Lecture 5: Sequential Multiple Assignment Randomized Trials (SMARTs) for DTR. Donglin Zeng, Department of Biostatistics, University of North Carolina

Size: px
Start display at page:

Download "Lecture 5: Sequential Multiple Assignment Randomized Trials (SMARTs) for DTR. Donglin Zeng, Department of Biostatistics, University of North Carolina"

Transcription

1 Lecture 5: Sequential Multiple Assignment Randomized Trials (SMARTs) for DTR

2 Introduction

3 Introduction Consider simple DTRs: D = (D 1,..., D K ) D k (H k ) = 1 or 1 (A k = { 1, 1}). That is, a fixed treatment is assigned to any patients at stage k. How shall we evaluate V(D)? Conduct a single-arm randomized trial. How shall we compare two and more different DTRs: D 1 vs D 2 (not necessarily simple ones)? Conduct a multiple-arm randomized trial.

4 Randomization Is Key Why randomization is needed? It provides representative coverage of the patient population. It controls all potential confounders which may bias the effects of comparing treatments. One essential mathematical relationship links observed outcome to potential outcome: E[R(d) H] = E[R(d) A = d, H] by randomization = E[R A = d, H] by consistency assumption. Consistency/SUTVA: R = R(d) if A = d and is not affected by the particular treatment assignments to the others.

5 Difficulty in Design Studies for Optimal DTRs Even for simple DTRs and only two treatment options at each stage, the total number of DTRs is 2 K. The optimal DTRs permit treatment decisions to depend on patient characteristics, H k, at each stage k so the total number of DTRs is infinite. Some smart design is needed.

6 Learn from Artificial Intelligence

7 What is Reinforcement Learning in AI Reinforcement learning (RL) is a machine learning method in artificial intelligence. Its history dates back to decades ago when AI engineers tried to mimic learning process in human brains. Different from commonly known machine learning, RL is a learning method of interactions between agent and unknown environment with feedback loop. It consists of exploitation and exploration. Exploitation: apply already learnt decisions Exploration: trial and learn from new environment.

8 Link between RL and DTR They both involve sequential decisions. They share the same goal: find optimal decisions to maximize some value. Equivalence between DTR and RL terminologies: treatment == action state/covariate==state reward outcome==reward treatment rule==policy

9 How RL Designs Studies in AI In artificial intelligence (AI), a single agent (Robot) trains and learns by itself for a particular task, for example, play chess to win. A typical design for this learning is a trial and error learning process. The single agent tries many different episodes, each with a given policy. It leans from these episodes and also updates policies in new episodes.

10 How Different RL Should Be Used in Medical Trials Instead of training one single agent with many policies, we train many patients, each given potentially different treatments. Randomization is necessary to have sufficient representation and avoid selection for treatments. This is essentially the idea of Sequential Multiple Assignment Randomized Trial (SMART).

11 SMART

12 SMART SMART: Sequential Multiple Assignment Randomized Trial (Lavori & Dawson 2000, 2004; Murphy 2005) Patients are sequentially randomized at each critical decision stage. Randomization probability may depend on current states of patients. Practical SMART Adaptive Pharmacological and Behavioral treatments for ADHD; Sequenced Treatment Alternatives to Relieve Depression (STAR*D) ; CATIE for schizophrenia; ExTENd for alcohol dependence; Adaptive therapy for androgen independent prostate cancer

13 Example: Two-Stage SMART Study The study (Kasari et al., 2014) was designed to study communication intervention for minimally verbal children with autism. The study aimed to test the effect of SGD, each stage lasting 12 weeks. SGD: speech-generating device; (JASP+EMT): blended developmental/behavioral intervention The second stage had another 12 week follow-up. The study started with 61 eligible children and 46 completed both stages.

