Podcast Summary: Think AI Podcast

Episode 2 – When Data Meets Health: AI in Healthcare That Actually Works

Host: Dave Goyal
Date: March 11, 2026

Episode Overview

This episode focuses on how AI and data strategies are transforming healthcare – not with flashy robotics or science fiction, but with practical, powerful changes to the data-driven “back office” functions that keep healthcare running. Host Dave Goyal, drawing upon 15+ years in healthcare IT and data, shares his experiences with how connecting, cleaning, and leveraging healthcare data can be the difference between organizational chaos and clinical excellence. The discussion moves from foundational data issues to real-world applications of AI, highlighting proven use cases and practical advice for AI-curious, skeptic, and enthusiast listeners.

Key Discussion Points & Insights

1. The Cost of Bad Data in Healthcare (00:00–03:30)

• Opening Story: Dave shares eye-opening stats about how U.S. hospitals lose over $260 million annually due to billing errors and data failures, not malpractice or lawsuits.
  • “A hospital loses money every single day. Not because of bad nurses, because of bad data.” (00:00)
• Impact: Errors in claims, missing information, or disconnected systems stall payments and hurt patient care.

2. Dave’s Career Beginnings: Problem-Solving with Data (03:31–09:30)

• Career Lessons: Early consulting in a global medical device company exposed Dave to data silos and the need for connected information.
  • “If you show up and solve problems, not just the ones they hired you for, but the ones nobody else wants to touch, doors open.” (04:40)
• Healthcare's Unique Stakes: Inaccurate data in healthcare means risk to patients’ lives – not just lost revenue.

3. Building Claim Systems from Scratch – A Data First Approach (09:31–15:30)

• The Project: Designing a claims operation from the ground up for a medical device company.
• Process Before Tech: Dave emphasizes starting with mapping the business process, then designing the system:
  • Automating data intake for claims, flagging errors instantly.
  • Creating real-time dashboards showing claims status and root causes for denials.
  • Automating denial workflows based on value and likelihood of recovery.
• Results: The company jumped from zero to $100 million/year in revenue by processing 100,000 claims per month – all due to data strategy and not "just" technology.
  • “Most companies do it backwards. They buy the software first and then they try to fit their process into it. That’s why most projects fail. We didn’t fail.” (15:02)

4. The Core Problem: Data Strategy, Not Just Technology (15:31–20:30)

• Data Overload, but Underutilized: Typical healthcare organizations have 15–20 different systems (EHR, billing, scheduling, HR, supply chain, quality metrics) – data is there, but not connected.
• Four Steps to Building Data Strategy:
  1. Assessment & Audit: Inventory data, usage, and decision gaps.
  2. Mapping Data Landscape: Document systems, flows, and breakpoints.
  3. Defining Architecture: Build warehouses/data lakes as single sources of truth.
  4. Analytics: Leverage clean, connected data to answer operational and clinical questions.
  • “It turns raw information into answers and then answers into action.” (20:20)

5. Non-Profit Hospitals & Data Challenges (20:31–27:00)

• Unique Constraints: Non-profits serve the most vulnerable but have tight budgets and fewer resources.
• Unlocking EHR Data: Dave describes building a "patient intelligence layer" on top of the EHR, enabling actionable insights for care teams.
• Key Use Cases:
  1. Chronic Disease Management: Dashboards identify patients with chronic conditions who need outreach.
  2. Emergency Department Utilization: Analyze patterns among frequent ER users to coordinate proactive interventions, improving outcomes and reducing costs.
  3. Quality Reporting and Compliance: Automated reports for regulatory compliance now take days instead of weeks.
• Quote:
  • “These are life and death questions. And the data to answer them exist… but it was locked, buried, inaccessible.” (23:12)

