Detecting potentially misdiagnosed dementia cases via classification
Machine Learning, datasets, Naïve Bayesian, Minimum distance, Neural Networks, ICD-10, READ, Nominal, Risk Factors… What does it all mean? How do familiar concepts from the healthcare world, map to the mathematical terminology and how does it all make sense?
This workshop is designed to provide a thorough exposition of four classification methods, namely, Minimum Distance Classifier, Bayes Classifier, Neural Networks and Random Decision Forest. The experience from a realistic scenario of trying to identify cases of potentially misdiagnosed dementia will be used to demonstrate the relevance between decisions, terminology at a technical level and how they are affected by the environment the data are produced and captured in.
The workshop is based on real world experience that has been captured in the following two publications:
- An in-depth exposition to Machine Learning classification terminology and methodology and how they relate to the real-world healthcare environment.
- Practical hands on application of the theory to a realistic scenario involving dementia.
- Participant's pack including: A Virtualbox instance with all required software and data set-up and ready to go. This is for the participants to keep. This is not a training tool, this is a production ready environment that participants can keep and apply what they learn from their course at their own time.
Maximum 20 participants. Participants must bring their own laptops which should be fairly new (purchased within the last 3-4 years).