Breakthrough in the diagnosis of genetic disorders
So-called artificial intelligence (AI) is increasingly used for the early diagnosis of diseases. Researchers have now developed a new AI technology that can accurately identify rare genetic disorders using a photograph of a patient's face.
Scientists from Tel Aviv University and FDNA Inc have developed a new form of diagnosis of rare genetic disorders that only requires a photograph of the patient's face. The doctors published the results of their study in the English-language journal "Nature Medicine".
AI surpasses medical diagnosis
The new artificial intelligence called "DeepGestalt" surpassed medical professionals and experts in identifying a number of syndromes in three different studies. Using this form of diagnosis could significantly improve the personalized treatment of patients.
Many genetic diseases have recognizable facial features
The current study found that eight percent of the population suffer from diseases with important genetic components. Many of these diseases show recognizable facial features, the researchers say. The new technology could, for example, identify the so-called Angelman syndrome, a nervous system disorder with characteristic features such as a wide mouth with widely spaced teeth. Strabismus could also be detected, a disease in which the eyes point in different directions.
How was the learning algorithm trained?
The results show how the most modern algorithms can be used to deal with demanding problems for which there is so far only a small amount of available data, the scientists explain. In the future, important applications could be developed and new genetic syndromes identified with the help of the technology. The learning algorithm was trained using 17,000 patient facial images from a database of patients with over 200 different genetic syndromes.
Possible disadvantages of the new technology
Since images of faces are easily accessible, this could lead to employers analyzing facial images, for example, and discriminating against people with pre-existing illnesses or medical complications, the study authors warned.
How reliable is the new AI?
The research team found that AI technology outperformed medical professionals in two separate sets of tests when it came to identifying a target syndrome among 502 selected images. In each test, the AI suggested a list of potential syndromes and identified the correct syndrome in the top 10 suggestions in 91 percent of cases. The technology uses the learning algorithm together with the facial features of the provided image and then creates a list of possible clinical pictures. All images used in the studies come from patients who have already been diagnosed with a disease.
AI's potential in healthcare is huge
The collection of ever larger and better curated medical data sets allows AI instruments to predict genetic mutations based on phenotypes of imaging, which can reduce the burden on health systems and improve patient care. There are still some limitations to be addressed to ensure that the proposed algorithms are robust, clinically accurate, and applicable to different age groups and ethnic groups in the hospital, the study's authors say. However, the potential of AI in healthcare is enormous. (as)