PROGRAM
Hands-on Workshops
March 7, 2019 9:00-16:30@Lecture Room 2, MD Tower 2F, TMDU
9:00-9:10
Welcome messages
Satoru Hashimoto (Representative Organizer)
Hidenobu Shigemitsu (Co-organizer)
Tetsuya Taga
(Executive Senior Vice President of Global Affairs, Tokyo Medical and Dental University)
9:10-9:30
Lecture
Electronic health record in quality improvement - Clinician’s perspective
Adrian Velasquez
9:30-9:50
Announcement of Hands-on Workshops
Leo Anthony Celi/Satoru Hashimoto/Euma Ishii
10:00-12:30
WS01-WS04
WS01
An introduction to causal inference
Ryo Uchimido/Wei-Hung Weng
Causal inference is an approach to estimate an average causal effect of treatment. Although machine learning/deep learning has developed many predictive algorithms with extraordinal accuracy, those algorithms are not sufficient in clinical decision making. For instance, identifying patients with poor outcomes might be very different from identifying the best treatment in a population. Causal inference could offer theoretical frameworks for the better choices of treatments. In addition, applying machine learning for causal inference has proposed as a next step of machine learning developments.
This course aims to provide a good start of learning causal inference approach. This course is mainly for participants who don’t have previous knowledge of causal inference. We will start from scratch by using a “DoWhy” Python library. This workshop has the following requirements: laptop with Anaconda 5.0.1 (Python 3.6 version) installed, and basic understanding of Python and Jupyter notebook. Required reading; https://arxiv.org/abs/1804.10846.
WS02
Matching & Causal Inference: New Directions
Aaron Kaufman/Adrian Velasquez
Matching is an important tool in causal inference, especially in the analysis of patient records, and it is one of the fastest-developing areas of research in statistics. In this workshop, we will discuss the basics of matching, the assumptions under which matching produces causal estimates, some of the classic techniques used to perform matching, and the methods used to measure match quality. We will apply these methods to MIMIC data related to patients with sepsis. Finally, we will introduce matching with text data, a new frontier in matching, and show how text data can improve match quality.
Prerequisites:
1) Basic experience with causal inference
2) Basic familiarity with the R programming language and R Studio
WS03
Data analysis using JIPAD, ANZICS APD, etc.
Shawn Sturland/David Pilcher/Satoru Hashimoto
Clinician researchers from Japan, Australia and New Zealand will demonstrate an approach to analysis of critical care registry data for research, with live interactive analysis of de-identified JIPAD data done in real time and in collaboration with participants. In two hours, we will attempt to construct a whole medical manuscript from scratch! For statistical analysis, STATA will be applied but no programming skills will be required.
WS04
Deep learning for medical imaging
Mengling (Mornin) Feng
Artificial intelligence (AI) is the buzzword for all things relating to technology these days. In particular, healthcare is seen as an area in which AI may be gainfully deployed to improve medical care, especially with big data, exponential computing power and a burgeoning demand on healthcare systems due to aging populations. In this talk, I will share one use case that my group has engaged recently.
Our use case focuses on the analysis of medical images, in particular, the mammogram images. My team recently participated the DREAM digital Mammogram challenged funded by the US white house fund and FDA. The aim was to improve the predictive accuracy of digital mammography for the early detection of breast cancer. The challenge was constructed around 640,000 mammogram images for over 80,000 patients. We developed a patch-based deep neural network structure to extract abnormal lesions and patterns from mammograms so to detect breast cancer. After months of training, the proposed AI model managed to achieve detection performance close to human experts. We are now working with the Breast Cancer Screening Center at National University Hospital aiming to develop and deploy an AI-aided tool to assist radiologists to achieve more accurate, consistent and faster breast cancer screening. Our workshop will share the fundamentals on how deep neural networks can be applied on medical image analysis, and I will also share the lessons learned via our mammogram project. Only pen and paper will be required.
12:30-14:00
Lunch
14:00-16:30
WS05-WS09
WS05
Tree based approaches for prediction
Tom Pollard/Stephanie Ko/Satoshi Kimura
This interactive workshop offers participants an introduction to tree-based approaches for prediction. We will begin by discussing some of the core concepts of tree models, walking through the process from training a model to testing performance. Following the introduction, participants will get hands-on experience using Python to analyse data in a cloud database. Some prior knowledge of programming is helpful, but not required. Please bring a laptop!
