Workshop: Machine learning for Clinical Microbiology
Registration for this workshop is via the ECCMID registration portal.
1. Click the link above to go to the registration portal
2. Log in with your ESCMID credentials
3. Select 'Individual Registration'
4. Progress until the 'Tickets' tab, where you can then select tickets for the workshop
Artificial intelligence (AI) already impacts our work in many ways. Some of this smart software enables the disruptive technologies changing the face of laboratory medicine. Wouldn’t it be useful to master AI and machine learning, and dictate how and where it is used in the laboratory? But in the real world of the busy clinical lab, very few staff have got the time or inclination to start from scratch, learning command line script or coding language they weren’t already familiar with.
The great news is that it doesn’t have to be that hard. Molecular microbiology is already exploiting specific applications of machine learning. It is only a small step to adapt bioinformatic skills to handle other types of clinical microbiology data. One of the attractions of machine learning is its ability to reveal previously hidden structure in complex data sets, without resorting to high-end statistical methods.
In this workshop, we will take you step-by-step through a complex, big data problem and show you how to use open source software to build data machines to make sense of your data set. The software platform we will use does not require any coding, prior bioinformatics experience or advanced statistics.
You will need to bring a laptop computer that operates Windows, macOS or Linux.
We will provide registered participants with instructions in advance how to download the free machine learning software suite, guidance on which modules to load, a set of raw data files and a simple familiarisation activity, to prepare your laptop computer beforehand.
This workshop is a refined version of a machine learning training course run by the workshop coordinator at the Department of Microbiology, Region Kroneberg, Sweden in October 2018, and will be supported by the staff who completed their training then. Though our main focus will be on using machine learning to accurately determine antimicrobial susceptibility test results from flow cytometry data, we will share other applications of machine learning in clinical microbiology.
- see the potential diversity of machine learning applications in clinical microbiology
- learn how to build a data machine that can determine an MIC
- discover how to curate your data
- follow a step by step guide to building a data machine without any coding
- find the hidden meaning in bacterial flow cytometry data
- use your analytic pipeline you build to determine an MIC
- put your machine learning pipeline to use in a calibration task
You don't need any coding skills prior to the workshop, but if you have even the basics you'll be able to develop your own machine learning toolkit for big data projects.
The workshop will be using Orange 3.20.1, which uses Python.
0935 Broad principles of machine learning
1000 Introducing open source machine learning software
1015 Step 1: caring for your data – scraping and cleaning
1115 Step 2: define your control population with Data machine 1
1200 Step 3: first look AST with Data machine 2
1300 Lunch break & troubleshooting
1400 Step 4: closer look at AST with Data machine 3
1500 Step 5: create your data machine ensemble
1530 Summative exercise
Dr Tim Inglis, FAST Lab, University of Western Australia & PathWest Laboratory Medicine
Dr Oskar Ekelund, Department of Microbiology, Region Kroneberg, Sweden
Dr Sofia Samajo, Department of Microbiology, Region Kroneberg, Sweden
Staff member, FAST Lab, University of Western Australia & PathWest Laboratory Medicine