For years Facebook, and companies like it have been working to improve Artificial Intelligence (AI) models to predict consumer behavior and improve advertising. But now, with Facebook's recent announcement of suicide prevention models in their AI, we may have a glimpse of what medicine in the future will look like.
As it turns out, the U.S government and even startup entrepreneurs are doing similar work. Suicide and drug addiction are tough topics, but let's take a look at the history here and what's being done to help all of us handle these issues better.
How Facebook AI predicts suicides
For several years, Facebook's AI development focused on one thing: learning consumer behavior to help improve their experience on the platform.
The AI team started out small and took days to complete experiments. Now it involves several hundred people in different departments running multiple experiments every day. Facebook advanced its AI quickly and at this point is using it for suicide prevention.
Now that any user can do their own live video, people can be exposed to something horrific. One case in point is the October 2017 suicide of a man in Turkey who shot himself live on the site while viewers pleaded for him to stop.
Facebook is using human facilitators and user reports, along with speech recognition, language processing, and other technologies. The software scans posts as they go live, watching for signs of distress and bringing it to trained response teams' attention.
This type of machine learning frees humans from combing through minutes of video or text. It also uses the site's reporting options, which were made more visible to users. Facebook says it can now notify authorities twice as quickly as it did before in many cases.
Battling a national suicide epidemic
Similar tech is now being used for veterans as well. In what has become a national epidemic, 20 veterans commit suicide each day. The government, military, and concerned citizens have turned to this type of software for help.
The Durkheim Project is one such effort. Veterans give the program permission to analyze their social media and mobile content. The machine intelligence predicts whether there's a risk of suicide based on their language. Then it notifies medical professionals and social workers so they can help the veterans. The Geisel School of Medicine at Dartmouth, the U.S. Department of Veterans Affairs (VA), Patterns and Predictions, and Cloudera joined forces for the project.
The initial test successfully discovered suicidal intent at 65 percent accuracy, higher than cutting-edge medical approaches can. This gives medical professionals a new, powerful tool to help spot potentially suicidal patients.
The project has drawn strong support from Congress and veterans across the U.S. Teams are continuing work so that all veterans can get this innovative type of care.
Identifying addiction from smartphone data
Organizations fighting drug addiction are using AI too on mobile platforms with apps, resources, and more.
New apps, like one from Triggr Health, provide an example. They collect and measure user screen engagement, texting habits, phone logs, sleep app data, and location services. These apps then use the data to help people resist cravings or triggers that cause drug relapse. That data is combined with information gathered from the participant, such as drug preferences, history of use, and trigger words. Machine-learning then spots potential risks and even notifies the customer's care team.
The underlying theme here is that language recognition and related technologies, even in their infancy, are starting to play a larger role in our health care. It's changing what's possible. Technological innovations like the seatbelt or polio vaccine resulted in thousands of saved lives, and it looks like AI is following in those footsteps.
Suicide and drug use are tremendously difficult subjects to talk about for most people. But, whether it's your favorite rock musician or a family member, these things hit close to home.
We can't help but take notice of how AI is taking on these vital issues and helping people. That's something to stay positive about.
If you have a small business, you have to wonder what you can do to use machine learning or natural language processing to make your products easier to use. How can we apply machine intelligence that's more attractive to newer generations of consumers who are increasingly comfortable with the technology?
These are certainly tantalizing questions. We're on the cutting edge here. And as machines start to do more and more things that only humans could once do, more opportunities will emerge for any of us entrepreneurs who think one step ahead.