Bing Xue

Recent Projects

Most of my projects are focused on generating better latent representations of multi-modal inputs, and inferring factual and counterfactual outcomes from the latent representation.

Large Language Models in Medicine

Exploit the clinical notes! The processing of open and unstructured clinical notes before and during surgeries is always tedious, as the notes can be arbitrarily long and written in various tenses/formats. We leveraged the recent advances of Large Language Models (LLMs) and developed a foundational language model to read the notes for us! We finetuned the pretrained LLMs with respect to the key aspects during perioperative care, and demonstrated that LLMs can accurately predict postoperative risks from the clinical notes! Our paper is in preparation.

Individualized Treatment Effect Estimation

JAMIA | Artificial Organs | KDD

We could have saved more people! In collaboration with National COVID Cohort Collaborative (N3C) from NIH and International Severe Acute Respiratory and Emerging Infection Consortium (ISARIC) from Oxford University, we investigated COVID-19 patients from 63 countries across 5 continents who were admitted to Intensive Care Units (ICU) during the pandemic. We evaluated the impact of their treatment assignment and responses, and developed several clinical assisting algorithms that could predict the best treatment options by considering both the their factual and counterfactual responses.

AI4Science

We are exploring a brand-new domain - can AI help physicians decode the material properties from their spectral imaging and vice versa? We have developed an innovative optical encoder-decoder architecture that nearly perfectly recover the spectral imaging from crystal responses! We are developing a meta-learning based algorithm that decodes various material properties by reading the spectra. I will work on publishing our results after my internship.

Realtime Sequential Modeling

Sleep Medicine | Sensors | British Journal of Anaesthesia

How to continuously update our beliefs with the inflow of new measurements/observations? We explored two directions. The first one was to design the statistical features that quantified the discrminant patterns in the time series. We have demonstrated that this solution was particularly suitable for wearale devices in first two papers. The second one was to update our posterior distributions as new observations flowed in. We have demonstrated its superior performance on complex multi-modal datasets.

Representation Learning in Healthcare

JAMA | KDD | ICDM

When learning from Electronic Health Records, we face 3 challenges:

How to learn from hundreds, even thousands of input features?

How to deal with partial observations where each patient only has a small subset of input features available?

How to model the rare postoperative complications?

We first examined the feature redundancy and data imbalance in the real clinical datasets, and evaluated various imputation stategies to deal with missingness (JAMA). After that, we developed a representation learning framework using Generative AI to model the input complexity and uncover its nonlinear relationship with the postoperative risks (KDD). To build clinical trust, we developed an explainable sequential model to identify the key events that cause risk elevation (ICDM).

Publications

Conference Papers:

Xue, B., Said, A., Xu, Z., Liu, H., Shah, N., Yang, H., Payne, P. and Lu, C., Assisting Clinical Decisions for Scarcely Available Treatment via Disentangled Latent Representation
KDD 2023 | paper | news


Xue, B., Jiao, Y., Kannampallil, T., Fritz, B., King, C., Abraham, J., Avidan, M. and Lu, C., 2022, August. Perioperative predictions with interpretable latent representation
KDD 2022 | paper | report by Medical Press | news by News Medical | news by Health IT Analytics | news by WahU Source | news by WashU Engineering | news by EurekAlert | news by SCIENMAG


Li, D., Xue, B., C., King, Fritz, B., Avidan, M.S., Abraham, J, and Lu C. (2022, Nov). Self-explaining Hierarchical Model for Intraoperative Time Series
KDD 2022 | paper


Zhang, Z., Xiang, Y., Wu, L., Xue, B. and Nehorai, A., 2019. Kergm: Kernelized graph matching
NeurIPS 2019 | paper


Shi, W., Xue, B. Guo, S., Goh, D. Y., and Ser, W. (2018). Obstructive Sleep Apnea Detection Using Difference in Feature and Modified Minimum Distance Classifier. EMBC 2018 | paper

Journal Papers:

Xue, B., Shah, N., Xu, Z., Yang, H., Marwali, E., Dalton, H., ... and ISARIC Clin- ical Characterisation Group. (2023). Validation of extracorporeal membrane oxygenation mortality prediction & severity of illness scores in an international COVID‐19 cohort.
Artificial Organs 2023 | paper


Xue, B., Shah, N., Yang, H., Kannampallil, T., Payne, P. R. O., Lu, C., and Said, A. S. (2022). Multi-horizon predictive models for guiding extracorporeal resource allocation in critically ill COVID-19 patients.
JAMIA 2023 | paper


Abraham, J., Bartek, B., Meng, A., King, C. R., Xue, B., Lu, C., and Avidan, M. (2022). Integrating Machine Learning Predictions for Perioperative Risk Management: Towards an Empirical Design of a Flexible-Standardized Risk Assessment Tool.
Journal of Biomedical Informatics 2023 | paper


Xue, B., Shi, W., Chotirmall, S., Koh, V., Ang, Y., Tan, X., and Ser, W. (2022). Distance-based detection of cough, wheeze and breath sounds on wearable devices.
Sensors 2023 | paper


Xue, B., Licis, A., Boyd, J., Hoyt, C. R., and Ju, Y. E. S. (2022). Validation of actigraphy for sleep measurement in children with cerebral palsy.
Sleep Medicine 2023 | paper | news by NeurologyLive


Jiao, Y., Xue, B., Lu, C., Avidan, M.S., and Kannampallil, T. (2021). Continu- ous Real-Time Prediction of Surgical Case Duration Using a Modular Artificial Neural Network.
British Journal of Anaesthesia 2022 | paper | Editorial Comment


Xue, B., Li, D., Lu, C., King, C.R., Wildes, T., Avidan, M.S., Kannampallil, T. and Abraham, J., (2021). Use of machine learning to develop and evaluate models using preoperative and intraoperative data to identify risks of postoperative complications.
JAMA network open 2022 | paper | Editorial Comment

Awards

2023, KDD'23 Student Travel Award.

2023, Co-PI of Big Ideas Grant.

2022, our work on perioperative care ("EnhanCed HandOffs (ECHO)) has been awarded R01 funding.

2015, Singapore and China's International Scholarship (SM2) for Undergraduate Students.

2014, 1st Prize in Singapore GreenTECH Competition.

2014, Singapore and China's International Scholarship (SM2) for Undergraduate Students.

2013, Singapore and China's International Scholarship (SM2) for Undergraduate Students.

2012, URECA Scholar, Nanyang Technological University.

2012, Singapore and China's International Scholarship (SM2) for Undergraduate Students.

2011, Singapore and China's International Scholarship (SM2) for Undergraduate Students.