How Can AI Help Improve Medication Therapy Management?

Every day, patients across the U.S. miss doses, delay refills, or forget a medication entirely. Not out of neglect, but because life is busy, instructions are confusing, or side effects are frustrating. Medication non-adherence, a challenge well-known in healthcare, affects far more lives than it should.

More than 125,000 Americans die each year due to complications from not taking their medications as prescribed. It’s one of healthcare’s most fixable problems and also one of its most expensive, driving up to 25% of hospitalizations and over $500 billion in avoidable costs.¹ Yet, these aren’t failures of individual willpower but signs of system gaps we can close.

Many patients meet adherence metrics yet remain at risk, as traditional systems often fail to detect hidden patterns until complications emerge.

Even with Medication Therapy Management programs and counseling calls scheduled, vulnerable patients often go unnoticed until complications arise. Traditional approaches rely on monthly refill reports or past adherence scores and tend to identify issues only after complications arise, when a patient's diabetes is already out of control or their mental health has significantly deteriorated. Some patients go unnoticed because, on paper, their adherence looks perfectly fine. They might pick up prescriptions on time but skip doses, split pills to save money, or stop taking them altogether due to side effects. These patterns rarely show up in standard reports, which highlights the limits of one-size-fits-all programs. What’s needed are intelligent, adaptive tools that identify risks before they escalate and reflect the complexities of real-world medication behavior.

Teams focused on Medication Therapy Management (MTM) know that improving adherence drives multiple outcomes: fewer complications, lower emergency utilization, and higher CMS Star Ratings.^{2, 3} Equipped with the right data, technology, and care strategies, they’re turning insight into early action. AI is already enabling promising advances in healthcare through better analytics, personalized care, and clinical decision support, though continued validation and oversight are key to realizing its full potential.⁴

Modernizing Medication Therapy Management with AI

A recent study published in the Journal of the American Pharmacists Association evaluated OpenAI's ChatGPT version 4o's ability to perform Medication Therapy Management tasks across 39 clinical patient cases of varying complexity. Results showed that ChatGPT accurately identified drug interactions, recommended alternative therapies, and generated management plans with high effectiveness, especially in simpler cases, highlighting its potential as a supportive AI tool in clinical pharmacy.⁵

While ChatGPT is being explored for use in healthcare, it's not currently approved for clinical applications due to concerns about its potential to generate false or misleading information - a phenomenon known as ‘hallucination’ - as well as challenges related to data security and patient privacy regulations such as HIPAA.⁶

Medication adherence is evolving with AI analytics tailored to the complexity of healthcare , delivering earlier interventions and more precise support. Instead of relying solely on human bandwidth, AI can continuously scan data like pharmacy refills, medical claims, and social factors to flag patients at risk, often before problems escalate. These systems can analyze hundreds of variables, from fill patterns and chronic conditions to prior hospital visits and social determinants, helping predict who might skip medications or face complications.

One of the most powerful advantages of AI in medication adherence is its ability to go beyond generic risk scores and enable smart, nuanced segmentation. Traditional Medication Therapy Management programs often group patients by broad criteria like age or number of medications, but AI can uncover deeper patterns, identifying not just who is at risk, but why. It can segment patients by behavioral traits, risk profiles, and likely response to different types of outreach. That means more targeted support: seniors with mobility issues might receive in-home visits or blister packs, tech-savvy but forgetful patients could get app-based reminders, and younger patients struggling with costs might be connected to copay assistance or lower-cost alternatives.

When we understand what patients truly need, we can support them in ways that feel personal, not prescriptive.

Early studies of AI-driven interventions show promising results, with improvements in medication adherence ranging from 6.7% to 32.7% compared to standard practices or control groups, suggesting that personalized, scalable care may enhance patient engagement and outcomes.⁷ Another study of patients with hypertension found that segmenting individuals by psychological and behavioral worldviews predicted adherence more effectively than age, income, or concerns about side effects or cost. The most adherent group had rates over six times higher than the least adherent. When we know who falls into which segment, we can focus resources where they’ll matter most.⁸

The smartest analytics only matter if they lead to real-world outcomes, and early signs from AI-enabled adherence programs are encouraging. Medicare plans that have adopted personalized medication management strategies are seeing improved adherence rates and lower total costs.⁹ Addressing medication gaps early not only reduces complications and hospital visits but also drives meaningful savings across the system.

For health plans and Medication Therapy Management leaders, this impact goes beyond clinical benefits. Medication adherence plays a critical role in Medicare Star Ratings, with three of the most heavily weighted measures - diabetes, hypertension, and cholesterol - directly tied to adherence.² Improving these metrics through smarter, targeted interventions can elevate a plan from mid-tier status to top-tier ratings. That shift brings more than just prestige. It opens the door to higher bonus payments, stronger member retention, and a more competitive position in the market

Context Gaps

Medication adherence may seem routine, but its consequences are anything but. With AI-powered analytics, healthcare teams no longer have to wait until a patient lands in the emergency room or misses a critical refill. These tools uncover hidden risks, reveal behavior patterns, and enable timely interventions, often before complications arise. For leaders in health plans and pharmacy programs, adopting AI isn’t about chasing trends. It’s a strategic decision to improve outcomes and boost operational performance. When risks are detected early and support is personalized, non-adherence becomes something we can anticipate and prevent. Smarter segmentation and precise risk prediction lead to more targeted, effective Medication Therapy Management programs.