WS06
Medical data science 101 for clinicians
Matthieu Komorowski/Kenji Wakabayashi
This workshop is specifically dedicated to clinicians with minimum or even zero prior knowledge in data science, coding or machine learning. I will offer suggestions on possible strategies to get up to speed on these topics and embrace the era of Health 2.0. The workshop will be based around building a simple mortality prediction model in SQL and Python from scratch from the MIMIC-III database, and along the way we’ll be discussing what skills are desirable or required for a doctor interested in specializing in data science, how to acquire those skills and what are the keys to success of a data science project. No need to wait! You will be able to start coding straight away during the workshop, so bring a laptop.
WS07
From ideation to entrepreneurship: The role of artificial intelligence in healthcare innovation
Dan Ebner/Leo Anthony Celi/Louis Agha-Mir-Salim
This workshop explores early idea generation, prototyping, and testing, as well as considerations of entrepreneurship within healthcare systems. Using the influx of artificial intelligence companies into healthcare as motif, we will introduce participants to the fundamental concepts of entrepreneurship, discuss the motives and barriers of integrating startups into health systems, and relate this to prominent examples of successes and failures. Within the session we will ideate on potential entrepreneurial projects in medicine, both in Japan and abroad, and then test these ideas in small, skill-diversified groups, with an opportunity to pitch to the rest of the workshop audience afterward. Interested teams will also be introduced to the thinkLab student network at Harvard/MIT and expanding abroad, which will offer support and mentorship for teams as they build their ideas and move toward launch. All backgrounds welcome, no previous experience required.
WS08
Software and tools for data analytics - GitHub, Jupyter notebook, Google Colabs, Slack
Kenneth Paik/Marta Fernandes
We are at a critical inflection point in clinical science, where we must address the reproducibility crisis. Not only do we need to be wary of the researchers and methods, but also cautious about biases in the underlying data. In this workshop, we discuss how Open Science and free access data can support and stimulate improved replicability in medicine. There will also be some directed hands-on primer with some open source tools, including executable notebooks and code repositories.
WS09
Natural language processing
Christina Chen/Euma Ishii/Naoaki Ichihara
There are valuable insights that can be gained from information hidden in free text clinical notes. In this workshop, participants will be introduced to the process of extracting data from free text. The first half of this session will consist of leveraging regular expressions (RegEx) to analyze formatted clinical notes. The second part will introduce a program called cTakes (visual natural language processing system) and SNOMED CT (standardized collection of medical terms). Participants will have plenty of hands-on practice.
Lectures
March 8, 2019 9:00-17:00@Google Japan
9:00-9:10
Welcome message
Yasuyuki Yoshizawa (President, Tokyo Medical & Dental University)
9:10-9:30
Startup; Summary of the 1st Datathon in Japan
Satoru Hashimoto/Hidenobu Shigemitsu
9:30-10:00
L01
Deconstructing the Hype around Medical AI
Leo Anthony Celi
10:00-10:30
Break
10:30-11:15
L02
When machine learning meets practice: A Singapore story
Mengling (Mornin) Feng
11:15-12:00
L03
Big data and AI in medicine
Hiroshi Tanaka
12:30-13:30
Lunch-time Lecture
How Leveraging National-level Big Data in Healthcare?
Haruki "Harry" Matsumoto
13:30-14:15
L04
Implementing AI in clinical practice
Kee Yuan Ngiam
14:15-15:00
L05
Realtime clinical text processing on ER/ICU database
Tomohiro Sonoo
15:00-15:30
Coffee Break
15:30-16:15
L06
How we have used data from the ANZICS registry for publications/QI projects/routine use
David Pilcher
16:15-16:30
Introduction to MIMIC
Tom Pollard
16:30-16:45
Introduction to GCP
Andrew Moschetti
16:45-17:00
Introduction to 10 CQs and mentors
Satoru Hashimoto
Datathon Workshop
March 9 &10, 2019 9:00-16:30@Google Japan
9:00-12:30
Datathon
12:30-13:30
Lunch-time Lecture
Challenges and opportunities of using big data in public health
Alex Cook
13:30-18:00
Datathon
16:00-17:00
Preparation for datathon; Introduction of 10 CQs and mentors
Datathon Workshop
March 10, 2019 8:30-17:00@Google Japan
9:00-12:30
Datathon
12:30-13:30
Lunch-time Lecture
Fujitsu’s cutting-edge AI technologies
Fumihiro Maruyama
13:30-14:30
Datathon, last spurt
15:00-16:30
Presentation from 10 CQ teams
16:30-17:00
Award and Wrap-up
Leo Anthony Celi/Mengling (Mornin) Feng/Satoru Hashimoto/Hidenobu Shigemitsu
Photo-taking