Ca-ssis/istock via Getty Images

Feature

AI helps docs, patients tackle medication adherence challenges

Healthcare organizations are using AI tools to cut down on patient resistance regarding treatment plans and improve medication compliance.

By
  • Brian T. Horowitz
Published: 31 Jul 2025

From drug prices to cognitive issues, there are many reasons why patients don't take their meds. Up to half of all patients decide not to follow prescribed treatment plans for chronic conditions such as diabetes or hypertension, according to a study published in The Permanente Journal.

Cost is often the biggest barrier. Sometimes patients may need to decide between paying their rent or mortgage and putting their food on the table and purchasing their medication, said Bryan Hill, chief technology officer and vice president of digital health and innovation in life sciences at IT services provider Cognizant.

Not only do patients consider the cost of care, but medication nonadherence results in about $100 billion to $300 billion in unnecessary health costs in the United States. About 20% to 30% of new prescriptions go unfilled, according to the Centers for Disease Control and Prevention.

Artificial intelligence tools, including AI agents, can ease the friction keeping patients from taking their meds and help them better manage chronic and complex treatment plans, said Hill.

Nonadherence can be intentional and nonintentional. Nonintentional adherence occurs when patients cannot remember whether they have taken their medication, or if they have a condition like dementia. However, with intentional medication nonadherence, patients make a conscious decision not to follow their doctors' orders to take medication, Hill explained.

People often stop taking medications after 30 days, Hill noted. Patients may stop because of concerns around side effects such as weight gain, Hill said. Or they may discontinue medication due to a lack of improvement. He noted that some medications like those for bipolar disorder may take several weeks to kick in, and patients may stop early.

Other causes include too many medications, a lack of symptoms, mistrust, worry and depression, according to the American Medical Association.

How agentic AI can improve medication adherence

AI allows providers to synthesize a lot of data points to study whether drugs will work, which can help with patient nonadherence. An AI interface could recognize a pattern of behavior and send reminders on a wearable device like the Apple Watch, according to Hill.

"I think the key is trying to steer someone's behavior in a way that they don't realize it's happening and that's the most supportive to them," Hill said.

He envisions AI agents being able to identify a behavior such as not taking meds or a lack of exercise and then triggering a text message to a family member, he said. However, these capabilities are in the early stages, according to Hill.

"I think we're starting to see them emerge," Hill said. "I think it's still early innings."

Although rudimentary examples of dosage reminders, like through text messages, exists today, he sees an opportunity to integrate data with electronic health records and payer data.

"Creating that bidirectional communication, that integration, is really the key to unlocking this," Hill said.

He also sees the potential for pharma to use AI to tackle patient friction and medication nonadherence. With pharma companies developing e-commerce platforms so patients can order medication directly from the manufacturer, they are laying the groundwork for an infrastructure that will enable AI-driven engagement, Hill suggested.

"You've got the financial transaction and supply chain, you've got tools to support to everyday living and you've got the drug delivery mechanism, which may be an injectable, and that gives you an opportunity to then track via the device whether or not somebody took their medication," Hill said. With companies collecting data within their own ecosystem, they have a "more complete set of frictionless data," he said.

Agentic AI apps emerge to help patients take meds

Hill highlighted four medication adherence tools that incorporate agentic AI capabilities to allow physicians to make decisions and adapt outreach based on patient data and behavior.

Epic's MyChart personal health portal now incorporates agentic AI and GPT-4 technology to allow doctors to send messages composed by an AI bot. Conversational agents initiate interactions and can suggest actions according to clinical and historical data.

The EveryDose cloud-based AI platform incorporates personalized reminders, adaptive analytics and autonomous care coordination. Its Intelligent Medication Cloud allows patients, care teams and enterprises to optimize medication management. The platform integrates with Epic on FHIR, Athenahealth, Veradigm (formerly Allscripts), NexGen and HL7 FHIR.

Meanwhile, Medisafe's Digital Drug Companion uses an inventory of notifications and reminders as well as guidance based on personas to improve patient adherence. Its Just-in-Time-Interventions platform uses behavioral science and a comprehensive data set along with machine learning to interact with patients and keep them on track with treatment. The company's Voice Intelligent Agent integrates across voice, SMS, app and the web to improve adherence.

Another company called AllazoHealth uses predictive modeling to anticipate medication behaviors. The technology allows physicians to learn from patient data, which enables them to personalize and automate outreach. The tool integrates with customer relationship management and hub service platforms using secure APIs.

Building more connectivity

As care becomes more personalized, more drug adherence solutions will emerge, according to Hill.

"I think these systems will have that added benefit of seeing these patterns, seeing the biological and behavioral activity within the individual patient," Hill said.

IT Infrastructure will need to support more connectivity and break down silos for AI-enabled solutions to be able to improve medication adherence, he added.

Hill suggested considering technologies that help with EHR and HIE integrations.

"For an IT leader, build for connectivity first, because if the data isn't flowing, if it's not connected to the UX, the user experience isn't going to matter for the patient," Hill said. "So that infrastructure has got to be there."

Brian T. Horowitz started covering health IT news in 2010 and the tech beat overall in 1996.

Dig Deeper on Artificial intelligence in healthcare