据美国麻省理工学院和哈佛大学的研究人员利用深度学习等 AI 技术发现了一类新型抗生素，可杀灭一类名为“耐甲氧西林金黄色葡萄球菌（MRSA）”的耐药性细菌。这种细菌以往每年会导致全美超 10000 人死亡。
研究人员使用了大幅扩展的数据集训练了一个深度学习模型，并测试了大约 3.9 万种化合物对金黄色葡萄球菌和来自肝脏、骨骼肌和肺部的 3 种人体细胞的影响，再将测试数据用以训练 AI 模型。通过训练，AI 模型将能够预测化合物的抗菌活性和对人体细胞的潜在毒性，研究人员发现了既能够杀死微生物，又能把对人体产生不良影响降到最低的化合物。
在两种小鼠模型（具有上述耐药性细菌的皮肤感染模型、被这种细菌感染全身的模型）的测试中，每种化合物都能将对应耐药细菌的数量减少 10 倍。
科学家从而成功确定了一种新的潜在抗生素类别，包含有两种具有潜力成为候选抗生素的化合物。
新闻翻译：
A new class of potential antibiotics has been discovered by researchers at the Massachusetts Institute of Technology and Harvard University using artificial intelligence technology. The antibiotics can kill resistant bacteria, including “耐甲氧西林金黄色葡萄球菌” (MRSA), which can cause death for tens of thousands of people in the United States every year.
Researchers used a large-scale dataset to train a deep learning model and tested the effects of about 39,000 compounds on金黄色葡萄球菌和来自肝脏、骨骼肌和肺部的 three types of human cells. The test data was then used to train the AI model. Through training, the AI model can predict the antibiotic activity and potential toxicity to human cells. Researchers have discovered compounds that can kill both microorganisms and reduce the adverse effects on humans.
In tests on two小鼠 models (skin infection models with resistant bacteria and models with systemic infections), each compound reduced the number of resistant bacteria by 10 times.
Scientists have successfully determined a new potential antibiotic class, which includes two compounds with potential to become candidates.
【来源】https://www.ithome.com/0/741/310.htm
Views: 2