How AI and machine learning are enabling early disease detection
It’s a simple enough premise: the sooner you can diagnose a disease, the more quickly you can treat it, ideally while saving money and improving the life of the patient in the process. In fact, according to Dana-Farber Cancer Institute, early detection of several cancers can lead to better outcomes, like overall survival1. For example, patients diagnosed with cervical cancer have an overall 5-year survival rate of 66%. When you compare patients diagnosed early (Stage I, 5-year survival ~92%) to those who are diagnosed with late-stage (Stage 4, 5-year survival <20%), it’s clear to see why early diagnosis matters2. But cancer isn’t the only disease where early detection could make a dramatic improvement for a patient. Some autoimmune and mitochondrial disorders are notoriously difficult to diagnose, often requiring the patient to undergo a ‘diagnostic odyssey’ of many years and countless doctors before they get the right diagnosis and proper treatment3, even though early diagnosis for diseases like rheumatoid arthritis can mean less joint damage in the future.
In MoneyBall Medicine: Thriving in the New Data-Driven Healthcare Market, I wrote about the ongoing deluge of data and technological revolution impacting healthcare today. Every component of a doctor appointment, lab test, or surgery can be boiled down to a disease classification, a discrete biomarker, or procedure code. But the value of these disparate data has been challenging to quantify and the multifactorial nature of data from a variety of places can be nearly impossible for humans to make sense of and use in clinically meaningful ways.
To Prognos, it’s this confluence of data and technology, including artificial intelligence and machine learning systems, that is bringing the promise of early detection to patients and doctors. Prognos has an ambitious mission to eradicate disease by detecting it earlier and earlier—ultimately, before the patient exhibits any symptoms or even develops the disease in the first place. Imagine being able to predict which patients are likely to develop diabetes or heart disease five, ten, or more years before their blood sugar starts to rise or their cholesterol reaches dangerous levels. You could target these patients with interventions, like diet, exercise, or pharmaceuticals, helping the patients live longer, healthier lives.
To make predictive analytics like I’ve described accurate and reliable takes massive amounts of data—like the billions of de-identified records Prognos has on more than 180 million patients. What’s more, Prognos has been able to articulate the value of these data points to a variety of partners in the diagnostics, life sciences, and payer industries. In this 3rd episode of the MoneyBall Medicine Podcast, I sit down with Prognos co-founder and Chief Medical Officer Jason Bahn to discuss how his company is leveraging the power of artificial intelligence and predictive analytics to enable early disease detection. We talk about the early challenges his company faced with a business model based on the value of data and where he sees AI in healthcare moving in the future. Join us for another episode of the MoneyBall Medicine podcast!