本文通过一路BBS站telnet客户端发布
Pam Belluck, Is It Boy Or Girl? A Test At 7 Weeks; Simple DNA analysis with high accuracy. New York Times, Aug 10, 2011.
http://www.nytimes.com/2011/08/10/health/10birth.html?_r=1&scp=1&sq=dna%20fetus&st=cse
Excerpt in the window of print: A fast, noninvasive measure of fetal DNA in the mother's blood
My comment:
(a) The report is based on
Stephanie A. Devaney SA et al, Noninvasive Fetal Sex Determination Using Cell-Free Fetal DNA; A systematic review and meta-analysis. Journal of American Medical Association (JAMA), 306: 627-636 (2011).
http://jama.ama-assn.org/content/306/6/627.abstract?sid=e1f99075-32bb-4f85-8a45-896302706c9e
(i) This is a review article, which, through statistics, reviews studies by others instead of conducting its own original scientific study.
* Why scientists have little confidence in Chinese medicine is because few research is done on it and because it has no objective, scientific tests.
* The following is about definitions of parametrics of an accurate diagnostic test. These definitions may be difficult for layperson to comprehend. If so, jump to the last item, (b).
(ii) Please see definitions in
sensitivity and specificity
http://en.wikipedia.org/wiki/Sensitivity_and_specificity
The studies under scrutiny all conduct tests on "Y chromosome sequence." So sensitivity means percentage of a male (who has Y) that is correctly called, and specificity, percentage of a female (who does not have Y) that is correctly identified.
(iii) The "95% confidence interval" measn within TWO standard deviations 標準差 on both tails.
http://en.wikipedia.org/wiki/Standard_deviation
(iv) This is the first research paper that proposed the use of "diagnostic odds ratio" (DOR).
Glas AS et al, The Diagnostic Odds Ratio: a Single Indicator of Test Performance. Journal of Clinical Epidemiology 56: 1129–1135 (2003).
http://vanguardia.udea.edu.co/cursos/Memorias%20precongreso%20aede/2.4.%20Estudios%20de%20pruebas%20diagn%F3sticas/Diagnostic%20Odds%20Ratio.pdf
A "single" indicator, because it combines sensitivity and specificity together into one.
* Formula (1) in the paper present the equation:
DOR = "TP/FN" devided by "FP/TN" = "sens/(1-sens)" devided by "(1-spec)/spec"
, where TP stands for true positive;
FN, false negative;
FP, false positive;
TN, true negative;
sens, sensitivity; and
spec, specificity.
* The paper, immediately following the formula, stated, "The value of a DOR ranges from 0 to infinity, with higher values indicating better discriminatory test performance."
(v) For Area Under Curve (AUC), see
The Area Under an ROC Curve. Division of General Internal Medicine (GIM), University of Nebraska Medical Center (UNMC), undated.
http://gim.unmc.edu/dxtests/roc3.htm
("An area of 1 represents a perfect test; an area of .5 represents a worthless test")
* You see, the figures in this web page has 0-1 in x axis and the same in Y axis, making the total area 1.
* Please click the "Previous Section" twice to understand how the figure is plotted.
(b) A handful of years ago, scientists learned that fetal cells flow in a mother's blood, even years after the pregnancy--similar to cancer cells flowing in a patient's blood. Tests were designed to detect either. There were technical difficulties, as measured by parameters mentioned above. Now the JAMA review asserts that using the fetal DNA to assign sex is good enough. Like ultrasound, there is always false positive and false negative.
--
|