Several key initiatives are currently underway resulting in NIH funding and published research.
Researchers are investigating the long-term sequelae of patients diagnosed with COVID-19. For instance, it was observed that over 20 percent of hospitalized patients with COVID-19 developed acute kidney injury (AKI), which was highly associated with reduced survival in these patients. While a proportion of these patients recovered from AKI, a significant proportion exhibited partial recovery and/or eventual chronic kidney disease. The Stony Brook Medicine COVID-19 Data Commons enabled researchers to identify these patients that are at risk of developing long-term sequelae from AKI in the setting of COVID-19, leading to a creation of a post-AKI clinic at Stony Brook Medicine. These efforts have now also extended to N3C-CD2H multi-center collaborative effort through the NIH (AKI subgroup co-led by Stony Brook investigators Mallipattu and Moffitt). Two external grant applications have been submitted based on preliminary analysis of AKI in the Stony Brook cohort, including an NIH R01.
Given the numbers of patients who developed severe illness requiring ICU care, the Stony Brook Medicine COVID-19 Data Commons is being used to investigate risk factors, outcomes and therapeutics. The power of collaborative data will help derive meaningful interventions. For example, few centers had large numbers of mechanically ventilated patients receiving rescue therapies, such as airway pressure release ventilation, nitric oxide and extracorporeal membrane oxygenation. The multi-center collaborative effort will provide actionable data as to whether these resource-intensive interventions improve outcomes. If they are efficacious, then the analysis will help guide the development of treatment protocols. Other elements being evaluated include the utility of prone positioning, effects of various medications including steroids, antivirals, and biologic agents, and complications, among others.
Stony Brook Medicine has tremendous strength in imaging informatics — several research groups are using Data Commons images in their research. Dr. Prateek Prasanna’s lab in BMI has demonstrated the potential for computerized analysis of pre-treatment chest radiographs, via radiomics and machine learning techniques, in predicting disease outcomes in patients with COVID-19 (under review). Machine learning classifiers to predict mechanical ventilation requirement and mortality were trained and evaluated on 353 baseline radiographs taken from patients who were COVID-19 positive. Two external grant applications have been submitted based on preliminary imaging analysis in the Stony Brook cohort.
The Cancer Imaging Archive (TCIA) de-identifies and hosts cancer images for public access. As part of a recent initiative, TCIA is now hosting COVID-19 imaging data. Stony Brook University is working closely with TCIA to provide over 1000 de-identified imaging studies and associated de-identified clinical data — a subset from the Stony Brook Medicine COVID-19 Data Commons.
Supporting High-Throughput COVID-19 Test Platform Validation
The Data Commons was instrumental in quickly identifying asymptomatic patients who were COVID-19 positive for testing of a new high-throughput COVID-19 testing platform. Using the manually extracted REDCap portion of the Data Commons, patients without reported symptoms on presentation were identified. Using the extracted discrete data elements, those identified patients with a COVID-19 swab performed at Stony Brook manually underwent chart review to determine the true asymptomatic cases. These pathology samples will be used for validation of the new laboratory testing. Because only a very small fraction of the Stony Brook patients with positive test results were asymptomatic, without the Stony Brook Medicine COVID-19 Data Commons, it would have been a tremendous challenge to identify clinically asymptomatic cases.