cool hit counter IBM's AI system that can predict if you'll get a mental illness just by watching you talk_Intefrankly

IBM's AI system that can predict if you'll get a mental illness just by watching you talk


Editor: yuxiangyu

Language is a fascinating tool that allows people to share ideas with each other. Often, language resonates when its clarity and accuracy are fully utilized. And language is also a tool that psychiatrists use to assess patients for specific psychiatric or psychotic disorders, including schizophrenia. However, these assessments often require trained professionals and specialized equipment.

And a team of IBM Research's computational psychiatry and neuroimaging group and universities worldwide. Together, they have developed an AI that can still predict psychotic episodes in patients with relative accuracy without the need for specialized personnel or equipment. Their psychological prediction study has been published in the Journal of World Psychiatry.

This group built on the results of the 2015 IBM study to demonstrate that differences in speech patterns between people at high risk of developing future psychosis and those who do not can be modeled using AI. In other words, they used natural language processing (NLP) methods to quantify the concepts of "verbal poverty" and "thoughtfulness" as syntactic complexity and semantic coherence, respectively.

Their AI then evaluated the speech patterns of patients whom the researchers had instructed to talk about themselves for an hour. "In our past research, we were able to construct a predictive model with manual scores that achieved 80 percent accuracy, but automated functions that achieved 100 percent," Guillermo Cecchi, principal investigator and manager of IBM Research's Computational Psychiatry and Neuroimaging group, told Futurism.

In their latest study, the researchers evaluated a larger group of patients engaged in a different type of speech activity: talking about a particular story they had just read. By using what they learned from their 2015 study to train their psychiatric prediction AI, Cecchi said, the team was able to build a retrospective model of patient speech patterns.

According to this study, this system can predict a patient's eventual psychotic episode with an accuracy of 83% already. Applying it to the patients in the first study, the AI would have predicted an accuracy of 79%.

IBM researchers' psychosis prediction AI could eventually help mental mental health practitioners and patients. As Cecchi had written in a 2017 IBM study: traditional methods of assessing patients are too subjective. He and his team believe that using AI and machine learning as computational psychiatry tools can eliminate this subjectivity and increase the likelihood of accurate assessments.

Cecchi told us, "This system can be used in the clinic. Patients considered at risk can be classified quickly and reliably, so that limited (always limited) resources can be used to treat those considered at high risk of developing a first-episode psychosis. "A layperson or someone from the clinic can evaluate remotely by sending an audio sample.

AI technology will bring about a real revolution in medicine, and as these advanced systems become mainstream, we will enter a new era in healthcare that will hopefully enable anyone, anywhere, to get the best diagnosis and treatment.

2、blockchain What is a graphene frame
3、Huawei unveils new AI development platform the Kirin 970 is firing on all cylinders

    已推荐到看一看 和朋友分享想法
    最多200字,当前共 发送