Jan 16, 2019 · 21m · mad
Building An Info Layer for Personal Health // Allon Bloch, K Health (FirstMark's Data Driven NYC)
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In a presentation at FirstMark's Data Driven NYC, K Health Co-Founder and CEO Allon Bloch introduces how machine learning and a massive longitudinal clinical dataset are transforming personal healthcare through AI-powered diagnostic tools. He demonstrates K Health's consumer application and outlines a vision for software-driven primary care that delivers accurate, personalized medical information at a fraction of traditional costs.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 4.7% of the talking time here. How this is scored →
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Bloch rejects the host's underlying premise regarding accuracy concerns by pointing out that traditional doctors do not even track their own accuracy rates and 60 million people currently rely on flawed WebMD searches anyway.
Hardest push from Matt ▶ 14:44 Matt Turck presses on AI liability and diagnostic errorsTurck directly challenges Bloch on the inherent fallibility of machine learning models and asks how the company handles responsibility when incorrect diagnostic analysis occurs.
Biggest teaching moment ▶ 7:45 Explaining differential diagnosis and the 'rule out' conceptBloch educates the audience on how clinical diagnostic thinking works, using the medical adage of ruling out common horses before rare zebras.
Matt holds his own ▶ 14:44 Matt Turck highlights machine learning failure modes in medicineTurck demonstrates domain understanding of machine learning limitations, noting that while ML excels at processing massive data, its lack of 100% reliability poses unique dangers in clinical diagnostics.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Data Silos and the Pitfalls of Online Search | 0 | 5 | 1 | 0 | This is a solo presentation segment without any host involvement. Allon Bloch highlights the flawed nature of searching symptoms on WebMD/Google and the absurdly inflated cost of US healthcare compared to military spending. | |
| The Core Concept: 'You Are Not Patient Zero' | 0 | 5 | 1 | 0 | The presentation continues without host participation. Bloch outlines K Health's core premise that 'you are not patient zero' and details licensing 20 years of unstructured data and 80 million doctor notes from an Israeli HMO. | |
| K Health Product Architecture and Care Model | 0 | 6 | 1 | 0 | Bloch conducts an extended app demo during his talk. He educates the audience on medical differential diagnosis and the 'rule out' principle using the medical metaphor of horses versus zebras. | |
| Automated Follow-Up and Life-Saving Case Study | 0 | 6 | 1 | 0 | Bloch wraps up his keynote presentation with a real-life user case study where K Health's follow-up protocol caught a ruptured appendicitis and saved a patient's life. | |
| Audience Q&A and Strategic Discussion | 5 | 6 | 4 | 6 | Host Matt Turck opens Q&A by challenging machine learning's reliability and liability risks in healthcare. Bloch pushes back on the premise, noting traditional doctors do not measure accuracy either and WebMD is far worse. |