Training AI to recognize facial features artificial intelligence can diagnose genetic syndrome with high accuracy
January 9, 2019 Source: Science and Technology Daily Author: Zhang Meng Ran
Window._bd_share_config={ "common":{ "bdSnsKey":{ },"bdText":"","bdMini":"2","bdMiniList":false,"bdPic":"","bdStyle":" 0","bdSize":"16"},"share":{ }};with(document)0[(getElementsByTagName('head')[0]||body).appendChild(createElement('script')) .src='http://bdimg.share.baidu.com/static/api/js/share.js?v=89860593.js?cdnversion='+~(-new Date()/36e5)];According to a paper published online by the British journal Nature Medicine on the 8th, an artificial intelligence can recognize rare genetic syndromes with high accuracy after receiving tens of thousands of real patient facial images. Scientists also stressed that because personal facial images are sensitive but readily available, they must be handled with care to prevent discriminatory abuse of the technology.
Various genetic syndromes exhibit unique facial features that can help clinicians diagnose. However, the number of possible syndromes is huge and it is not easy to identify them correctly. The use of artificial intelligence may help diagnose the genetic syndrome, but the training dataset used in the early studies of this possibility was small and only identified a small number of syndromes.
This time, Yalong Gurovich, a researcher at FDNA Analytical Technologies, and colleagues used a facial image of more than 17,000 patients to train a deep learning algorithm. All of these patients were diagnosed with hundreds of genetic syndromes. The images used in the study came from a community platform where the clinician passed the patient's facial image. The research team used two independent test data sets to test the performance of artificial intelligence, each of which contained hundreds of patient facial images previously analyzed by clinical experts. For each test image, artificial intelligence lists various potential syndromes in a certain order.
In both sets of tests, in about 90% of cases, the first 10 recommendations from artificial intelligence included the correct syndrome, which exceeded the performance of clinical experts in the other three experiments. Although the test dataset used in this study is relatively small in scale and is not directly compared to other existing identification methods or human experts, the results suggest that artificial intelligence is expected to prioritize rare genetic syndromes in clinical practice. Level division and diagnosis.
Researchers say further research is needed to optimize the ability to identify artificial intelligence and compare it to other diagnostic methods.
Gummy Jelly Candy,Yummy Gummy Candy,Gummy Heart Candy,Sugar Free Gummy Candy
Montreal Shantou Food Co., Ltd , https://www.montrealsnack.com