Lone Star Health Group — a Texas-based network of 18 primary care clinics serving 120,000 patients — needed a HIPAA-compliant AI platform to reduce appointment no-shows, improve chronic disease care plan adherence, and automate prior authorization workflows that were consuming 2.8 hours of clinician time daily.
Lone Star Health Group operates 18 primary care clinics across the Dallas-Fort Worth metroplex, serving a patient population of 120,000 — with significant proportions of diabetic, hypertensive, and COPD patients requiring active chronic disease management. Their operational challenges were well-defined: a 22% appointment no-show rate costing an estimated $2.1M annually in lost revenue; prior authorization requests consuming 2.8 hours per clinician per day; and no systematic mechanism for identifying patients falling off their care plans before an emergency room visit demonstrated the gap.
In US healthcare, HIPAA compliance is non-negotiable and the technical requirements are specific: PHI must be encrypted at rest and in transit, access controls must be role-based with audit trails, Business Associate Agreements are required for every technology vendor, and any AI system making clinical-adjacent decisions must include robust human oversight mechanisms. Building a useful product within these constraints — at the quality level required for clinician adoption — is harder than building the same product in any other industry.
We designed a three-module platform: an AI patient engagement layer, a care gap intelligence system, and a prior authorization automation workflow.
XGBoost model trained on 3 years of appointment history — predicts no-show probability per appointment 72 hours out with 81% accuracy, triggering personalized outreach via SMS, voice, or portal notification based on patient preference.
Identifies patients overdue for preventive screenings, HbA1c checks, medication refills, and follow-up visits — generating a prioritized daily outreach list for care coordinators with AI-drafted outreach messages.
AI extracts clinical data from encounter notes, matches against payer formulary requirements, and auto-generates prior auth requests — reducing average completion time from 47 minutes to 8 minutes per request.
HIPAA-compliant conversational AI for patient portal — answers common questions about lab results, appointment scheduling, prescription refills, and referrals, with immediate escalation to clinical staff for clinical questions.
Clinic-level and network-level view of care gap closure rates, chronic disease control metrics (average HbA1c by clinic, blood pressure control rates), and financial impact of no-show reduction.
BAA-ready architecture: AES-256 encryption, role-based access with multi-factor authentication, complete PHI audit logs, automatic session timeout, and de-identification pipeline for analytics.
The prior authorization automation alone would have justified the entire investment. But watching the no-show prediction model work in real time — and seeing our care coordinators reach patients who we would never have called otherwise — that is when I realized this platform changes how primary care operates.
Dr. Jennifer Walsh Chief Medical Officer · Lone Star Health Group