.png)
Today, we’ve got a slew of interesting developments with one central theme: all of this new progress in STEM is fueled at least in part by Artificial Intelligence!
Researchers from the University of Copenhagen and Harvard University recently published a study chronicling their use of Artificial Intelligence to identify patients with the highest risk of developing pancreatic cancer--using nothing more invasive than a patient's medical history. After training on a population of about six million patient histories, the Artificial Intelligence was accurately able to pick out those who would later go on to develop pancreatic cancer based solely on the medical issues they had previously suffered from. Pancreatic cancer is notably difficult to screen for due to the pancreas' secluded location and sensitivity to invasive procedures, so predictive approaches like this one could prove vital in making sure that the testing is applied only in those with the greatest risk. The technology needs further testing and refinement before it can be approved for use in real medical situations, but it may someday serve as a crucial part of the fight against this tricky cancer. Check out the article above to learn more and find the link to the original study!
A while back, we covered the development of a new stent-like implant that would help transcribe a person's brain activity; now, scientists have found a way to non-invasively harness Artificial Intelligence to accomplish the same goal. By having a patient listen to stories while their brain is being scanned with magnetic resonance imaging (MRI), the researchers were able to train an Artificial Intelligence to later transcribe the patient's thoughts by comparing their brain activity in new situations to the brain activity observed during the training period. While the end result certainly contained some errors, it still communicated the point fairly well--and did so without the need for implants or a restricted list of words to use. This new technology holds a lot of promise for benefiting those who have otherwise lost their ability to communicate, like some stroke survivors. Check out the article above to see some examples of the AI in action and learn about how the researchers are approaching privacy concerns in light of this new technology.
Another area where Artificial Intelligence has recently been applied is in the realm of mRNA vaccines. mRNA vaccines show immense promise for addressing a variety of illnesses, but they are limited by the instability of their titular mRNA strands. The speed at which mRNA degrades means that vaccines designed around it may produce a lackluster immune response compared to standard vaccines. However, not all mRNA molecules are made equal; the same protein can be encoded in mRNA in millions of ways, and some variations are much more stable than others. Knowing this, researchers have opted to harness the language-processing skills of some existing AI to assess the near-infinite number of potential mRNA combinations and find the single most stable combination. When applying this new method to the spike protein encoded by our current COVID-19 vaccines, the researchers were able to find a highly-stable mRNA transcript that, among other things, improved immune response a whopping 128 times! This new development may be crucial in refining mRNA vaccine technology into its most safe and useful form, and researchers are eager to continue improving it as more mRNA vaccines begin development. Check out the article above to learn more and find the link to the original study!
One of the greatest medical concerns in the modern era is the lack of new antibiotics to address the growing number of antibiotic-resistant illnesses. Though there are many research efforts ongoing to combat this problem, one effort that has recently produced favorable results involves the use of Artificial Intelligence. Researchers trained an Artificial Intelligence on data derived from attacking one of the most prevalent--and dangerous--highly-resistant "superbugs" (Acinetobacter baumannii) with various drugs and then had the AI comb through a list of several thousand untested chemical compounds to pick out which ones showed the most promise for addressing the hazardous bacteria. The AI pared the list down to 240 promising candidates, which researchers then tested. This testing led to the discovery of the powerful new (experimental) antibiotic called abaucin, which is highly effective at killing A. baumannii--and only A. baumannii. This AI-bolstered method of combing through immense lists of candidates to find only the ones with the greatest chance of being successful could be crucial in the race for new antibiotics. Instead of having to manually test thousands of chemicals, researchers would only have to go through a few hundred. If this technology can be refined and applied on a larger scale, it could ramp up the production of new drugs immensely and potentially even provide highly-specialized antibiotics that would generate resistance much more slowly. Check out the article above to learn more!
Comments