AI to find use to reveal human emotions, say researchers

A new AI-based wireless signals could help to know inner emotions of individuals, according to a new study carried out at Queen Mary University, London.

The study demonstrates how radio waves can be used to measure heart rate and breathing signals. Furthermore, the study demonstrates how radio waves can predict the feelings of an individual in the absence of any other visual hint, such as facial expressions.

To examine this, during the initial phase, participants were asked to watch a video selected by researchers. This was to evoke one of the four basic emotions: joy, anger, sadness, and pleasure. Meanwhile, researchers emitted harmless radio signals while the individuals were watching the video, like the ones discharged from any wireless system such as WiFi or radar. The radio signals were aimed toward the individuals, and then the signals that bounced back measured. Thereafter, the changes in the signals caused by slight body movements analyzed, which helped the researchers to unveil hidden information about the heart and breathing rates of an individual.

In fact, previous research used similar wireless or non-invasive methods to detect emotions. However, for earlier such initiatives, data analysis depended on the use of classical machine learning approach, wherein an algorithm is used to find and classify emotional states for the data.

On the other hand, for the new approach, researchers employed deep learning techniques. For these techniques, an artificial neural network learns its own features from raw data that is time dependent. Using deep learning techniques, researchers revealed emotions could be detected more accurately than traditional machine learning methods.

Interestingly, deep learning allows to analyze data in a way similar how human brain would work in the event of different layers of information and to make connection between them.

AI finds use for drug repurposing consequent upon study

In a breakthrough development, researchers have developed a machine-learning method to determine how to improve the outcomes of existing medications for diseases these are not prescribed. The method involves crunching massive amounts of data.

In fact, the objective of the work is to expedite drug repurposing – a concept not new in the pharma sector. For example, Botox injections that were first used to treat crossed eyes now find use to treat migraine, and are a top cosmetic strategy to lessen the appearance of wrinkles.

However, to discover new uses of existing medication is a mix of few things. It is time-consuming, involves expensive randomized clinical trials, and serendipity. The success of such efforts ensure a drug deemed effective for one medical condition will be useful for some other condition as well.

Meanwhile, researchers at the Ohio State University developed a framework that combines massive datasets pertaining to patient care with high-powered computation. This leads to arrive at repurposed drug molecules and estimated effects of existing medications on predefined outcomes.

Whilst the focus of the study is to propose repurpose of drugs to prevent stroke and heart failure in patients with coronary artery disease- the flexible nature of the framework enables its application for most diseases.

“Besides this, the work showcases how artificial intelligence can be used to examine how a drug works on a patient. In addition, the framework expedites hypothesis generation and speeds up clinical trials,” said a scientific associate at the Ohio State University.

Nonetheless, drug decisions will remain with clinicians.

Notably, drug repurposing is an attractive vocation. This is because it could lower the risk related with safety testing of new medications.