The Limitations of Machine Learning in Medical Imaging

In this episode of This Much I Know, Managing Partner Carlos Espinal speaks with Emi Gal and Diego Cantor, co-founders of Ezra, about how digital healthcare will become. Every year, 20 million people are diagnosed with cancer, 10 million of which are detected late. Ezra's core mission aims to change this equation through their next-generation cancer screening technology that uses AI & MRIs. Carlos kicks the discussion off with an observation: healthcare solutions are now increasingly moving from reactive to proactive, as the ability to predict medical diagnoses ahead of time is growing. This statement translates well into some of Ezra's findings: while 13% of their scans suggest a potential cancer patient, a whopping 70% of their findings are clinical findings (ie. liver disease, aneurysms, etc.). The three also cover the balance between data privacy and medical progress within the field of medical imagery, as any kind of algorithm requires data to be trained on. Finally, Emi and Diego touch on their long-term aspirations for Ezra, and what it is going to take for the world to have equal access to such technologies. Tune in to this episode — it's a really good one.