CHEST 2023 On Demand Pass-Big Data in Sepsis: From Epidemiology to Diagnosis

Video Transcription

Video Summary

In this session, three speakers discussed the use of big data, machine learning, and artificial intelligence (AI) in sepsis research and clinical practice. The first speaker, Vinny Liu, discussed the promises and pitfalls of predictive models in critical care. He presented data showing the increase in predictive models in the literature, but noted that there has not been a clear improvement in their performance over time. He also highlighted challenges such as heterogeneity in sepsis definitions and the lack of standardized data preprocessing. The second speaker, Tim Buchman, discussed the use of big data to understand sepsis epidemiology. He presented data showing the increasing burden and costs of sepsis in Medicare beneficiaries. He also highlighted the need for careful analysis of big data to avoid biases such as immortal time bias. The third speaker, Laura Evans, discussed the use of sepsis phenotypes to improve clinical care and research. She emphasized the importance of accurately identifying subgroups of sepsis patients who may respond differently to treatment. Lastly, Haley Gershnger, provided an editor's perspective on the use of big data, machine learning, and AI in sepsis research and academic writing. She highlighted the potential pitfalls of using these tools, such as poor quality research and immortal time bias. She also discussed ethical considerations and the potential use of AI in manuscript reviewing. In summary, the speakers discussed the potential of big data, machine learning, and AI to improve sepsis care and research, but also highlighted the challenges and ethical considerations associated with their use.

Meta Tag

Category Critical Care

Session ID 1147

Speaker Timothy Buchman

Speaker Hayley Gershengorn

Speaker Vincent Liu

Speaker Kathryn Pendleton

Speaker Steven Simpson

Track Critical Care

Keywords

big data

machine learning

artificial intelligence

sepsis research

clinical practice

predictive models

sepsis definitions

data preprocessing

sepsis epidemiology

immortal time bias