fbpx

The AI revolution will leave no stone unturned, and no industry untouched. In sectors from manufacturing to healthcare, the rise of complex machine learning programs promises to completely overturn old ways of doing business–for the better.

Behavioral Health is one area where we haven’t heard much about AI’s potential impact–even if it is one field that will be completely transformed by its introduction. Expect to see positive developments: more accurate diagnoses, more effective data collection (and as a result, more     effective treatments), and from this increased efficiency, cheaper costs.

In short, everyone wins–even workers. Executed properly, no one has to lose their job to an algorithm. Let’s take a closer look.

Tracking outcomes

AI’s great strength is in data collection and analysis. I know this firsthand, having built a successful business out of applying algorithms to behavioral health operations. But AI can go far beyond just streamlining billing processes or cutting out tedious phone calls with insurers. In fact, AI can help doctors tailor their strategies to patients, based on their individual medical history.

We need evidence-based strategies–yesterday. Consider the mental health and substance abuse epidemics that we face today: in 2015, opioids killed 33,000 peoplesurpassing gun homicides for the first time in American history. No area has remained untouched, from big cities to rural towns alike; as families are ripped apart, orphaned children are filling foster homes at an unprecedented rate.

With a public health problem tearing apart our society, one would think that there would be plenty of research on the treatments that work and the ones that don’t. Yet this is not the case: unfortunately, there doesn’t seem to be a scientific consensus, as the addiction studies field suffers from a lack of meaningful regulation or oversight. There are no uniform guidelines, licensing requirements, or best practices, either at the federal or state level.

Even if Congress were to pass more robust legislation tomorrow, there’s no guarantee that this bill would have the right priorities. Would we legalize marijuana to replace opioids, for instance? Ban opioids entirely and turn to alternatives? To really grasp the extent of this problem, look no further than this feature from The New York Times: the victims profiled are prescribed a wide range of different treatments, from therapy to medications like Suboxone.

This is where AI comes in, as it can help track outcomes, determining the most effective courses of action not simply for the opioid crisis in general, but also for individual patients. Consider this: one major strength of AI is its ability to process reams of data rapidly, at a pace no human (or team of humans) can ever hope to match. In 2003, Cochrane, an independent network of healthcare researchers, estimated that it would take 30 years to review all existing randomized trials (not including new research since that time).

These efforts have already begun. Watson, IBM’s flagship machine learning program, has already been tasked with researching the journey into addiction, unearthing why certain patients prescribed opioids become addicted while others do not. Another, similar study, carried out by Watson and MAP Health Management, aims to identify patients at risk of relapsing, and intervene with treatment and intervention.

Though neither of these efforts attempt to determine the effectiveness of different types of addiction treatments, they still use many relevant methods: precise data on metrics from types and quantities of opioids to associated diagnoses, patient timelines, and even insurance claims. It’s only a small step further to use AI to track and determine the best treatments to address the opioid epidemic and even provide specific services on a case-by-case basis. For instance, one person could receive buprenorphine and see an EMDR therapist to address childhood trauma, while another could receive talk therapy and medication to address their depression.

AI can even diagnose mental health disorders

Given that AI is already being used to find and assist patients who are likely to relapse (or become addicts in the first place), the next step would be AI diagnoses. Of course, it’s daunting to trust a machine with your vital data. But AI is already providing highly accurate diagnoses in a range of fields, from lung cancer to heart disease. Mental health is no different, as the field of computational psychology demonstrates. One study from the Cincinnati Children’s Hospital Medical Center shows that machine learning can be almost 93 percent more accurate in identifying suicidal tendencies than humans.

The key to the AI’s success lies in the details: AI was able to take into account many more variables than humans, who often looked for a single risk factor like major depression. This task is tailor-made for machine learning programs, which can improve and evolve their understanding independent of human intervention by gathering and processing data. One program can even consider factors like tone, vocabulary, sentence length, and even muddled meaning in diagnosing psychosis.

We still have a long way to go before we can introduce this sort of technology into every clinic or office. But we’re getting there: there are smartphone apps that partner with licensed therapists and even programs that can look at Instagram photos to assess mental health risks.

The AI-Human Partnership

Still, there’s one lingering question: can we apply AI to Behavioral Health without displacing humans?

Yes, because AI is unlikely to ever replace humans completely. For instance, psychologists have only about a one percent chance of being replaced, only slightly less likely than medical practitioners (two percent), a group which includes therapists, psychiatrists, and so forth. Much of this has to do with the nature of these duties: it’s not easy to automate a role that requires abstract thinking (such as creativity and complex problem solving) or social intelligence (such as empathy and negotiation skills). Behavioral health has all of the above, in abundance.

Though machine learning may diagnose patients more accurately than humans, it only serves as an aid for providers. Think of it this way: algorithms will assist physicians by spotting mistakes, listing any possible risk factors, or recommending the best possible course of action. Even apps that help spot depression don’t necessarily treat them on their own; the vast majority bridge the gap between having a friendly voice to talk to and professional help.

Here, I speak from experience. Repetitive work is the easiest to automate, so you’d think that I would have replaced my operations associates with AI at the first opportunity. In reality, my staff simply shifted to higher-level responsibilities. Now, thanks to accurate analytics and data, the operations team (with the support of AI) is helping my clients and I chart new business strategies, weed out inefficiencies in our processes, and track progress. We are getting far more done–with far less people.

That, ultimately, is perhaps the best way to summarize the impact of AI on Behavioral Health: do more with less. AI can identify best practices, collect important patient data, help tailor treatments to match individuals, and elevate the standard of care. Fewer patients will fall through the cracks, and overworked, stressed providers will reduce their chances of missing key symptoms or warning signs. AI can help create a safety net that won’t leave patients behind–without sacrificing any jobs.

******

Ali Beheshti is an entrepreneur with extensive experience in behavioral health. He is the Founder and CEO of Zealie, an enterprise technology company poised to transform the behavioral health industry.

Read More

Privacy Policy

©2019 ZEALIE LLC. All Rights Reserved.

Call Now Button

Log in with your credentials

Forgot your details?