Predict patient appointment no-show probability using multi-factor risk modeling incorporating historical patterns, social determinants, behavioral signals, and operational context to enable targeted intervention strategies.
This skill builds and applies a multi-factor predictive model for patient appointment no-shows. It integrates historical attendance patterns, social determinants of health (SDoH), scheduling characteristics, and behavioral signals to generate per-appointment risk scores. The model enables targeted outreach interventions (reminder calls, transportation assistance, telehealth conversion) to reduce no-show rates while maintaining equity across patient populations. National benchmarks indicate average no-show rates of 18-23% across ambulatory settings, with significant variation by specialty, payer, and demographics.
| Input | Description | Format |
|---|---|---|
appointment_history | Historical records with status (kept, no-show, cancelled, rescheduled) | De-identified CSV/JSON |
patient_features | Demographics, insurance, zip code, preferred language, portal status | JSON array |
scheduling_features | Lead time, day of week, time of day, appointment type, provider, location | JSON object |
visit_history | Prior visit frequency, last visit date, no-show history count | JSON object |
sdoh_indicators | Area-level deprivation index (ADI), transportation access, broadband access | JSON object |
weather_data | Historical weather conditions for appointment dates (optional) | JSON object |
reminder_history | Prior reminder contacts (type, timing, response) | JSON array |
no_show_prediction:
model_performance:
auc_roc: number
precision_at_30pct_threshold: number
recall_at_30pct_threshold: number
calibration_error: number
fairness_metrics:
- group: string
fpr: number
fnr: number
appointment_scores:
- appointment_id: string
predicted_probability: number
risk_tier: string
top_risk_factors:
- factor: string
contribution: number
recommended_intervention: string
session_summary:
- session_date: date
provider_id: string
scheduled_count: number
predicted_no_shows: number
recommended_overbooks: number
population_insights:
overall_predicted_rate: number
rate_by_specialty: object
rate_by_insurance: object
rate_by_adi_quintile: object
top_modifiable_risk_factors: array
Apply a Predict-Stratify-Intervene-Measure cycle:
Feature importance analysis should use SHAP (SHapley Additive exPlanations) values to ensure model transparency and enable clinically meaningful risk factor communication.
Example: Multi-Specialty Ambulatory Network