Networks as a Predictor for the Spread of Cancer

In class, Dave has talked about survival rates of cancer patients . Models and networks are cropping up in other useful ways in the medical field around cancer as well. I came across a study by 6 doctors who used an artificial neural network to see if this could be an effective way of predicting patterns in lymph node metastasis. 

An artificial neural network (ANN) is a computational model that can process vast amounts of data and complex relationships and find patterns. An ANN "learns" from the data and changes accordingly. Wikipedia has a page about ANN that does a nice job explaining what they are an ANN is and how it is applied in many different contexts. 

Lymph node metastasis is particularly significant because in certain kinds cancer, namely gastric and esophageal, the lymph system is the first place to which the cancer spreads. So an accurate prediction of this could help catch and effectively treat the cancer in its early stages. 

In this study, the doctors were specifically looking to determine if genetics were a leading factor in the predictability of lymph node metastasis. By using a mass of data collected by the National Cancer Center in Japan the doctors created an artificial neural network, used more data to refine it, and then tested its accuracy. In a separate study in Germany, the same ANN was was able to predict the incidence of lymph node metastasis as accurately as 96% for certain sub-types of cancer. This could mean more accurate, more efficient, and thus more effective surgeries and postoperative care, with fewer side-effects. The way surgeries are done now, lymph nodes that are not cancerous but appear irregular for other reasons are resected. The removal of lymph nodes not only decreases immunity, but can interfere with lymphatic drainage leading to such unpleasant conditions as lymphedema. Perhaps the use of an artificial neural network could prevent damage to the real living network that is the lymphatic system.