Search This Issue

Home Site

Advertise Here - Reach those who use and understand AI Technology.

   Page 25
Sample Issue - Subscribe TODAY to see the entire issue
18.3 Table of ContentsBottom of PageSite Map

and its growth shows no signs of stopping. The human brain simply cannot deal with so much information. Therefore, it is necessary to employ artificial intelligence techniques in the process of scientific research. Artificial intelligence provides increased speed and "capacity" compared to the human brain, which proves incredibly useful to researchers.
     Specifically, neural networks are a very useful tool. Neural networks mimic the behavior of the human brain, so they are often used in applications where systematic thought processes are important. Processing units are interconnected and communicate like biological neurons. Successful and relevant connections are emphasized whereas irrelevant connections are downgraded. This mimics the conditioned learning of biological organisms. This dynamic learning is vital in the research process since science does not remain static.

Cancer Prediction
     Cancer is a difficult disease to research. Cancer results from everyday cells that, in effect, "lose their mind." In normal, healthy cells, the process of cell replication is highly regulated. Many "checkpoints" are in place to make sure that all steps are done properly. These "checkpoints" are usually molecules that are created or degraded which signal the cell to either hold in the stage where it is or to move on to the next stage of cell division. For example, one such molecule believed to be involved in several different cancers is called p53. This molecule is often absent or mutated in extracts from tumor cells, indicating that perhaps p53 is a molecule that signals the


cell to stop dividing. If p53 is not produced, then cells will divide too rapidly and cancer can result.
     Since no two cancer physiologies are exactly alike, it is notoriously difficult to make a prediction about whether the patient is likely to survive. Even with two cases of the same type of cancer, one person may survive for many years while another will die within months. A myriad of different physical factors contribute to the outcome. This is one reason why cancer is such a frightening disease - no one can tell you whether your particular complement of genes will put you at an advantage or disadvantage.
     Researchers at the National Cancer Institute (NCI) are working on creating a model that will help predict the prognosis of cancer patients. Particularly, they are working with a type of cancer called Neuroblastoma. Neuroblastoma is a childhood cancer. It usually begins with cells of the adrenal gland and spreads, creating tumors in the neck, chest, abdomen or pelvis. Using DNA microarray analysis (see PC AI Volume 18, Issue 2 - "Microarrays and Artificial Intelligence") the researchers studied the gene expression profiles of cancer patients. The microarrays consisted of about 25,000 genes and the analysis was repeated for 49 patients. In order to connect the microarray results with certain outcomes, the 49 patients were chosen because their outcomes were known. Some of the patients survived for more than three years without any cancer-related issues. Others died due to the disease. Feeding this information into an artificial neural network, the researchers found they were able to predict whether a patient would

To Page 24

18.3 Table of Contents
Top of Page

To Page 29


PC AI Magazine - -
e-mail: PCAI.TH @ - Comments: webmaster @