For the moment, artificial intelligence remains confined to the realm of science fiction, but it is important to note that the same cannot be said about machine learning, which is when computers learn through their experiences in much the same manner as living entities. Given that computers possess a number of important strengths relative to their flesh and blood counterparts, it should come as no surprise to learn that machine learning is seeing use for a wide range of purposes in a wide range of fields, with medicine being no exception to this rule. In particular, it is interesting to note that machine learning is about to be applied to the war on cancer, which remains the recipient of billions and billions of dollars from sources situated all around the world on an annual basis.
To elaborate, the war on cancer receives so much funding for a couple of reasons. First, cancer rates are rising throughout the world in spite of the fact that it is one of the most common killers of humans, which can be connected to the simple fact that more and more people are living longer and longer. After all, cancer occurs when mutations cause cells to spread dividing out of control without being checked by natural processes, which is rare because few mutations cause such problems and fewer mutations still can enable the cancerous cells to elude the attention of the immune system. However, the longer that someone lives, the higher the chance that such a mutation will come up because of the increased number of opportunities, thus leading to a higher rate of cancer. As a result, there is no one who is exempt from the terror of cancer, which in turn, means that there is enormous demand for cancer cures. Second, it should be noted that combating cancer is neither simple nor straightforward. After all, the limitations of current techniques and technologies mean that most cancerous tumors have to be treated by being cut out through surgical procedures before the remaining cancerous cells are eradicated using either radiation or chemical treatments. Since neither radiation nor chemical treatments are all that precise, this means that the fight against cancer is often a grueling process with a whole host of unwanted side effects, which is why there is so much interest in better cancer cures with fewer problems.
However, it is important to note that while much progress is being made on new cancer cures such as nanobots that will reduce the side effects of chemical treatments by delivering their chemical payloads right to the cancerous cells so as to avoid affecting their non-cancerous counterparts, there is still enormous interest in tried and true solutions such as cancer drugs, as shown by the latest news out of the Pharmaceutical Artificial Intelligence group at Insilico Medicine, Inc. In brief, said team has shown that generative adversarial networks can be used to come up with new cancer drugs, meaning that its next step will be the creation of an actual generative adversarial networks-based engine that will be used for said purpose. Something that could not just speed up research and development in the field of pharmaceuticals but also increase the chances of success in clinical trials, which often involve enormous amounts of time, effort, and other resources for nothing that can be turned into a commercial product.
What Are Generative Adversarial Networks?
Before elaborating on this occurrence, it is important to explain generative adversarial networks. In brief, generative adversarial networks come up with useful outputs by pitting a pair of neural networks possessing machine intelligence against one another in a competition. One neural network is generative in nature, meaning that it is responsible for coming up with outputs that are as close to true examples of the desired outputs as possible. In contrast, the other neural network is discriminative in nature, meaning that it is responsible for sorting true examples from false examples of the desired outputs. By pitting these two neural networks in competition with one another, the generative neural network becomes better and better at coming up with outputs that resemble true examples of the desired outputs, while the discriminative neural network becomes better and better at discarding examples that fail to meet the necessary criteria of the desired outputs. Combined, this means that the generative adversarial networks become better and better at producing the desired outputs over time as they learn more and more from their experiences, much more so than what humans can manage on their own so long as the pair of neutral networks have been programmed in the right manner.
How Will Generative Adversarial Networks Be Used in Combating Cancer?
The Pharmaceutical Artificial Intelligence group has proven that generative adversarial networks can be used to come up with new molecules that possess potential as cancer drugs based on robust parameters set by their users. It has not created such an engine so far, but its paper means that it can be expected sometime in the near future, which could have profound implications for the field of pharmaceuticals as a whole. After all, while the new molecules created by the generative adversarial networks will still have to undergo clinical trials to confirm their effectiveness for their intended purpose as well as the extent of their side effects in those who use them, this could be a much superior alternative to the current methods for coming up with new molecules, which in turn, could mean not just better results in less time but also a radical transformation in the structure of the industry as a whole. Something that is particularly true because while the Pharmaceutical Artificial Intelligence has focused on the creation of potential cancer drugs, there is no reason to think that generative adversarial networks could not be put to use in coming up with new molecules with potential uses in combating other medical conditions, which could mean the same radical transformation for those particular parts of the field of pharmaceuticals as well. Summed up, while there are still numerous challenges to be overcome before generative adversarial networks will be used as this paper suggests that they can be used, the field of pharmaceuticals promises to be extremely interesting in the near future, meaning that interested individuals should make sure to keep a close eye on the latest developments.