I am a PhD student in the MRC BSU at the University of Cambridge. I am working on the project Clinical prediction under informative presence: Exploring what patterns and timing of repeated observations in routinely collected data can tell us about disease risk under the supervision of Dr Jessica Barrett and Dr Brian Tom.

My research interest lies in developing machine learning models for temporal data with a particular focus on inequalities in medical care access and delivery.

Publications

Papers

DeepJoint: Robust Survival Modelling Under Clinical Presence Shift: December 2022; NeurIPS Workshop; V. Jeanselme, G. Martin, M. Sperrin, N. Peek, B. Tom, J. Barrett; Code: Github

Imputation Strategies Under Clinical Presence: Impact on Algorithmic Fairness: December 2022; Machine Learning for Health; V. Jeanselme, M. De-Arteaga, Z. Zhang, B. Tom, J. Barrett; Code: Github

Constrained clustering and multiple kernel learning without pairwise constraint relaxation: June 2022; Springer - Advances in Data Analysis and Classification; B. Boecking, V. Jeanselme, A. Dubrawski; Code: Github

Neural Survival Clustering: Non-parametric mixture of neural networks for survival clustering: April 2022; Conference on Health, Inference, and Learning (CHIL); V. Jeanselme, B. Tom, J. Barrett; Code: Github

Sex differences in post cardiac arrest discharge locations: December 2021; Resuscitation Plus; V. Jeanselme, M. De-Arteaga, J. Elmer, S. M.Perman, A. Dubrawski

Deep Parametric Time-to-Event Regression with Time-Varying Covariates: April 2021; AAAI Spring Symposium on Survival Analysis; C. Nagpal*, V. Jeanselme*, A. Dubrawski

Prediction of Hypotension Events with Physiologic Vital Sign Signatures in The Intensive Care Unit: August 2020; Critical Care; J. H. Yoon*, V. Jeanselme*, A. Dubrawski, M. Hravnak, M.R. Pinsky, G. Clermont

Preprint

Leveraging Expert Consistency to Improve Algorithmic Decision Support: July 2022; M. De-Arteaga, V. Jeanselme, A. Dubrawski, A. Chouldechova

Extended Abstracts

Using observation processes to predict survival: A deep learning ap- proach to joint modelling: IBC - July 2022; V. Jeanselme, G. Martin, M. Sperrin, N. Peek, B. Tom, J. Barrett
Dynamic Phenotypes Preceding Hypotension in Intensive Care: ESICM - October 2021; V. Jeanselme, A. Dubrawski, M. R. Pinsky, G. Clermont, J. H. Yoon
A Case For Federated Learning: Enabling And Leveraging Inter-hospital Collaboration: ATS - Philadelphia, PA, USA - May 2020; S. Caldas, V. Jeanselme, G. Clermont, M. R. Pinsky and A. Dubrawski
Robustness Of Machine Learning Models For Hemorrhage Detection: ATS - Philadelphia, PA, USA - May 2020; V. Jeanselme, A. Wertz, G. Clermont, M. R. Pinsky and A. Dubrawsk
Cross-correlation Features of Vital Signs Enable Robust Detection of Hemorrhage: ISICEM - Brussels, Belgium - Mar. 2020; V. Jeanselme, A. Wertz, G. Clermont, M. R. Pinsky and A. Dubrawsk
Predicting Hypotension Episode With Numerical Vital Sign Signals in the Intensive Care Unit: ISICEM - Brussels, Belgium - Mar. 2019; J. H. Yoon, V. Jeanselme, A. Dubrawski, M. Hravnak, M. R. Pinsky and G. Clermont
A Real-time Photoplethysmography Signal Artifacts Removal System in ICU: ESICM - Paris, France - Oct. 2018; Y. Chen, J. H. Yoon, V. Jeanselme, M. Hravnak, A. Dubrawski, M. R. Pinsky and G. Clermont
Prediction for Hypotension Episode with Multigranular Data in the Intensive Care Unit: ATS - San Diego, CA, USA - May 2018; Received ATS Abstract Scholarship Award; J. H. Yoon, V. Jeanselme, Y. Chen, A. Dubrawski, M. Hravnak, M. R. Pinsky and G. Clermont

Talks

Using observation processes to predict survival: A deep learning approach to joint modelling: IBC - July 2022; V. Jeanselme, G. Martin, M. Sperrin, N. Peek, B. Tom, J. Barrett
Neural Survival Clustering: Non-parametric mixture of neural networks for survival clustering: CHIL - April 2022; V. Jeanselme, B. Tom, J. Barrett
Deep Parametric Time-to-Event Regression with Time-Varying Covariates: AAAI Spring Symposium on Survival Analysis - April 2021; C. Nagpal*, V. Jeanselme*, A. Dubrawski

Teaching

AI4ALL

I mentored highschool students during AI4ALL summer programms. In order to learn the basics of machine learning and programming in python, I created the House Pricing Project using data from Kaggle.

Teacher Assistant

I have been teacher assistant for Applied Data Science at Carnegie Mellon University (Spring 2020) and for Principles of Biostatistics at Cambridge (Fall 2021).

Visualization

Visalization is essential to understand the data at hand. I created a simple tool to visualize transformation from orignal space, to TSNE and PCA.

Clustering for labeled data can also be visualized with this tool which allows to display the hierachy given the class distribution in subbranches.

I also had some interest in visualizing sound

Open Source

Reproducibility is key in science, I publish the majority of my code in order to allow other researchers to make further analyses and help me to improve my code and findings.

A few tools that I use:

MovingWindowDefinition
SklearnTS: A tool for working with time series.

Some projects that I wanted to work on:

Image Compression is a simple use of singular value decomposition for data compression.
SearchEngine allows to search in a set of files.

I have also implemented code for a few papers that I find particularly interesting:

And also some analysis of Kaggle datasets:

How to learn vocabulary ?

Contact

If you are interested in working with me, send me an email at vincent.jeanselme@gmail.com or contact me @JeanselmeV

Vincent Jeanselme

PhD student in Biostatistics at the University of Cambridge