Clinical practice is the direct treatment of the individual
patient. Clinical science is the study of how
to do clinical practice. Clinical research consists of
systematic contributions to clinical science.
From these three definitions follows a method of doing clinical research
that delivers results specific to the individual patient. The fundamental
idea is to extract from a very large electronic medical record system
a small collection of previous patients who are very similar to the
patient at hand. The experiences of those patients are used to compute
the probabilities of benefit and harm for any therapy for which there
is sufficient information. The physician explains to the patient
how he or she can use these specific results to make a decision.
The technology of patient-specific research
requires comprehensive knowledge of the medical record system, the
ability to extract event-stream datasets and to manipulate them for
research purposes, and the construction of a network of algorithms
for computing results for the individual patient.
Conventional medical research does not provide patient-specific
information, and also does not provide probabilities of benefit and
harm.
All of the necessary steps are described in
Patient-Specific Research by Mikel Aickin. Click on
The Book to read the
first section.
It is easy to describe how patient-specific information might
be delivered to the visit room, but it is difficult to create
and run such a patient-specific system. This book describes in
detail the technical issues of going from electronic medical
records to an evidence-based report tailored to the individual
patient. It also explains the theoretical issues involved in
using non-intervention data for research purposes.
Patient-Specific Research - Contents and First Chapter
exclusively at Amazon
The Patient's Question-Unanswered, by Mikel Aickin
What is the one question that every patient should want to ask, but which
none do? Why does no one have the answer to this question?
From Medical Records to Clinical Science, by Mikel Aickin & Charles Elder
Research directly from medical records is generally disparaged. This article attempts
to provide an explanation for researchers about the rationale, methods, and problems
with such research. It is a simplified introduction to a large and dense methodology literature.
Analysis of Nonintervention Studies, by Mikel Aickin
This is a much more technical handling of the issues discussed in the preceding article.
Complete Propensity Scores, by Mikel Aickin (unpublished)
Conventional propensity scores do not solve the problems of confounding in non-intervention
studies. This article suggests augmenting the treatment-propensity score with outcome-propensity
scores. Taken together they do succeed in removing the unwanted
effects of common causes.
Critique of Conventional Research Methodology, by John Heron
Now 30 years old, this article put forward a clear argument for how standard medical
research did not (and does not) serve the interests of the individual patient.
Perils of Evidence-Based Medicine, by Ben A. Williams
Elaboration on the fallacy of division, applying statistical results from collections
of patient as if they applied to individual patients.
Group Analysis versus Individual Response: The Inferential Limits of Randomized
Controlled Trials, by John M. Kelley & Ted J. Kaptchuk
Points out the variability in individual patient outcomes even when group outcomes are clear.
The authors interpret this to mean that it is difficult to identify "responders", by which they
mean patients who responded to the specific treatament effect, not for other reasons. They almost
take the next step, to observe that what was needed was more information on each individual
patient.
Bridging Psychology and Biology, by Stephen M. Kosslyn, John T. Cacioppo, Richard J. Davidson,
Kenneth Hugdahl, William R. Loveallo, David. Spiegel, & Robert Rose.
Draws lessons for psychological therapy from the "obvious" fact that people have variable
biological responses to things that happen to them. Argues that therapy should be driven
by the search for an appreciation of individual differences.
Title
These downloadable programs run in R, a free statistics and data management program.
Download the ES programs as textfiles. Each file contains one program
One of the programs loads all the others. All you have to do is source("Eload.r")
to make the programs available in R. The advantage is that you can
modify them.
Download the ES programs as an R package. You
only need to do require("ES") to make the programs available.
The advantage is convenience, and the disadvantage is that
you cannot modify them.