The prostate-specific antigen (PSA) test is the front-line clinical test for telling men whether they likely have prostate cancer. Yet it isn’t always accurate. In fact, roughly 75% of men with elevated levels of PSA do not have cancer (what’s known as a false positive). But even more dangerously, sometimes men without elevated PSA levels do have cancer (what’s known as a false negative).
Just this past week, the actor Ben Stiller shared his own experience with the PSA test, calling it “the prostate cancer test that saved my life” because his doctor gave him a test to establish his baseline PSA level when he was 46 and then repeated tests over time. This time-series data — even over a short period — helped detect and treat Stiller’s cancer early as his doctors observed rising PSA levels.
While any given test may have its share of false positives and negatives and other flaws for predicting disease in a one-off fashion, they can be made extremely predictive when considered over time. See for example recent studies from women’s health, which have shown that screening at multiple points over time identifies ovarian cancer earlier.
But time-series data is just part of the answer. The greater opportunity here lies in bringing together and combining various longitudinal measurements much more broadly. And that’s where Q comes in. By aggregating several different existing, well-established technologies and tests over time and using computation to better interpret data, Q can greatly increase the effective accuracy of these tests. This in turn dramatically increases our ability to catch disease early, early enough to treat.
How is this possible? As healthcare becomes more of an information science, computation and data can play a more significant role. No human being could effectively combine a flood of inputs from many disparate tests. But software can!
Moreover, in addition to broader testing for baseline and trends, software can leverage data from existing tests to yield other untapped insights. For example, with the low cost of genomics — because bio has its own Moore’s Law — one could use genetic material to better interpret PSA tests. Recent results show how knowing your “snips” (SNPs or single nucleotide polymorphisms) in DNA would lead to a much more predictive interpretation of PSA.
With these approaches, the tri-fold axes of multiple tests, time, and population can all be brought to bear in a unified way. Genomics would give insight into how to better read tests for a given patient and understand the specific challenges they face. Combine that with time series data and we get even greater accuracies. And finally, as more patients participate, one could predict even subtle aspects earlier with higher accuracy due to aggregating patient experience and outcomes from millions of people.
The approach that Q is taking can revolutionize healthcare by bringing the best of technology and computer science to the healthcare science and practices we already have in place. The co-founders themselves embody this combination of skills: serial entrepreneur and computer scientist Jeffrey Kaditz most recently worked to bring transparency in consumer finance; Michael Snyder is the genetics chair at Stanford and Director of Personalized Medicine who also pioneered digitization of human biofluids; and chief medical officer Garry Choy is an MD, MBA, and practicing clinical radiologist.
I’m delighted to announce our investment in Q’s recent funding round and excited to partner with this team to help bring their vision to fruition.