Self-harm and harm to others are two of the most worrisome behaviors of people with autism spectrum disorder (ASD).

But what if you could predict these behaviors before they happen?

Advances in autism research and wearable biosensor technology are giving doctors new avenues to approach the disorder that could have significant impact on harmful behavior.

In the United States, about have been diagnosed with ASD, according to the Centers for Disease Control and Prevention (CDC).

One of the agency’s was that “nearly 28 percent of 8-year-old children” with the disorder can exhibit self-harming behavior.

But there could be help on the way.

Matthew Goodwin, PhD, an assistant professor at Northeastern University, tells Healthline that using data from wearable biosensors “can predict with about 79-80 percent accuracy based on the last three minutes of your physiology whether or not you are going to aggress to someone else or aggress to yourself” within the next minute.

Goodwin presented his findings this week at the in San Francisco.

Read more: Helping parents who have a child with autism »

How the technology works

The technology he is using isn’t necessarily groundbreaking, but its applications are.

The sensors are placed in various key locations on an ASD patient’s body. They take a variety of physiological and biometrical data.

Accelerometers placed on the wrist and arm track limb movement in three-dimensional space. Skin surface temperature, electrodermal activity (sweating), and heart rate give indications of the autonomic nervous system.

Currently, an ASD diagnosis relies on behavioral studies rather than physiology, but the data collected from these sensors has a myriad of applications.

In his presentation this week, Goodwin focused primarily on two applications for biosensor data in ASD patients.

The first of those applications is in analyzing stereotypical motor movement — things like arm flapping, rocking back-and-forth, finger flicking — that are defining parts of ASD. Currently the function of these movements is not entirely understood within the scientific community.

Goodwin suggests that these actions could be communicative. They could regulate stress or even help with proprioception, although none of this is certain. That’s why more research is needed.

Read more: Why girls shows signs of autism later than boys »

Easier way to study patients

Currently, ASD patients are recorded on video to study stereotypical motor movement. Their movements are then broken down and coded frame by frame.

Researchers study how long movements occurred, how often they occurred, and what their pattern was.

It’s not practical says Goodwin.

His research utilizes data from biosensors to create an algorithm that can identify stereotypical motor movements.

“What we have demonstrated,” he said, “is that with 50 examples across about 15 individuals with autism, that algorithm can perform with between 90 and 99 percent accuracy” compared with traditional coded video.

Using computer data analysis dramatically increases the scope of how much data on ASD patients can be recorded and studied.

“Now you have a scalable solution,” he says.

Read more: Brain chemical doesn’t do its job in people with autism »

Bridging the communication gap

Goodwin’s second application of biosensors looks closely at indexes of the autonomic nervous system that can tell things about an ASD patient they may not be able to communicate.

Because the disorder is typically defined by deficits in social communication, it can be hard “to identify an internal state and then relay that to someone else,” said Goodwin.

What biometrical data can do is fill in the gaps created by that lack of communication.

Even if the person with ASD is incapable of saying how they feel, physiology is able to show whether the person is feeling excited or relaxed based on factors such as heart rate and the presence of sweat.

Read more: Alternative treatments for autism »

Potential new ways to treat autism

Autism envelopes a variety of symptoms and severities — which makes it difficult to treat.

Using biosensors to collect information may help to create better tailored treatments for differing individuals within the autism spectrum.

Thomas Frazier, PhD, the chief science officer of advocacy group , says, “I think this approach could be useful as a way to monitor treatment progress in clinical trials. More research may be needed to identify reliable epochs of data collection and sensitivity to change, but given what is written this could be a useful treatment tracking approach.”

Goodwin hopes to make the biometrical sensors available commercially. Unlike other sensors and wearable devices, these sensors are developed to be form-fitting in a way that is tolerable for those with ASD.

The devices, he says, have peer review in the medical community and “they are on the pathway to being certified as a medical device with FDA approval, but we are not there yet.”

“This idea that you can record biology and detect a departure in a normal signal, which is predictive of a future behavioral event that something bad is about to happen,” he said. “I think that would be a tremendous innovation, if you could keep people, themselves, and those around them safer because we’re not caught off guard.”