Machine learning algorithms have proven to accurately predict opioid overdose risk, which could allow doctors to tackle the opioid epidemic by honing their resources to high-risk people.

That’s according to research published in JAMA that was designed to “develop and validate a machine-learning algorithm to predict opioid overdose risk among Medicare beneficiaries with at least 1 opioid prescription.”

The US opioid epidemic has reached staggering heights in recent years, adding hundreds of billions of dollars in US healthcare costs and taking the lives of over 40,000 people in the US in 2017 alone. With healthcare organizations staggering under the weight of the epidemic's costly consequences, researchers have turned to artificial intelligence (AI) in the hope of finding some light at the end of the tunnel.

The result of the project is a machine learning tool that proved to be powerful in assessing overdose risk. According to the report, the algorithms developed accurately identified individuals at high risk of an opioid overdose. The algorithms sorted 560,000 Medicare beneficiaries into risk-based groups. And the machine learning tool's groupings were almost entirely accurate: More than 90% of overdose episodes occurred in the high-risk group.

Moreover, the AI-powered tool proved a better predictor than traditional methods. For example, the traditional method used by the Centers for Medicare and Medicaid Services (CMS) classified 70% of people who experienced an overdose as low-risk.

Moving forward, say observers, AI-powered tools can help healthcare firms channel their drug abuse spending where it's needed most and curb the opioid epidemic's consequences.

In particular, machine learning could guide health firms to better allocate precious opioid prevention resources to patients who need them most. In addition, providers and payers could develop more cost-efficient opioid prevention programs by targeting resources at the 25% of patients prescribed opioids who suffer 90% of overdoses.

Another benefit of targeting opioid prevention resources at high-risk patients is that it could reduce costly overdoses. If opioid prevention spending is concentrated among high-risk populations — versus spread out across the population — payers and providers may have a higher chance of preventing overdoses.