AI in Medicine: More Than a Hypothetical
The average American is optimistic about how artificial intelligence (AI) can change healthcare. Anyone who has ever experienced or been near to a difficult health problem understands that the hope that a rapidly developing technology offers is not a small one. And while medical progress is often cited as one of the most valuable and tangible benefits of AI, it is much more than a talking point.
Over recent years, antibiotic development has slowed while antibiotic resistance has increased. Developing new drugs to treat bacterial infections is neither fast nor cheap. Researchers at the Massachusetts Institute of Technology (MIT) are using AI to bridge the divide between need and resources, and have already discovered two new compounds that demonstrate antibacterial activity. MIT News described how “the researchers employed two different approaches: First, they directed generative AI algorithms to design molecules based on a specific chemical fragment that showed antimicrobial activity, and second, they let the algorithms freely generate molecules, without having to include a specific fragment.” While the mechanics of deep learning AI technology and genetic algorithms might seem distant, the infections targeted by these newly discovered compounds are quite common: MRSA (staph) and gonorrhea, for which Louisiana ranks second nationally in reported cases.
At the University of Southern California (USC), neurodegenerative disease specialists are challenging long held beliefs about the treatability of Alzheimer’s disease. Because “AI sees what the eye cannot,” researchers are better able to interpret and respond to imaging of the human brain. Using AI to examine genes, speech and language patterns, and brain activity means that diagnoses can come earlier and more accurately. Beyond diagnosis, the same methods are being used to formulate novel treatment options, like brain-machine interface systems (BMI). USC also reports that their programs’ embrace of AI has contributed to progress on pharmaceutical breakthroughs and fast-tracked drug discovery.
The AI model, Evo 2, is trained on 128,000 genomes and could hold the key to predicting and understanding genetic disease. Scientists are working with the model to understand its “brain” so that they can interpret its findings, which possess detail and insight into the human genome previously unimaginable. Cancer is among the diseases that stand to be impacted by continued breakthroughs with the Evo 2 model. More time, research, and innovation are all necessary ingredients in the continued development of a tool that would transform the diagnosis of genetic diseases.
The above breakthroughs are neither abstract nor trivial. AI is making tremendous progress on healthcare’s toughest problems, and while more work remains, the innovations of the last year alone are worth celebrating.
