Kempton Presley on Building the AI Data Foundation for Medication Adherence — AdhereHealth | VIVE 2026

Key Takeaways

• Medication non‑adherence causes ~125,000 deaths annually
• Non‑adherence costs US healthcare up to $300 B each year
• AdhereHealth unifies claims and clinical feeds into a single clean dataset
• Robust data prevents AI hallucinations and unsafe clinical recommendations
• Tech‑enabled clinicians intervene only for patients identified as high‑risk

Pulse Analysis

The push toward AI in health care often focuses on flashy large language models, yet the true bottleneck remains data quality. In the United States, medication non‑adherence accounts for roughly 125,000 deaths and $300 billion in wasted spending each year, according to NCQA. These figures underscore the urgency of a reliable data pipeline: fragmented claims, delayed lab results, and inconsistent electronic health record formats create a "dirty" data environment where AI can misinterpret signals, leading to costly mis‑triage or even patient harm. Companies that prioritize rigorous data ingestion, cleansing, and normalization—like AdhereHealth—lay the groundwork for trustworthy AI interventions.

AdhereHealth’s strategy blends advanced data science with a tech‑enabled clinical services team, turning raw, siloed information into a unified architecture that flags patients at risk of medication non‑adherence. By automating best‑practice protocols rather than existing bad behaviors, the platform ensures that any downstream generative AI operates on evidence‑based pathways. This approach not only mitigates the risk of AI hallucinations but also aligns with the shift toward value‑based reimbursement, where providers are financially incentivized to close care gaps and improve outcomes.

As generative AI moves from hype to operational reality, the competitive advantage will belong to firms that have already invested in data integrity. A solid data foundation enables scalable, real‑time decision support, reduces manual chart reviews, and supports precise clinician outreach. For health plans and risk‑bearing providers, this translates into lower readmission rates, higher medication adherence, and ultimately, better financial performance under value‑based contracts. AdhereHealth’s model illustrates how data‑first thinking can turn AI from a risky experiment into a reliable engine for population health improvement.