I have updated the simulation package with 2 VBA functions, one for simulating finite discrete variables, and another (Cholesky()) for generating correlated random variables. The Sampler macro and example spreadsheets have been updated to generate examples of both of these (including an example 3 correlated normally distributed variables, and a way to generate arbitrary correlated random variables). ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Trial Management using Time-to-Event datahttps://youtu.be/H4fwAC01iHE The presentation of my work is here: (Annual meeting of the American Statistical Association, August 2006). Modeling Multiphase Clinical Trials: Time to Completion, Study Management ===================================================
Simulation using Excel: A Clinical Study ApplicationThis (hypothetical) spreadsheet simulation evaluates the impacts on timeline and budget of various combinations of high, medium, and low quality sites, and compares 2 options for performing an interim analysis: either after a fixed number of patients, or after a fixed length of time. Since enrollment rates are uncertain, the former option is more of a challenge for planning timelines, as the time of the interim analysis will vary, while the later risks having too few patients in the interim analysis.
Simulated costs for multiple scenarios are compared against a conventional point estimate of costs for the study, showing a high risk of going over budget because of uncertainty in screen failure and enrollment rates.
The next update will incorporate the SIPmath™ Modeler Tools for Excel v3.0. These tools should simplify some aspects of writing a simulation in Excel, but more importantly allow a simulation to be written that uses the results of other simulations as part of it's simulated universes, and publish it's results for use by other simulations and reports. They are found at: (NB: these require later versions of MS Excel, and the code does not seem stable yet.)
≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣
Introducing Tolstoy Targets: Motivation, Principles and ExamplesUpdated slides for my pet graphics project, Tolstoy Targets, including motivation, basic principles, and examples.
Next update will include applications to risk based study management. https://sites.google.com/site/dennissweitzer/home/tolstoy-targets
* ~º -=> ---===< ><> ><;;> <;;> -;=;> -;=;< -;=;o >-|o >-|OModeling & Simulation in Managing Studies: Some Exampleshttp://www.slideshare.net/denswei/clinical-study-modeling-simulationhttp://www.slideshare.net/denswei/clinical-study-modeling-simulation
========================================= Randomization Metrics:Jointly assessing predictability and efficiency loss in covariate adaptive randomization designs A presentation given August 8, 2013 at the Joint Statistical Meetings in MontrealRandomization methods generally are designed to be both unpredictable and balanced between treatment allocations overall and within strata. However, when planning studies, little consideration is given to measuring these characteristics, nor are they examined jointly, and published comparisons between methods often are not useful. In order to compare randomization performance, we simulated various covariate-adjusted randomization methods , and compared efficiency & unpredictability graphically and statistically using proposed metrics. Publication:
Slide set: * ~º -=> ---===< ><> ><;;> <;;> -;=;> -;=;< -;=;o >-|o >-|ORandomization: Too important to gamble withRandomization is a central part of Randomized Controlled Trials (RCTs) which are considered the 'gold standard' of clinical evidence. This talk is a basic introduction to the topic of randomization: It outlines how randomization is key to reducing various kinds of bias in a study; Explains several common randomization techniques; Proposes measurements of predictability and imbalance of a randomization method; And compares the performance of the methods in a series of simulations.
Summarized as a poster for the Society for Clinical Trials Annual meeting, May, 2013Presented to the Delaware ASA monthly meeting in October, 2012* ~º -=> ---===< ><> ><;;> <;;> -;=;> -;=;< -;=;o >-|o >-|OAnimated explanation of Tolstoy Targets: This simple & animated explanation starts from the very basics & includes examples from project management and chemical screening.
* ~º -=> ---===< ><> ><;;> <;;> -;=;> -;=;< -;=;o >-|o >-|OResearch Study in Pancreatic CancerTHE EFFECTS OF PREDNISONE, MONTELUKAST, CETIRIZINE, RANITIDINE, POTASSIUM PHOSPHATE, MAGNESIUM, AND VERAPAMIL ON PANCREATIC ADENOCARCINOMA STAGE IV.
..is approved and ready to enroll. This is a small proof of concept study to examine the benefits of this combination therapy in managing late stage pancreatic cancer symptoms. While there is no comparator group, I plan to do a comparison between enrolled patients and patients from the SEER database (a cancer registry) matched on baseline characteristics, and including year of diagnosis as a covariate to adjust for improvements in treatment over time. (NB: Only descriptive statistics will be compared).
