You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
111 lines
5.2 KiB
TeX
111 lines
5.2 KiB
TeX
\documentclass[../Main.tex]{subfiles}
|
|
\graphicspath{{\subfix{Assets/img/}}}
|
|
|
|
\begin{document}
|
|
|
|
As noted above, there are various issues with the analysis as completed so far.
|
|
Below I discuss various issues and ways to address them that I believe
|
|
will improve the analysis.
|
|
|
|
\subsection{Increasing number of observations}
|
|
|
|
The most important step is to increase the number of observations available,
|
|
specifically the number of trials matched to ICD-10 codes with corresponding
|
|
population estimates in the Global Burden of Disease Dataset.
|
|
Improvements in Large Language Models may make this data more accessible, or
|
|
the data may be available in a commercial dataset.
|
|
|
|
|
|
\subsection{Enrollment Modelling}
|
|
|
|
One of the original goals of this project was to examine the impact that
|
|
enrollment struggles have on the probability of trial termination.
|
|
Unfortunately, this requires a model of clinical trial enrollment, and this
|
|
data is missing from my dataset.
|
|
In most cases the trial sponsor reports the anticipated enrollment value
|
|
while the trial is still recruiting and only updates the actual enrollment
|
|
after the trial has ended.
|
|
Some trials do publish an incremental record of their enrollment numbers,
|
|
but this is not the norm.
|
|
It may be possible to impute the enrollment process if a suitible model
|
|
can be created.
|
|
% Due to the bayesian model used, this would be easy to incorporate
|
|
% \cite{mcelreath_statisticalrethinkingbayesian_2020}.
|
|
|
|
There has been substantial work on forecasting
|
|
multi-site enrollment rates and durations by
|
|
\cite{
|
|
tozzi_predictingaccrualrate_1996,
|
|
carter_applicationstochasticprocesses_2004,
|
|
anisimov_modellingpredictionadaptive_2007,
|
|
zhang_stochasticmodelingprediction_2010,
|
|
zhang_jointmonitoringprediction_2012,
|
|
zhang_modelingpredictionsubject_2012,
|
|
heitjan_realtimepredictionclinical_2015,
|
|
jiang_modelingvalidatingbayesian_2015,
|
|
deng_bayesianmodelingprediction_2017,
|
|
lan_statisticalmodelingprediction_2019,
|
|
zhang_simplerobustmodel_2022,
|
|
urbas_interimrecruitmentprediction_2022,
|
|
bieganek_predictionclinicaltrial_2022,
|
|
avalos-pacheco_validationpredictiveanalyses_2023,
|
|
}
|
|
but choosing between the various single and multi-site models presented is
|
|
difficult without a dataset with which to validate the results.
|
|
% In addition to needing a well calibrated model, I would require more trials,
|
|
% specifically those that report their enrollment incrementally so
|
|
% that there is data on what happens when enrollment is slower than anticipated.
|
|
% It may also be possible to estimate the probability that enrollment goals
|
|
% have been met if data can be extracted that details planned observation times.
|
|
% Of course, this is speculative at this point.
|
|
%FIXTAG: Avoid speculation here.
|
|
|
|
% \subsection{Improving Population Estimates}
|
|
%
|
|
% The Global Burden of Disease dataset contains the best estimates of disease
|
|
% population sizes that I have found so far.
|
|
% Unfortunately, for some conditions it can be relatively imprecise due to
|
|
% its focus on providing data geared towards public health policy.
|
|
% For example, GBD contains categories for both
|
|
% drug resistant and drug suceptible tuberculosis, but maps those to the same
|
|
% ICD-10 code.
|
|
% In contrast, there is no category for non-age related macular degeneration.
|
|
% Thus not every trial has a good match with the estimate of the population of
|
|
% interest.
|
|
% Finding a way to focus on trials that have good disease population estimates
|
|
% would improve the efficiency of the analysis.
|
|
% %FIXTAG: What am I trying to say here. IHME is among the best data sources.
|
|
% % How do I propose getting other data? Should probably just remove this.
|
|
|
|
\subsection{Improving Measures of Market Conditions}
|
|
|
|
% Deficiency: cannot measure effect of market conditions because of endogenetiy of population and market conditions (fatal diseases)
|
|
|
|
In addition to the fact that many diseases may be treated by non-pharmaceutical
|
|
means (e.g. diet, physical therapy, medical devices, etc),
|
|
off-label prescription of pharmaceuticals is legal at the federal level
|
|
\cite{commissioner_understandingunapproveduse_2019}.
|
|
%FIXTAG: Discuss how there isn't much data about off label prescription (I have a source)
|
|
These two facts both complicate measuring competing treatments,
|
|
a key part of market conditions.
|
|
One way to address non-pharmaceutical treatments is to concentrate on domains
|
|
that are primarily treated by pharmaceuticals.
|
|
Another way to address this would be to focus the analysis on just a few specific
|
|
diseases, for which a history of treatment options can be compiled.
|
|
%FIXTAG: Get rid of 'another', doesn't match context
|
|
This second approach may also allow the researcher to distinguish the direction
|
|
of causality between population size and number of drugs on the market;
|
|
%FIXTAG: join better to prior sentence
|
|
for example, drugs to treat a chronic, non-fatal disease will probably not
|
|
affect the market size much in the short to medium term.
|
|
This would require identifying diseases that are prime candidates and then
|
|
trials and drugs associated with those diseases.
|
|
% This allows the effect of market conditions to be isolated from
|
|
% the effects of the population.
|
|
% %FIXTAG: I am already proposing these as fixes
|
|
|
|
% Alternative approaches
|
|
% - diseases with constant kill rates? population effect should be relatively constant?
|
|
|
|
\end{document}
|