RSS Highlands Local Group: Why most prediction models cannot support treatment decisions and AGM

Date: Tuesday 07 December 2021, 3.00PM - 5.00PM
Location: Online
Online - joining instructions will be emailed to those registered
Local Group Meeting
Book now


Share this event

Currently in the development of prediction models ad hoc approaches are used for patients who start a treatment. Patients may for instance be excluded or censored or the treatments are ignored. Consequences of such analysis choices on the interpretation of the calculated risks are typically neglected which may lead to wrongly interpreted risks. In particular, risks may not provide the input to treatment decisions they are expected to give. In this talk, Nan will propose  'predictimand' framework to deal with treatment in clinical prediction models.  Analogous to the ICH estimand guideline for clinical trials, the ‘predictimand' framework is comprised of different questions that may be of interest when predicting risk in relation to treatment started after baseline. Nan will provide definitions of the estimands matching these questions, giving examples of settings in which each is useful.
 
This event will be followed by the local group AGM

Speaker
: Nan van Geloven, Assistant Professor, Leiden University Medical Center.
Time: 3 - 3.55pm
Title: Why most prediction models cannot support treatment decisions

Abstract: Currently in the development of prediction models ad hoc approaches are used for patients who start a treatment. Patients may for instance be excluded or censored or the treatments are ignored. Consequences of such analysis choices on the interpretation of the calculated risks are typically neglected which may lead to wrongly interpreted risks. In particular, risks may not provide the input to treatment decisions they are expected to give.
We propose that the way treatment is dealt with in clinical prediction models should be guided by a clearly defined prediction estimand. Analogous to the ICH estimand guideline for clinical trials, we propose a ‘predictimand' framework of different questions that may be of interest when predicting risk in relation to treatment started after baseline. We provide definitions of the estimands matching these questions, giving examples of settings in which each is useful.
We illustrate the impact of the predictimand choice in a dataset of patients with end-stage kidney disease where part of the patients receive a kidney transplantation during follow up and in published prediction models for Covid-19 patients.

4.05pm - 5pm RSS Highlands Local Group AGM
 
 
Nan van Geloven, Assistant Professor, Biostatistics, Leiden University Medical Center.
 
David McLernon for RSS Highlands Local Group   
 
Book now