6. AI in Healthcare: Separating Hype from Reality (27:01–37:20)

• Moving to AI: After years in traditional data, Dave pursued formal study in AI and healthcare at Harvard to bridge real needs with advanced tech.
• Key Point: AI’s biggest value is not replacing clinicians or robotics, but improving back office processes.
• Real, Practical Use Cases (with examples and results):
  • AI-Powered Claims Management: Predict denials before submission, reducing refusal rates by 20–30%.
  • Automated Medical Coding: AI reads clinical notes, assigns first-pass codes, with humans as final reviewers, saving up to $500K in coding costs.
  • Prior Authorization Automation: Streamlines paperwork, reducing 40-minute tasks to 5 minutes, freeing nurses for patient care.
  • Predicting Patient No-Shows: Smart scheduling to reduce no-shows by 25%.
  • Readmission Risk Prediction: Stratify discharge risk to drive proactive follow-up.
  • Intelligent Document Processing: OCR and NLP turn scanned images and faxes into structured data automatically.
• “Six use cases, all back office. None of them involve replacing a single clinician.” (36:44)

7. Privacy & Governance: The Real Risk (37:21–40:10)

• HIPAA is Non-Negotiable: AI systems in healthcare must comply with privacy law, with robust encryption, access controls, and audit trails.
• Main Message: The real risk isn’t AI, but fragmentary, insecure manual systems.
  • “If you deploy AI with proper data governance… you actually make patient data safer. Not less safe, but safer.” (39:15)
• Advice to Skeptics: Demand compliance but don’t let fear block overdue improvements.

Notable Quotes & Memorable Moments

| Timestamp | Speaker | Quote | |-----------|---------|-------| | 00:00 | Dave | “A hospital loses money every single day. Not because of bad nurses, because of bad data.” | | 04:40 | Dave | “If you show up and solve problems, not just the ones they hired you for, but the ones nobody else wants to touch, doors open.” | | 15:02 | Dave | “Most companies do it backwards. They buy the software first and then they try to fit their process into it. That’s why most projects fail. We didn’t fail.” | | 23:12 | Dave | “These are life and death questions. And the data to answer them exist… but it was locked, buried, inaccessible.” | | 36:44 | Dave | “Six use cases, all back office. None of them involve replacing a single clinician.” | | 39:15 | Dave | “If you deploy AI with proper data governance… you actually make patient data safer. Not less safe, but safer.” |

Timestamps for Key Segments

• 00:00–03:30 — The Cost of Bad Data in Healthcare
• 03:31–09:30 — Dave’s Early Lessons in Healthcare Data
• 09:31–15:30 — Building Claims Systems: A Process-First Approach
• 15:31–20:30 — Data Strategies in Healthcare: Building a Unified Approach
• 20:31–27:00 — Non-Profit Hospital Stories & Impactful Use Cases
• 27:01–37:20 — How AI Actually Works in Healthcare (6 Back Office Use Cases)
• 37:21–40:10 — Privacy, Security & The Real Risk of AI

AI Tip of the Day (40:11–42:30)

• Main Advice: Instantly summarize complex documents using AI. Steps:
  1. Upload to any AI tool (ChatGPT, Claude, Gemini, etc.).
  2. Prompt: “Summarize this document in plain English. Give me five most important points. For each point, tell me what action I should take. Keep your language simple, no jargon.”
  3. Use follow-up prompts to deepen your insights (“What are the three biggest risks?”, etc.).
• Encouragement: Try it for two minutes with any document in your workflow – skeptics should read both original and summary for comparison.

Final Thoughts

• Closing Tone: Dave emphasizes action – “Take one idea from this episode and turn it into action.”
• Preview: Next episode will shift to manufacturing and its data challenges, with lessons that echo this episode’s themes of process-led innovation.
• Call to Action: Listeners are encouraged to subscribe and share, especially with healthcare executives who need to hear how data and AI can be transformational.

Flow & Usefulness

This episode is packed with real-life healthcare IT stories, clear implementation steps, and cautionary notes about data and AI done right vs. wrong. Dave’s examples are tactical and actionable for healthcare professionals but also relevant for any organization struggling to harness its data. The tone is pragmatic, demystifying AI by focusing on measurable process improvements rather than far-flung automation dreams.

For even more context or to go deeper on any segment, refer to the listed timestamps above for the specific discussions!