Arjun Chakraborty

Surgical site infections (SSI), infections at the surgical site that occur after surgery, impact more than a hundred thousand patients a year in the United States. They increase the risk of death after surgery, lead to complications like cellulitis and sepsis, and incur significant healthcare costs. Surveillance of SSI can guide interventions to reduce SSI rates. The current mainstay of SSI surveillance is manual chart review, which is expensive and time consuming. Automated surveillance systems addressing these drawbacks typically rely on a limited number of data modalities from the electronic health record (EHR). They predominantly use rule-based approaches or conventional machine learning algorithms to retrospectively predict whether a surgical case resulted in an SSI. This limits the performance and domain adaptation capability of published gold standard automated surveillance systems.
In contrast to past published gold standard automated surveillance approaches, we used a data-driven deep learning framework using structured data, clinical text data, and temporal information from the EHR of surgical cases to build an automated surveillance system. Our main findings were:
1. We found that a purely data-driven deep learning approach using multimodal data can outperform previous published gold standard rule-based or conventional machine learning-based approaches on the task of SSI prediction.
2. We found it was possible to build models that could domain adapt to a diverse set of domains. Using the data representation and modeling strategies above.
3. Large language models, specifically generalist foundation models (e.g., Llama 3) can be used to offer previously unrealized gains on performance.