Emma Rocheteau

ML for Healthcare PhD Student

University of Cambridge

Hello!

I am a PhD student at the University of Cambridge working on machine learning problems for healthcare, supervised by Pietro Liò. I am part of the AI group in the computer science and technology department, but I am originally from a medical background (my medical studies are paused between the 4th and 5th years to allow me to complete this PhD).

I have previously completed parts IA and IB (years 1 and 2) of the pre-clinical medical and veterinary sciences tripos, part IIA (intercalation year) of the engineering tripos and one year of clinical medicine (on the wards in the hospital setting).

Download my CV.

Interests

Deep Learning
Graph Neural Networks
Electronic Health Records
Healthcare
Medicine

Education

PhD in Deep Learning (in progress), 2021
University of Cambridge
MB BChir (in progress), 2017
University of Cambridge
BA in Engineering/Preclinical Medicine, 2016
University of Cambridge

Research Interests

My work has focused on predicting patient outcomes in the Intensive Care Unit (ICU). When designing my deep learning models, I am often inspired by my knowledge of clinical decision making.

For example, for time series processing in Electronic Health Records (EHRs), I use temporal and pointwise convolution to efficiently extract patient trajectories over time – a method inspired by clinicians.

I am also working on using graph neural networks to link the experiences of similar patients. The rationale is that when clinicians make decisions they will typically lean on their past experience, especially if they are dealing with a rare disease.

If you’re interested in the kind of stuff I do, follow me on twitter!

Featured Publications

Emma Rocheteau, Pietro Liò, Stephanie Hyland

April 2021 In CHIL Conference 2021 (also ML4H at NeurIPS 2020 and HSYS at ICML 2020)

Temporal Pointwise Convolutional Networks for Length of Stay Prediction in the Intensive Care Unit

Our model, Temporal Pointwise Convolution (TPC), is specifically designed to mitigate common challenges with Electronic Health Records, such as skewness, irregular sampling and missing data. We have achieved significant performance benefits of 18-68% over the Long-Short Term Memory (LSTM) network, and the Transformer.

Emma Rocheteau, Catherine Tong, Petar Veličković, Nic Lane, Pietro Liò

January 2021 In W3PHIAI-21 and DLG-AAAI'21 at AAAI 2021

Predicting Patient Outcomes with Graph Representation Learning

Our model, LSTM-GNN, is designed to take advantage of similarity between patients in the EHR (established using the diagnoses). First, it processes the time series data for each patient with the LSTM component, before sharing information within the neighbourhood of patients via the GNN. This is an alternative way of presenting diagnoses information (the common approach is to use an encoder in the late stages of a model). We found that using both methods together gains the best performance.

Other Papers

Quickly discover relevant content by filtering publications.

Emma Rocheteau, Doyoon Kim (2020). Deep Transfer Learning for Automated Diagnosis of Skin Lesions from Photographs. In ML4MH at NeurIPS 2020.

PDF Cite Poster arXiv

Jacob Deasy, Emma Rocheteau, Katharina Kohler, Daniel J. Stubbs, Pietro Barbiero, Pietro Liò, Ari Ercole (2020). Forecasting Ultra-early Intensive Care Strain from COVID-19 in England. In medRxiv.

PDF Cite Code Poster medRxiv

Experience

Academic Director

Institute for Medical AI

Aug 2020 – Present

Institute dedicated to advancing patient-centered care with medical AI.

Researcher Intern

Microsoft Research

Jul 2019 – Oct 2019 Cambridge, UK

13-week internship with the healthcare intelligence team investigating length of stay and discharge location prediction in the ICU.

Data Science Intern

Featurespace

Jul 2018 – Sep 2018 Cambridge

8-week internship involving feature engineering mouse/keyboard data to identify financial fraud using a random forest classifier.

Physiology Supervisor

University of Cambridge

Jan 2017 – Present Cambridge

I teach 11 first year undergraduate medical students cardiovascular, respiratory, renal, digestion, endocrinology and nerve and muscle electrophysiology. My role as supervisor involves tutorials each week, plus marking work (mainly essays) and answering any questions the students may have.

Recent & Upcoming Talks

Temporal Pointwise Convolutional Networks for Length of Stay Prediction in the Intensive Care Unit

CHIL Proceedings Paper Talk. The original talk can be viewed at 01:59:00 into Day 1 Proceedings on the CHIL website. There is a Q&A immediately following the talk.

Apr 8, 2021 3:00 PM Online

Predicting Patient Outcomes with Graph Representation Learning

W3PHIAI Spotlight at AAAI 2021

Feb 8, 2021 6:45 PM Online

See all events

Reviewing Experience

Machine Learning for Healthcare Conference (MLHC) 2021
Machine Learning for Preventing and Combating Pandemics (MLPCP) Workshop at ICLR 2021
Trustworthy AI for Healthcare (TAIH) Workshop at AAAI 2021
AMIA 2021 Informatics Summit
Machine Learning for Health (ML4H) Workshop at NeurIPS 2020*
Graph Representation Learning and Beyond (GRL+) Workshop at ICML 2020
Graph Representation Learning (GRL) Workshop at NeurIPS 2019
Machine Learning for Health (ML4H) Workshop at NeurIPS 2019

*My ML4H reviews were explicitly recognised as excellent by metareviewers

Awards

2nd Runner-Up for Best Short Paper Award, W3PHIAI Workshop at AAAI 2021
Best Talk Prize, Oxbridge Women in Computer Science Conference 2020
Best Poster Prize (49 submissions), 2nd HBP Student Conference 2018
AC Comfort Essay Prize, Royal Society of Medicine, 2017
Cambridge Medical Journal Essay Competition, 1st Prize, 2017
3rd Year Computer-Based Project Prize, Department of Engineering, 2016
Gold Rob Clarke Award for Undergraduate Research in Physiology, 2016
Churchill College Medical and Veterinary Prize, 2014, 2015 & 2016
Gold Award in the British Biology Olympiad, Royal Society of Biology, 2013
Gold Award in the British Chemistry Olympiad, Royal Society of Chemistry, 2013
Gold Award (& top 100) in the British Physics Olympiad, University of Oxford, 2013

Projects

TPC

A model for processing time series in Electronic Health Records. It outperforms the LSTM and Transformer models on MIMIC-IV and eICU.

LSTM-GNN

The LSTM-GNN model uses similarity between patients to obtain a better representation of the diagnosis information in the Electronic Health Record.

ICUnity

An initiative to harmonise the MIMIC-III and AmsterdamUMCdb databases.

Contact

ecr38@cam.ac.uk
Churchill College, Cambridge, CB3 0DS
Follow Me
Chat on LinkedIn