This paper published in Nature Magazine last week – and the on-point commentary by Eric Topol, MD and Sandeep Burugupalli – is a great example of how life sciences companies can advance precision medicine outside of oncology. The study used ~46k healthy patients who had at least five CBCs (the most common blood test panel) measured over a 20 year period. It shows that individuals have setpoints that are quite consistent over time, and that deviations from these baseline values are predictive of a number of common conditions. The study prompted me to take a look at Dandelion Health's data to see how we could contribute to this research. It turns out we have 434k patients with at least 5 CBCs across at least 5 years. We also have associated longitudinal data from these patients – other labs, diagnoses, ECG waveforms, radiology images & reports, pathology reports, etc. Who wants to pull the thread on this insight with us? We can potentially use this line of thinking to: - Enrich clinical trial cohorts - Structure preventive clinical trials - Explore the use of deviations from CBC setpoints as companion diagnostics Drop me a DM if you’re interested in discussing. Congrats to the authors on a great paper: Brody Foy John M. Higgins Veronica Tozzo Rachel Petherbridge Maxwell Roth Robert Hasserjian Camille Powe and others. Article link: https://lnkd.in/eEEG-BNM Topol commentary: http://bit.ly/3ZxVVqs Burugupalli commentary: https://bit.ly/4fn4LgA
Great work Sandeep!
Healthcare AI R&D, Strategy, Operation, GTM & Regulatory, Neuroradiologist, AI MedTech Founder, Advisor, Board Member, Investor Stanford MS, Healthcare Management, Biomedical Informatics and AI, Inventor
1wFascinating research potential here, Nicholas. Building on your ideas, these could be research directions: 1. Multi-modal correlation studies: Investigate how CBC setpoint deviations correlate with ECG changes and imaging findings across your 434k patient dataset. This could uncover novel early warning signals for disease progression. 2. Population stratification: Analyze how CBC baselines cluster across different demographic groups to identify subpopulations that might benefit from personalized reference ranges. 3. Machine learning applications: Develop predictive models that combine CBC trajectory analysis with your other longitudinal data to forecast disease risk and treatment response. I look forward to seeing what you come up with!