These are a small subset of projects we have developed in collaboration with companies and researchers.
The emergency department is often the doorway to interactions with the health system. That means that the impact on revenue when patients leave without being seen extend far beyond the index event. This graph shows revenue before and after an index visit after adjusting for patients’ utilization characteristics. Patients who left without being seen are in blue and others are in orange.
We developed a change point model to identify shifts in claim denial rates for a broad array of visit types. This graph shows an uptick in denials for c-sections beginning in late June.
A deep learning model we developed on high-throughput molecular data is able to distinguish a systemic inflammatory response (red), from infection by bacteria (blue) and viruses (green).
We published work with a client demonstrating that their clinical decision support tool led to decreased total episode costs. Episode costs when providers were compliant with the tool’s recommendations are in red and non-compliant episodes are in green.
Genetic testing can identify patients who are unlikely to benefit from particular chemotherapeutics. The patients avoid unnecessary toxicity and dramatically reduce cost of care. We worked with a client to demonstrate the value of their genetic testing support for clinical decision making. The top line shows cost of care when an unnecessary chemotherapeutic is used and the bottom line shows without.
Panic disorder can lead to a complicated medical journey. We were able to develop a model to predict panic disorder based on historical claims that was used to target a psychosocial intervention.
We developed an interactive tool that allows live modifications of assumptions and immediately populates the consequences of those assumptions on return on investment estimates.
