Artificial intelligence, especially in the form of machine learning, keeps popping up in reports and studies of new and developing health and medical technologies. We’ve written about machine learning used to target Huntington Disease, predict pain level, and identify suicidal adolescents. By feeding algorithms, the machines, huge batches of identified and categorized input, subsequent applications for diagnosis or treatment alternatives present one of the most promising medical tech developments ongoing. When diagnosis or treatment options are a single level binary machine learning is a valuable tool, but when there are multiple possibilities, machine learning’s assistance with differential analysis shines.
Researchers at the NYU School of Medicine recently published a study in Nature Medicine in which machine learning distinguished between two types of lung cancer that are often difficult to differentiate. The algorithms not only recognized adenocarcinoma and squamous cell carcinoma with 97% accuracy but also detected which if any of six genetic mutations commonly associated with lung cancer were present. Genetic differentiation accuracy ranged from 73% to 86%, depending on the gene.
Certain types of cancer such as adenocarcinoma respond to therapies that target specific genes. Conventional genetic tests can take weeks to produce results; faster testing allows care teams to start cancer treatment sooner, which is always a good thing. Using AI for genetic screening has implications for every part and process in the human body influenced by genes, which could result in an explosion of improved treatments.