3/30/26

Validating the Impact of Cardio Algorithms

We have built this data platform, this longitudinal platform that connects the different multimodal data sources, especially within cardiology, to the EMR and to long-term longitudinal outcomes, where we can validate algorithms. There are many people who would like to understand how algorithms perform and what the impact is of utilizing an algorithm that they are not able to do so, because locally, even if they could, they maybe wouldn't be able to do so for a representative population. So we've partnered with the AHA to basically create a neutral AI assessment lab where people who are looking to implement an algorithm, such as a health care provider, can come to us and say, hey, we're considering this algorithm. The algorithm developers themselves can come to us and basically ask that the algorithm go through the assessment lab.


We've created a pipeline where they can basically send us the algorithm without us ever having to look at it or compromise their IP or anything like that. We run it on a curated dataset appropriate for their algorithm. And we produce a performance report that tells you about the algorithm's performance overall across different subgroups within different samples. And then we do an impact analysis, which is basically a study of the clinical and economic impact of implementing this algorithm in patient workflow or in clinical workflow. So if you were to actually take this algorithm and apply it in real life, what would have happened to the patients? Would we have seen reduced hospitalizations because you're treating them earlier, so they would have avoided adverse outcomes? What's the cost associated with that? Like, are their costs averted? Is there a positive ROI? 


And so we create this report that assesses the algorithm, assesses its potential impact. We can provide that to stakeholders such as healthcare providers, so that they can make an informed decision before they implement an algorithm within their clinical workflow and start impacting their patients' and providers' lives. So working with the AHA, they are connected to obviously many hospitals, many partners within the ecosystem, and can help with basically well-informed algorithm uptake in the cardiology space by directing people towards the assessment lab to get an independent assessment of an algorithm's value and its performance.


Previous

Building Dandelion's Multimodal Data Engine

Next

Model Trial Design to Improve Real-world Outcomes