AI Model Wiki

    Skip to main content
    20 min
    Lesson 4 of 6

    Real-World Case Studies in AI Pharmacy Practice

    From the course AI for Pharmacists

    How pharmacy systems and health networks have deployed AI — and the patient safety and efficiency outcomes documented.

    Kaiser Permanente: AI Drug Interaction Alerts at Scale

    Kaiser Permanente operates one of the largest integrated health systems in the US, with pharmacists reviewing over 50 million prescriptions annually. The system deployed an AI-enhanced drug interaction checking tool that goes beyond standard drug database lookups to incorporate patient-specific factors: renal function, age, concurrent diagnoses, and genetic test results.

    Outcomes:

    • Clinically significant alerts (those requiring pharmacist action) increased from 12% to 31% of all alerts — a 158% improvement in alert specificity
    • Alert fatigue dropped: pharmacists overrode fewer alerts, and the override rate fell from 94% to 71%
    • Adverse drug event rates in enrolled patients fell by 22% over 18 months

    Key lesson: Reducing false-positive alerts is often more impactful than adding new alerts. AI that filters noise produces better safety outcomes than AI that generates more warnings.

    CVS Health: AI-Powered Medication Adherence Outreach

    CVS deployed an AI adherence platform that analyzes refill patterns, medication type, patient demographics, and past outreach responses to predict which patients are most likely to discontinue therapy and what intervention (phone call, text, pharmacist consultation) will be most effective.

    Outcomes:

    • Adherence improvement of 8.3 percentage points for chronic disease medications (diabetes, hypertension, cholesterol)
    • Cost of adherence outreach reduced by 34% through AI-targeted interventions vs. blanket outreach programs
    • Pharmacist time spent on routine adherence calls reduced by 40%, freeing capacity for clinical consultations

    Key lesson: AI prediction of both non-adherence risk AND optimal intervention type is more valuable than prediction alone — the action recommendation is where ROI is realized.

    NHS England: AI Formulary Optimization

    NHS Clinical Commissioning Groups piloted AI-driven formulary analysis to identify opportunities for cost-effective prescribing changes within therapeutic equivalence guidelines.

    Outcomes:

    • Identified £14M in potential annual savings across a region of 1.2M patients
    • Prescribing change implementation rate of 68% when AI recommendations included clinical evidence summaries
    • Time for formulary review reduced from 6 months to 6 weeks

    Key lesson: AI recommendations achieve higher adoption when they come packaged with the clinical evidence rationale, not just the cost or efficiency data.

    Patterns Across All Three

    • AI that reduces alert fatigue (specificity) outperforms AI that adds more alerts (sensitivity)
    • Combining prediction with intervention recommendation delivers 2-3x more ROI than prediction alone
    • Evidence packaging drives clinician adoption of AI recommendations

    Key Takeaways

    • •Kaiser's patient-specific AI alerts cut override rate from 94% to 71% and reduced adverse events by 22%
    • •CVS adherence AI improved chronic disease adherence 8.3 points and cut outreach costs 34%
    • •NHS formulary AI cut review time from 6 months to 6 weeks and identified £14M in savings
    • •AI recommendations achieve higher adoption when packaged with supporting clinical evidence
    0 of 6 lessons marked complete

    Progress is kept in this browser only.