Skip to main content

IOMICS Healthcare


Daniel Corkill, PhD, Chief Science Officer


IOMICS Healthcare


Life Sciences and Healthcare

Business Challenge

COVID-19 presented enormous challenges to organizations and society in 2020. A life-threatening complication of COVID-19 is Acute Respiratory Distress Syndrome (ARDS). ARDS is the rapid onset of widespread lung inflammation, accumulation of fluid, and an extensive immune response that compromises breathing and leads to cascading problems, permanent damage and sometimes death. Earlier detection of the impending onset of ARDS would allow clinicians to intervene more effectively. The challenge is monitoring patients for subtle patterns that indicate that ARDS is likely to develop and alerting clinicians with a very low false-positive rate. Premature or unnecessary ARDS interventions increase resource usage, staff risk and patient comorbidities, and reduce a patient’s likelihood of survival.

Analytics Solution

In June 2020, IOMICS Healthcare began developing an advanced warning system for ARDS onset in intensive care (ICU) patients in collaboration with a New York City hospital consortium. A significant innovation was using our data-engineering capabilities to transform observational patient data that also included when ARDS had been clinically diagnosed to train models that predict ARDS development a specified number of hours in advance of the human diagnosis. We then used our award-winning cognitive computing platform to competitively train hundreds of ARDS-onset decision support models using millions of hourly clinical observations from thousands of ICU patient stays. These models achieved earlier onset prediction with acceptable accuracy and low false-positive rates.


IOMICS Healthcare’s lead researcher on the ARDS-onset project was invited to present this work to the COVID-19 CDSS Initiatives Program at NIH. We are now working with the National COVID Cohort Collective (N3C) Clinical Data Working Group and using additional data from the N3C Data Enclave to explore how the use of more extensive patient data, such as medical history, comorbidities, and omic assay data, might further improve point-of-care decision-support models.