ML Algorithm to Detect Rare Clinical Events

Slide 1

Slide 1 text

Haque Ishfaq PhD Researcher Montreal Institute for Learning Algorithms (MILA); School of Computer Science McGill University, Canada @haqueishfaq Samara Sharmeen Resident Surgeon VIVANTES Netzwerk für Gesundheit GmbH RWTH Aachen University, Germany @ssharmeen Hassan Sami Adnan Medical Researcher Faculty of Health, Medicine and Life Sciences, Maastricht University World Health Organization (WHO) @hsami Hassan Saad Ifti DPhil Researcher Department of Engineering Science University of Oxford, United Kingdom @saadifti Atia Amin Researcher Department of Biology University of South Dakota, United States @atiaamin1 ML Algorithm to Detect Rare Clinical Events

Slide 2

Slide 2 text

Success of Deep Learning Source: Stanford CS 231N

Slide 3

Slide 3 text

Step 1: Create a massive labelled dataset

Slide 4

Slide 4 text

Step 2: Train a large deep neural network Source: Stanford CS 231N

Slide 5

Slide 5 text

Step 3: Evaluate Performance

Slide 6

Slide 6 text

But is that all? What if our training set training data

Slide 7

Slide 7 text

And then during test time ... test data

Slide 8

Slide 8 text

Not just a toy example! Perils of biased

Slide 9

Slide 9 text

Not just a toy example!

Slide 10

Slide 10 text

TVAE: Triplet Based Variational - Triplet Loss and Metric Learning - Generative Model - Variational Autoencoder

Slide 11

Slide 11 text

What is Metric Learning? training data

Slide 12

Slide 12 text

But is that all? What if our training set training data Representation

Slide 13

Slide 13 text

What is Metric Learning? Positive Anchor Negative

Slide 14

Slide 14 text

What is Metric Learning? Positive Anchor Negative Learning

Slide 15

Slide 15 text

What is Metric Learning? Positive Anchor Negative Positive Anchor Negative Learning

Slide 16

Slide 16 text

Variational Autoenconder? Source: http://kvfrans.com/variational-autoencoders-explained/

Slide 17

Slide 17 text

Triplet Based Variational Autoencoder

Slide 18

Slide 18 text

MNIST Dataset

Slide 19

Slide 19 text

Triplet Based Variational Autoencoder

Slide 20

Slide 20 text

0 2000 4000 6000 8000 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 Digital Health Deep Learning Artiﬁcial Intelligence Number of publications on PubMed, on three different topics Emerging trend in medicine

Slide 21

Slide 21 text

Problems of classification - Difficulty in defining clinical classifications - Risk of misdiagnosis - Delay in screening or treatment - Missed opportunity for detecting rare events

Slide 22

Slide 22 text

Benefits of representation - More relaxed paradigm - Physicians in the loop to deal with red flags - Early detection of rare events (e.g. microcalcification in breast cancer)

Slide 23

Slide 23 text

tomorrow’s Talk… Simulation Model for 3D-printed Drug Development

Slide 24

Slide 24 text

PySTEMPlus @PySTEM Haque Ishfaq PhD Researcher Montreal Institute for Learning Algorithms (MILA); School of Computer Science, McGill University, Canada @haqueishfaq Samara Sharmeen Resident Surgeon VIVANTES Netzwerk für Gesundheit GmbH; RWTH Aachen University, Germany @ssharmeen Hassan Sami Adnan Medical Researcher Faculty of Health, Medicine and Life Sciences, Maastricht University; World Health Organization (WHO) @hsami Hassan Saad Ifti DPhil Researcher Department of Engineering Science; University of Oxford, United Kingdom @saadifti Atia Amin Researcher Department of Biology University of South Dakota, United States @atiaamin1 Outreach platform for underprivileged children PyHEALTH Bridging IT and Healthcare Follow us on Twitter for latest updates on our research. Our next event Hosting the ﬁrst ever Python coding camp for children in Mym, Bangladesh!