While I am not involved in Precision Medicine (PM) in any way, I would like to ask two questions – one statistical and one syntactical.
The one-page summary states “Statistics is the science of learning from data, and accounting for the relevant uncertainties”. That science is mentioned is good – this is an essential and distinctive aspect of statistics. My concern is that the statement as written is overly general, as there are other means of learning from data that are not statistics. For example, reading anecdotes learns from data but is not statistical. We can learn from a single example – I only need to be run down by a street car once to be advised against it - but this is not a statistical method; neither is one desirable. It may be useful further state that statistics is a mathematical science involved in learning from a body of data. While we are well aware that "data" is exclusively plural, this understanding may not be widespread outside the scientific community.
My second question relates to the sentence “PM complements traditional precision medicine implemented by physicians at a bedside using their knowledge and training with data and evidence-based strategies.” This appears to contrast PM with PM. I expect this is quite clear to persons familiar with PM – I am quite unfamiliar with it. Yet, the tone of the document leads me to believe it is written for a learned but general audience. How is PM in this context distinct from “traditional precision medicine”?
Thank you!
Original Message:
Sent: 05-28-2015 08:07
From: Michael Mout
Subject: Obama's Precision Medicine Initiative
Just a quick note to remind non-practitioners of the danger of significance testing in the prescence of large samples, as may be he case when looking at results from large medical studies. Quite often we will see very significant results (p<.0001) when the actual effects are negligible. This is particullarly true when a for profit company or a entity with a specific point to make is trying to proof their point. We should always look not only at the statistical significance, but the medical or business significance. This is particularly true in end of life care where a particular procedure may extend life by a few days, or weeks at a tremendous cost.
As they said in Watergate, always follow the money.
------------------------------
Michael Mout
MIKS
------------------------------
Original Message:
Sent: 05-27-2015 10:59
From: Steve Pierson
Subject: Obama's Precision Medicine Initiative
Dear All,
Thanks to those of you who provided me information on the important role of statistics/statisticians in precision medicine. Based on that, here is a draft one-pager: www.amstat.org/misc/PrecisionMedicineOnePagerMay27DRAFT.doc.
I'd welcome any comments from the statistical community to improve the document. I'd like to finalize it in the next few weeks for use in our advocacy so please provide comments by noon Eastern on June 5.
Below is a paste of the text from the one-pager.
Thanks,
Steve
Statistics and Precision Medicine
What is Statistics and What is Precision Medicine (PM)?
• Statistics is the science of learning from data, and accounting for the relevant uncertainties. As such, it permeates the physical, natural, and social sciences; public health; medicine; business; and policy.
• PM can be broadly framed as increasing the use of a patient’s specific characteristics in determining right treatment for a given individual at a given time under the given circumstances. PM complements traditional precision medicine implemented by physicians at a bedside using their knowledge and training with data and evidence-based strategies.
Precision Medicine Challenges of Statistical Nature
• Complex data challenges arise in a broad spectrum of PM activities, ranging from genomic medicine to comparative effectiveness research. As such, many of the statistical challenges of Big Data apply.
• When used for developing treatment approaches, data analyses of large data sets should recognize that empirical correlations do not necessarily reflect cause and effect. The latter are investigated using a combination of substantive knowledge and statistical techniques called “causal inference”.
• As each individual is unique, personalized care is inherently based on smaller pools of individuals, making inference more difficult.
• It is essential to ensure the representative nature of large cohort for precision medicine, in order for the results to generalize beyond the initial source of information.
Statistical Objectives in Precision Medicine
• Develop solid foundations for drawing rigorous inferences from electronic health records and other large datasets.
• Develop methods to incorporate rigorous statistical analyses of large datasets into decision analyses tailored to the individual patient.
• Develop data sharing approaches that facilitate learning from the small and highly dispersed set of cases similar to the patient for whom a decision is to be made.
• Develop better methods for validation of risk prediction models, and facilitate sharing of data so that investigators can validate their models
Roles of Statisticians on Precision Medicine Teams
• Statisticians develop approaches to best collect and use the data to inform which treatment is best for a given individual at a given time under the given circumstances.
• Statisticians translate the scientific question into a precise data-based question: this includes carefully describing data structure, the underlying system that generated the data and what we are trying to assess or predict.
• Statisticians develop data analytics for the health scientists to test and further develop theories in order to develop more potent interventions.
• Statisticians are experts in enhancing communication across disciplinary boundaries such as the computational, health, and behavioral disciplines.
The American Statistical Association (ASA) is a scientific and educational society of 19,000 members who serve in industry, government and academia in more than 90 countries, advancing research and promoting sound statistical practice to inform public policy and improve human welfare.
------------------------------
Steve Pierson
Director of Science Policy
American Statistical Association
------------------------------
Original Message:
Sent: 01-30-2015 11:17
From: Steve Pierson
Subject: Obama's Precision Medicine Initiative
Dear ASA Members,
The President announces this morning his Precision Medicine Initiative to accelerate the pace of efforts underway to help make precision medicine the norm rather than the exception (http://www.whitehouse.gov/blog/2015/01/30/precision-medicine-initiative-data-driven-treatments-unique-your-own-body and http://www.nih.gov/precisionmedicine/).
We at the ASA want to make sure the important role of statisticians in precision/personalized medicine is recognized. We're starting by asking you to let us know what you think are the important statistical challenges to be addressed. Please email me at pierson@amstat.org. Suggestions for how we can frame this broad space would also be helpful. (For example, I've heard the work can be categorized into these two broad categories: (i) Right patient for the treatment and (ii) Right treatment for the patient.)
We'd also appreciate hearing about some of the work you are doing. ASA's PR Coordinator Jeff Myers may us some of information provided for media pitches about statistician involvement in this area.
We anticipate a large response to this request so please use email to provide us this information: pierson@amstat.org. I apologize in advance if I don't respond to each email.
Thank you!
Steve
-------------------------------------------
Steve Pierson
Director of Science Policy
American Statistical Association
-------------------------------------------