Approximate Bayesian Computation – Part 2

Wednesday, 24th July Time: 09:00 – 11:00 (BST) Table of Contents (powered by https://videoken.com) 0:03:17 Announcements 0:04:10 About Babylon 0:04:35 Together we can put accessible and affordable healthcare in the hands of every person on earth 0:05:15 We have 3 services 0:06:59 Transforming Healthcare in Rwanda 0:07:23 Babylon is expanding around the world 0:07:49 Areas of research 0:08:52 Some of Babylon Health's Publications 0:09:32 Approximate Bayesian Computation – Part 2 0:10:59 Summary of the First Part 0:12:12 Outline 0:13:38 Approximate Bayesian Computation 0:14:12 The Metropolis-hasting Algorithm 0:16:29 ABC-MCMC Method 0:21:54 ABC-MCMC - Why Does it Work? 0:26:42 Sequential ABC 0:27:53 Sequential ABC - A Schematic Representation 0:28:55 Population Monte-Carlo 0:32:21 Population Monte-Carlo for ABC 0:33:18 Sequential ABC - The Approach (1) 0:37:50 Sequential ABC - The Approach (2) 0:39:02 Sequential ABC - The Algorithm 0:44:49 Choice of the Perturbation Kernel 0:47:11 Properties of Optimal Kernel 0:49:16 Derivation of Optimal Kernel 0:52:31 Gaussian Random Walk Kernels 0:58:01 Computational Efficiency and Kernel 1:00:00 Computational Cost of Perturbation Kernel 1:02:46 Choice of the Threshold Schedule 1:05:14 Drawback of the Quantile Approach 1:10:32 Outline 1:19:39 Model Selection 1:20:15 Bayesian Approach for Model Selection 1:22:57 Computation of the Evidence 1:24:14 Generic ABC for Model Choice 1:28:43 ABC for Model Selection 1:31:41 Potential Issues 1:32:26 ABC Approximation to a Bayes Factor 1:38:24 Sufficient Statistics and Model Selection 1:41:19 How To Select Summary Statistics? 1:42:53 Example: Laplace Vs Normal 1:50:39 Example: Leukocyte Migration 1:56:11 Take-Home Messages 1:58:08 References