Research Area: Mortality Prediction

Neo-Bedside Monitoring Device for Integrated Neonatal Intensive Care Unit (iNICU)

Neonatal intensive care units generate massive amounts of physiological data from bedside monitors, ventilators, infusion pumps, blood gas analyzers, and other medical devices. Traditionally, these data are manually documented at regular intervals, leading to transcription errors and the loss of valuable high-resolution clinical information.

This paper introduces Neo, an affordable bedside Internet of Things (IoT) device that integrates heterogeneous NICU medical devices into the cloud-based iNICU platform. Neo continuously acquires real-time physiological signals, synchronizes data from multiple vendors, automatically generates nursing charts, and provides clinicians with unified patient trends through web and tablet interfaces.

The platform combines embedded hardware, cloud infrastructure, and machine learning to support early prediction of neonatal morbidity using continuous physiological signals such as heart rate, respiratory rate, and oxygen saturation. Validation using data collected from 92 preterm infants demonstrated the feasibility of real-time clinical analytics for supporting early intervention and improving neonatal care.

Neo established one of the earliest vendor-agnostic bedside device integration platforms specifically designed for neonatal intensive care, laying the foundation for AI-assisted neonatal monitoring systems.

Development and Validation of High Definition Phenotype-Based Mortality Prediction in Critical Care Units

Accurate prediction of neonatal mortality remains one of the most challenging problems in critical care. This study developed a High Definition Phenotype (HDP) framework that integrates electronic medical records, laboratory results, bedside physiological monitoring, medications, nutrition, and derived physiological features into a unified time-series representation of each patient.

Clinical data from 1,546 neonatal admissions across eight Level III NICUs were collected, with 757 patients containing complete fixed, intermittent, and continuous datasets for model development. Two predictive approaches—a Logistic Regression Model (LRM) and a Long Short-Term Memory (LSTM) deep learning model—were developed and compared with established neonatal severity scores including CRIB, CRIB-II, SNAP-II, and SNAPPE-II.

Both AI models consistently outperformed traditional severity scores, with the LSTM achieving an AUC of up to 0.96, while also providing continuously updated mortality risk predictions throughout hospitalization. The framework demonstrates the potential of integrating multimodal clinical data into real-time bedside decision support systems capable of detecting patient deterioration earlier than conventional approaches.