Awesome my AI! 5min recording to diagnose genetic diseases

If you give a 5min recording, how much information can you get from this recording?

According to the latest research by the University of Wisconsin–Madison's Waisman Center and the Wisconsin Institute for Discovery, relying on 5 minutes of recording is enough to determine whether a person is susceptible to genetic-related inheritance. Sexual disease.

1 Machine learning to identify fragile X chromosome syndrome

Recently, this study was published on Scientific Reports under the title "Automated screening for Fragile X premutation carriers based on linguistic and cognitive computational phenotypes". Researchers use machine learning to analyze hundreds of voice records and accurately identify fragile X chromosomes in the pre-mutation phase. Chromosomes with this characteristic increase the risk of neurodegenerative diseases, infertility, etc. The offspring of the population of this chromosome are prone to fragile X syndrome.

Fragile X syndrome is caused by a mutation in the DNA during the formation of the X chromosome in the human body. It is mainly characterized by mental retardation and physical disability. Currently, millions of people worldwide have early fragile X chromosome mutations. .

2 Machine learning - artificial intelligence calculation program to make diagnosis easier

Professor Marsha Mailick, deputy dean of the University of Wisconsin Graduate School, said that there are still no effective diagnostic methods for these pre-mutation conditions. In many cases, patients are unaware of their risk.

Diagnosing early fragile X chromosome mutations is a time-consuming, costly task that requires a lot of resources. To this end, the team of Professor Marsha Mailick hopes to develop a rapid, economical and effective screening method.

Ever since, they developed a machine learning-artificial intelligence calculation program. It is reported that this new type of artificial intelligence robot can "train" through existing data and then analyze new information.

Kris Saha, an associate professor of biomedical engineering at the University of Wisconsin, added that they had to spend hours analyzing and annotating each record in the first place, and it took less than a second to complete the work.

3 Past applications of machine learning

In 2012, a study led by Professor Jan Greenberg, co-author of the study and vice president of the University of Wisconsin, analyzed a 5 min speech record of a mother talking about a child with early fragile X syndrome. Studies have shown that a warm, positive family atmosphere created by parents can reduce a child's behavioral problems.

Audra Sterling, an associate professor of communication science and disease at the University of Wisconsin, used the same recordings to study the results. The results showed that there was a strong correlation between age and speech disorder in the middle-aged and older women with a pre-mutation fragile X chromosome.

The results of the above studies indicate that recording can track the progression of disease in elderly patients with early fragile X-chromosome mutations; systematic speech recording analysis can yield valuable information about families with pre-mutative fragile X chromosomes.

However, previous speech feature coding was time consuming and required clinical expertise, but the methods used in the new study do not require these features.

4 Create a language cognitive function module

Professor Kris Saha, Professor Jan Greenberg, Professor Audra Sterling, Professor Marsha Mailick and graduate student Arezoo Movaghar jointly designed an initial machine learning algorithm that intelligently differentiates patients into two groups: patients with mothers carrying vulnerable X chromosomes and Carrying the mother.

Researchers have also created lists of language and cognitive functions based on recording and machine learning algorithms, such as the average length of sentences in a record or the number of padding pauses, such as "hmm", "ah" or "oh" ("um," "ah , " or "oh.") pronunciation method, these features can very effectively distinguish the difference between the two groups.

The researchers first analyzed the recordings of 100 5-minute mothers talking about children with fragile X-chromosome syndrome, and then analyzed the recordings of another 100 mothers of children with autism spectrum disorders. Based on these remarkable features, the accuracy of machine learning algorithms can be as high as 81%.

According to researchers, the use of machine learning screening methods to diagnose 1,000 patients with early fragile X chromosome mutations in the population can save more than $11 million compared to genetic testing alone.

5 More than fragile X chromosome diagnosis

According to Professor Marsha Mailick, the study is the first step toward a faster, more cost-effective screening process, and they plan to expand screening for other populations, such as men with vulnerable X chromosomes.

Professor Kris Saha said that the machine learning algorithms developed in this study are not limited to fragile X chromosome diagnosis and can be used to diagnose other genetically related diseases in the future.

Graduate student Arezoo Movaghar wants to simplify the way data is collected. He is developing an app for personal medical problems that can track 5 minutes of voice samples, data and even audio recordings from mobile phones or home mini-sounds, and then let machine learning algorithms work to complete the data. The goal of collection. (Original topic: Awesome my AI! 5min recording can diagnose genetic diseases, cheaper than using only genetic testing)

References: 1) Machine learning can detect a genetic disorder from speech recordings2) Automated screening for Fragile X premutation carriers based on linguistic and cognitive computational phenotypes

Reprint, please indicate the source: HC3i China Digital Medical Network
[Editor: Wen Ying TEL: (010) 68476606]

  • 【Collection of this page】 【Copy Link】 【Print】

Man Bath Robe

Best hot selling men bath robe, waterproof dry robe, dry Change Robe, surf change robe, windproof cahnge robe, swim bath robe supplier.

Welcome discuss the customized design and logo.


Selling point:

1. support customized design and logo

2. support small initial order

3. support customized color

4. support sample service


Golden Garment 12Towel Robe 1



Man Bath Robe,Bath Robe,Hotel Bath Robe,Dry Robe Bath Robe

Suzhou Golden Gamrnet MFG Co.,Ltd , https://www.svchangerobe.com