!!
Thanks for this
well explanation, very usefull! thanks!
This is fascinating. Do you know much about the development and validity of the Dawes Redman criteria for automatic analysis of cardiotocographs??
Why is causation not important in prediction modelling? Wouldn't the inclusion of potentially spurious factors be problematic when trying to apply the model in other samples?
that is why you need external validation for your models. To determine the causality of a factor, you need to know the causal model, which we do not many times. Check Miguel Hernan's lectures on the topic.
@@jekamito Isn't that just implicitly saying we do need to know causal factors, whether they are obtained from external validation or causal modelling.
Great work good sir
Wonderful presentation Sir. thanks for sharing!!
GREAT LECTURE THANK YOU
Fantastic lecture
Very helpful information
thanks for share, great presentation!