Research Summary

Early Detection of Sepsis in Paediatric Intensive Care Units through Integration of Clinical and Big Data

Computational decision-support can save lives of critically ill patients through early recognition of actionable events. The discovery of such events from the deluge of Big Data requires interdisciplinary efforts and bridging expertise. A highly devastating condition in the paediatric and neonatal Intensive Care Units (ICU) is ‘sepsis’ and the primary goal of this project is its early detection through computational integration and modeling of routine ICU data. Further, striking a balance in the patient management plan is highly critical. On one hand, delay in the recognition and treatment of sepsis leads to mortality-rates as high as 50% in India, while on the other, over-aggressive antibiotic therapy is a risk factor for death itself.  Stratification of suspected sepsis into sub-classes is expected to guide treatment plans. Thus, the overarching theme of this project is use Big Data analytics in combination with domain expertise to derive models and assist patient management plans. This would be carried out without disrupting the clinical care, using routinely generated multi-dimensional data. The collaborative effort between AIIMS, IIT-Delhi and Stanford School of Medicine (amongst other partners) is not only aimed at discovery but to also to fuel crosstalk and translation at the interface of medicine and data-science.

Figure Legend: The window of opportunit, y is proposed to lie in the network of subtle-physiological and clinical parameters. This may lead the clinical manifestations of the disease by hours, even days. This network shall be mined, modeled and exploited using techniques of Artificial Intelligence, Machine Learning and the Big Data ecosystem. The end goal is to obtain patient-centric interpretable models to save lives and to spur new knowledge in the septic response.