What happened

US health insurer UnitedHealth Group's use of its AI-powered ALERT system to identify patients receiving what it considers excessive mental health therapy has led to denials of coverage for many individuals, despite the model being banned in three US states, according to non-profit newsroom ProPublica.

ALERT (Algorithms for Effective Reporting and Treatment) is a suite of algorithms that was originally designed to identify patients at risk of suicide or substance use but was repurposed to analyse clinical and claims data for unusual treatment patterns and flag what the company considers "overuse" of therapy.

ProPublica's investigation revealed that the company's algorithms flag patients based on treatment frequency, such as those attending more than 30 sessions in eight months or having sessions twice a week for extended periods. The result is a significant risk of losing access to necessary mental health services for vulnerable populations.

Despite having been ruled illegal in California, Massachusetts and New York, the company's Optum subsidiary has persisted in employing what are seen as arbitrary thresholds and cost-driven strategies to manage mental health care access.

Why it happened

The use of algorithms to limit care stems from a broader trend in the insurance industry in which  companies seek to maximise profit by controlling costs. 

Although the US' Mental Health Parity and Addiction Equity Act mandates equal access to mental health care compared to physical care, insurers are allowed considerable discretion in determining what constitutes "medically necessary" treatment. 

The regulatory environment enables practices that prioritise financial considerations over patient needs, often resulting in stricter limitations on mental health services.