Thursday, November 14, 2002

bookmark: connotea :: del.icio.us ::digg ::furl ::reddit ::yahoo::

Bio-IT World Computer model predicts disease variant based on genetic defect

"University of California, San Diego (UCSD) bioengineers have for the first time used a computer model to relate specific genetic mutations to exact variations of a disease. This is the first model-based system for predicting phenotype (function of the cell or organism) based on genotype (an individual's DNA).

In the study, published in Genome Research (Vol. 12, Issue 11, 1687-1692, November 2002), Bernhard Palsson and his team at UCSD's Jacobs School of Engineering reviewed genetic information from patients who have an enzyme deficiency that causes hemolytic anemia. Physicians have recorded some 150 DNA sequence variations that could be involved in this type of anemia. By inserting the specific DNA sequences into a computer model for red blood cell metabolism, Palsson accurately predicted which mutations would result in chronic hemolytic anemia and which would cause a less severe version of the disease."

redux [06.26.02]
BioMedNet In cancer, microarrays moving rapidly toward clinical use
[requires 'free' registration]

"The ability to see the whole tumor at once in a big picture of gene expression will help scientists understand the "complicated and heterogeneous" biology of cancer, he predicts. It is becoming increasingly clear that there are as many cancers as there are individuals; the same tumor from the same individual may, at different points in time, have different expression patterns."

""Although prognosis is what grabs headlines," he said, "the progress, I have always thought, is in increasing the understanding of the nature of the disease.""

redux [06.20.02]
Science Daily Gene Expression Profiles Predict Survival Of Lymphoma Patients After Chemotherpy

"Patterns of genes that are active in tumor cells can predict whether patients with diffuse large B-cell lymphoma (DLBCL) are likely to be cured by chemotherapy, scientists reported today in the New England Journal of Medicine. Researchers analyzed thousands of genes in lymphoma biopsy samples from patients with DLBCL and determined that the activity of as few as 17 genes could be used to predict patients' response to treatment. "We're able to reliably predict the survival of these patients using data from a small number of genes, indicating that this technique should be entirely manageable for routine use," said National Cancer Institute (NCI) investigator Louis M. Staudt, M.D, Ph.D., the senior author on the study."

redux [05.31.01]
Wired News Fingering Cancer Genes

"Genes have fingerprints just like fingers, which got one cancer researcher thinking.

Since the FBI uses neural networks -- a type of artificial intelligence built to imitate neuron function in the brain -- to sift through masses of computerized fingerprint data to solve crimes, why not do the same for genetic fingerprint data?"

""We trained (the neural networks) to recognize this is one cancer and this is another and this is not a cancer," Kahn said. "Eventually it learned to recognize particular features that were particular for cancer.""

Family Physicians' Electronic Network Diagnostic Algorithms: results at last!

"We seem to forget, sometimes, that the first researchers in AI that chosen medicine as a problem domain did so, not because of an interest in medicine, but because of an interest in diagnosis as an example of intelligent behavior. Medical diagnosis was one example (perhaps a poor one given that there are much simpler and easier models in other physical systems). Automated diagnosis has rarely interested the medical community, not because of a fear of removing the human element (we've already done that with our reimbursement system) or of replacing humans with machines but, more simply, because diagnosis (as most people view it), is not really the problem. Most clinicians manage some form of diagnosis and most patients are treated appropriately. What is needed is better information on the utility of information and the means to obtain it which least stresses the system."