From Data to Dollars: Improving Medical Billing Accuracy Using NLP
summary
A hospital team builds an automated SQL-to-R pipeline that pulls free‑text notes via odbc and stringr, applies a keyword list plus regex to extract billing minutes, and flags outliers under 10 or over 100 minutes. The tool processes ~285,000 critical‑care notes and correctly captures billing data for 93% of over 250,000 notes versus 40–60% from vendor SmartPhrases, and it runs monthly with manual review for edge cases.
AI-written summary · about
chapters (14)
- 0:00Medical billing challenges
- 1:46Clinician billing issues
- 2:54SmartPhrase and LLM limits
- 4:20AI constraints in healthcare
- 6:30Custom NLP tool from free text
- 7:56Automating extraction with R and SQL
- 8:18Keyword dataset for billing phrases
- 9:01NLP function for time extraction
- 10:28Detecting billing errors
- 11:11NLP vs SmartPhrase validation
- 12:13Monthly review and feedback
- 13:41Expanding NLP to specialties
- 14:27Future analytics and refinement
- 15:16Q&A on HIPAA and improvements
related talks
11mEnsuring Reliability in Healthcare Data Engineering
41mAccelerating privacy-enhancing data processing
05mHow generative AI can support medical staff
31mBig data for tiny patients: Applying ML to pediatrics
21mTransforming Healthcare Through Data-Driven Intelligence
26mNatural Language Processing for Healthcare: Current Trends...
52mBig Data - Top Ingredient to Improve Healthcare
15mTransforming Healthcare with BI & AI