Peter Jacko peter.jacko [a] gmail.com

Senior Lecturer (Associate Professor) in Management Science Department of Management Science, Lancaster University Management School, LA1 4YX, UK Senior Decision Scientist, Berry Consultants, Abingdon, UK At Lancaster University, I am also member of: Foundations theme, Data Science Institute research groups: Health Systems, Simulation and Stochastic Modelling and Optimisation STOR-i: Statistics and Operational Research Centre for Doctoral Training in Partnership with Industry

:: Group on Optimal Adaptive Learning

G.O.A.L. is an informal research group led by Dr. Peter Jacko, founded in 2018, although his research interests in these topics have been materialising since 2005 when he started his PhD. The mission of G.O.A.L. is to:

• develop, compare and popularise (i) stochastic models, (ii) exact and approximate analytical methods, such as reinforcement learning and approximate dynamic programming, and (iii) computational and simulation methods for addressing finite-horizon variants of the multi-armed bandit problem
• develop and share efficient and user-friendly software implementations of these methods

with the vision of bringing these techniques to practice, especially:

• to design patient-centric (such as response-adaptive and/or covariate-adjusted) randomised clinical trials and other sequential experiments
• to develop optimal adaptive policies for other application areas

:: Collaborators

      Sofia S. Villar, Cambridge University
      James Wason, Newcastle University & Cambridge University
      Thomas Jaki, Lancaster University
      Kevin Glazebrook, Lancaster University
      Stephen Chick, INSEAD

:: Students

      Faye Williamson (PhD student), Lancaster University
      Amin Yarahmadi (PhD student), Lancaster University

:: Publications

      Bofill Roig et al. (2021) On Model-Based Time Trend Adjustments in Platform Trials with Non-concurrent Controls, Submitted 12/2021 (paper)
      Villar, S. S., Jacko, P. (2021): Bandit Procedures for Designing Patient-Centric Clinical Trials, chapter in The Elements of Joint Learning and Optimization in Operations Management (chapter)
      Williamson, S. F., Jacko, P., Jaki, T. (2021): Generalisations of a Bayesian Decision-Theoretic Randomisation Procedure and the Impact of Delayed Responses, in Computational Statistics & Data Analysis (paper)
      Gibson, L., Jacko, P., Nazarathy, Y. (2021): A Novel Implementation of Q-Learning for the Whittle Index, in Proceedings of ValueTools (paper)
      Williamson, S. F. (2020): Bayesian bandit models for the design of clinical trials (PhD Thesis)
      Jacko, P. (2019): BinaryBandit: An Efficient Julia Package for Optimization and Evaluation of the Finite-Horizon Bandit Problem with Binary Responses, Submitted 06/2019 (working paper)
      Jacko, P. (2019): The Finite-Horizon Two-Armed Bandit Problem with Binary Responses: A Multidisciplinary Survey of the History, State of the Art, and Myths, Submitted 06/2019 (working paper)
      Williamson, S. F., Jacko, P., Villar S. S., Jaki, T. (2017): A Bayesian Adaptive Design for Clinical Trials in Rare Diseases, in Computational Statistics & Data Analysis (paper)

:: Projects

      09/2020-08/2024: "Methods for Sustainable Testing of Biomedical Interventions", a proposal for the UKRI Future Leaders Fellowship Scheme, UK (funding decision expected Feb 2020). This is a major £1.75M project with a team of 3 postdoctoral researchers and the above mentioned collaborators, involving several partner organisations (Sheffield CTRU, NICE, AstraZeneca, Roche) and a prominent multidisciplinary advisory board (Don Berry, Jack Bowden, Emilie Kaufmann, Ger Koole, Nigel Stallard, Christina Yap)
      02-08/2018: "Modelling and Simulation of Clinical Trial Designs" with Phastar, UK and the Trial Design Modelling Centre at AstraZeneca, UK

:: Talks

Peter has given a number of talks on this topic to a range of audiences, from biostatisticians in pharma industry to academic researchers in fields including applied mathematics, operational research, stochastic modelling, statistics and machine learning:
Upcoming:
Past:
      10/2021: INFORMS Annual Conference, Anaheim, CA, USA (virtual)
      04/2021: IMA-ORS Conference on Mathematics of Operational Research, Birmingham, UK (virtual)
      07/2020: Berry Consultants, USA
      03/2020: Loughborough University, UK (abstract, slides)
      06/2019: PSI Conference, London, UK (abstract, slides)
      08/2018: Advanced Analytics Centre, AstraZeneca, Cambridge, UK (abstract, slides)
      06/2018: PSI Conference, Amsterdam, The Netherlands (abstract, slides)
      05/2018: Workshop on Multi-Armed Bandits and Learning Algorithms, Rotterdam, The Netherlands (abstract, slides)
      02/2018: Product Development Biometrics Biostatistics Group, Roche Pharmaceuticals, Welwyn Garden City, UK (abstract, slides)
      12/2017: Queueing Theory Workshop, Obergurgl, Austria (slides)
      04/2017: InSPiRe (Innovation in Small Populations Research) Conference, Warwick, UK (poster)
      04/2017: IMA and OR Society Conference on Mathematics of Operational Research, Birmingham, UK (abstract, slides)
      02/2017: Workshop on Stochastic Models, Statistics and Their Applications, Berlin, Germany (extended abstract, slides)

Our results have also been presented by our collaborators and students at PSI 2016 (Germany), IBC 2016 (Canada), CEN ISBS 2017 (Austria), and at a number of workshops and seminars.

:: Software

      10/2019: https://peterjacko.shinyapps.io/binarybandit-app/: A web-based interactive application (R Shiny App) that provides visualization of data obtained from extensive computational experiments using the BinaryBandit package for Julia programming language. It includes and extends the results presented in my 2019 working paper and in the PSI 2019 conference and the PSI 2018 conference slides (see above). The app is under continuous development.
      08/2019: https://github.com/PeterJacko/BinaryBandit: A package in Julia programming language for optimization and evaluation of the multi-armed bandit problem with binary (success/failure) responses. Some results obtained using this software are presented in my 2019 working paper and in the PSI 2019 conference and the PSI 2018 conference slides (see above). The package is under continuous development.

Last modified: 11/2018