For more details, see: http://www.nashedspa.com/index.html
=========================================January 2012 update:
Outline•How
to do Simulation in Excel
•Notes
on using Inverse Probability Functions
(For Verification, Validation, and Sensitivity)
• Some tools: Excel Macros, Visual Basic Functions
• Examples from Clinical Trials
•Probability
Management in SIPS, SLURPS, & DIST
Slides and spreadsheets at:
≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣
My resume summarized as a word cloud:
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Answer (mine):
Variability
Standard Deviation Shows
Variance Calculates
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
These 4 graphs plot project outcomes (for instance, Adverse Events, Efficacy goals, lab findings, schedule & budget). Which projects are on target? Which are in trouble? These
are splatter plots, a graphic I created to summarize outcomes of clinical studies by multiple criteria. However, it is suitable for any
project with multiple objectives, each of which can be categorized as Good/Uncertain/Bad on a scale of 0 to 10. The defining metaphors are a threefold theme of Targets, Traffic
Lights, and Tolstoy. ∫∬∭∮∯∰∱∲∫∬∭∮∯∰∱∲∫∬∭∮∯∰∱∲∫∬∫∬∭∮∯∰∱∲∫∬∭∮∯∰∱∲∭∮∯∰∱∲∫∬∭∮∯∰∱∲∫∬∭∮∲∫∬∭∮∯∰∱∲
##
Excel
is a general purpose spreadsheet which is widely used & understood,
but rarely used by itself for simulations. Industrial grade simulation
in Excel requires add-in packages, while "casual" simulations are
highly limited in scope (and could be described as "toy" simulations).Quick & Simple Simulation using MS ExcelHowever, the Data Table function in MS Excel can be used to execute substantial simulations, without requiring cumbersome programming "tricks" or VBA coding to store iterated results. The result is an arbitrarily large results table in which each row is one iteration of the simulation, and each column is a random variable generated in the simulation. A small number of additional probability functions are easily programmed using VBA to make Excel a general purpose simulation package. Because VBA is interpreted, use of VBA functions can greatly limit the speed of a simulation. However, for simulations of modest size and complexity, the ease and familiarity of working in Excel, outweigh the disadvantages of speed. Presented to the Delaware ASA monthly meeting in October, 2011is: (1) A power point tutorial on implementing basic simulations in Excel (2) An Excel spreadsheet template containing macros which ease writing simulations, some missing probability functions written in VBA, and examples. (3) An Excel spreadsheet examples of specialized topics. (3a) Project management examples of using simulation in a CPM (Critical Path Method) context. Quick & Simple Simulation using MS Excel
≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣
This is a tongue-in-cheek statistics talk that I gave sometime around the 2004 election cycle in the Bush vs Kerry match-up. In it, I outlined some quirky models for predicting election results, but also the approach used by Nate Silver of simulating state outcomes from models of polling data to predict the electoral college results. (This was long before Nate Silver was famous though).
Election Forecasting, 2004 * ~º -=> ---===< ><> ><;;> <;;> -;=;> -;=;< -;=;o >-|o >-|O
Sign of the Timings: Predicting Time of Completion in Multiphase Survival Clinical TrialsStudying maintenance of clinical effect typically requires clinical response for a minimum amount of time on treatment before randomization. If randomized, patients are then followed until treatment failure or withdrawal, and the trial halted after a pre-specified number of events. For ethical and cost reasons it is desirable to minimize the number of patients enrolled and randomized, and to predict the time of the last event under multiple scenarios. We describe a data-driven stochastic simulation for two such trials in which: Each phase is modeled as a competing event process; Distributions of event times are derived from Kaplan-Meier survival curves from available data; Parameter uncertainty is modeled based on K-M survival estimates; Withdrawals and events occur at similar overall rates, though at different times; Predictions are updated as information is accrued. Presented to: the Delaware ASA monthly meeting in September, 2006; Annual meeting of the American Statistical Association, August 2006;Modeling Multiphase Clinical Trials: Time to Completion, Study Management +++++++++++++++++++++++++++++++++++++++++++++Quick and Dirty Simulation in Excel I have often needed to do simple, flexible and quick ad hoc
simulations while designing clinical trials (and other things). Excel
provides almost everything one needs to create an iterated simulation,
missing only a few common random variables, a way to repeat the
calculations using random variables, and a way to collect the results.
Commercial add-in products provide these features using custom
programming, however I found one simple programming "trick" using
iterated circular references to provide key functionality for iterated
simulations.Presented to the Delaware ASA monthly meeting in February 2010. |