14 was conducted by an independent data-coordinating center. Diagram of The Autism Study parent-report and observation during study adminis tration of the NLS. FIGURE 1 Participant flow through trial. Note: JASPþEMT ¼ spoken mode of JASPER plus Enhanced Milieu Teaching; JASPþEMTþSGD ¼ spoken mode of JASPER plus Enhanced Milieu Teaching plus Speech Generating Device. Figure: SMART Design of Autism Study (Kasari et al. 2014) JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY VOLUME 53 NUMBER 6 JUNE

15 Advantages of SMART Research questions to be answered from a SMART Main effects of treatments What is the better initial treatment, JASP+EMT or JASP+EMT+SGD? What about the slow-responders: intensify or not? Effects of embedded DTR JASP+EMT intensify vs JASP+EMT+SGD intensify vs JASP+EM JASP+EMT+SGD Exploring optimal treatment strategy (deep tailoring) intensify or not in the second stage dependent on additional intermediate outcomes?

16 General Advantages Valid comparisons of different treatment options at different stages due to the virtue of randomization. Discover adaptive treatment strategies that are embedded in the SMART trial. Inform the development of adaptive and deeply tailored treatments (using potentially high-dimensional biomarkers).

17 Unbiased Value Estimation under SMART Recall the value function associated with D = (D 1,..., D K ) V(D) = E[R(D)] = E[R(D 1,..., D K )]. Let π k (a, h) be the randomization probability P(A k = a H k = h). Start from stage K: E[R(D 1,..., D k 1, D K )] { [ ]} = E E R(D 1,..., D K ) H K, A K = D K (H k ) { [ = E E R(D 1,..., D k 1, A K ) I(A K = D K (H K )) ]} H K π K (A K, H K ) [ = E R(D 1,..., D k 1, A K ) I(A ] K = D K (H K )). π K (A K, H K )

18 Continue to Prior Stages Now at stage K 1, we continue and repeat the same derivation to obtain E[R(D 1,..., D k 1, D K )] [ = E R(D 1,..., A K 1, A K ) I(A ] K 1 = D K 1 (H K 1 ), A K = D K (H K )). π K 1 (A K 1, H K 1 )π K (A K, H K ) Continue backwards till stage 1: V(D) = E [ R(A 1,..., A K 1, A K ) I(A 1 = D 1 (H 1 ), A K = D K (H K )) K k=1 π k(a k, H k ) [ ] = E R I(A 1 = D 1 (H 1 ), A K = D K (H K )) K k=1 π. k(a k, H k ) The value function for D can be estimated using the average reward outcomes from the patients whose treatments follow D, weighted by their randomization probabilities. ]

19 An Alternative Interpretation In SMART, a particular treatment assignment takes probability K π k (A k, H k ). k=1 If we run a single-arm trial for a given D, the treatment assignment is K k=1 I(A k = D k (H k )). To use SMART to estimate the expected reward in the D-specified trial, important sampling theory gives V(D) = E [ R I(A 1 = D 1 (H 1 ), A K = D K (H K )) K k=1 π k(a k, H k ) In conclusion, SMART provides unbiased estimation and comparisons between DTRs so potentially leads to valid estimation of optimal DTRs. ].

20 Demo of SMART Trial

21 NSCLC trial In treating advanced non-small cell lung cancer, patients typically experience two or more lines of treatment, and many studies demonstrate that three lines of treatment can improve survival for patients. 1st-line 2nd-line 3rd-line

22 1st-line therapy Scagliotti et al., 2008 { Cisplatin + Gemcitabine Cisplatin + Pemetrexed Sandler et al., 2006 { Paclitaxel + Carboplatin Paclitaxel + Carboplatin + Bevacizumab (Avastin) Pirker et al., 2008 { Cisplatin + Vinorelbine Cisplatin + Vinorelbine + Cetuximab (Erbitux)

23 2nd-line therapy Docetaxel Approved regimens Pemetrexed Erlotinib (approved for third-line, too) Timing issue Immediate or delayed docetaxel? (Fidias et al., 2007; Ciuleanu et al., 2008) Comparison of PFS suggests immediate, but OS did not reach significant difference. Merits of pemetrexed and erlotinib remain unclear.

24 Important clinical questions 1. Among many approved 1st-line treatments, what treatment to administer? 2. Then, at the end of the 1st-line treatment Among approved 2nd-line treatments, what treatment to administer? When to begin the 2nd-line of treatment? Immediate Progression Death 1st-line 2nd-line Possible treatments Possible treatments and initial timings

25 Reinforcement Learning Has a well-known framework for optimizing sequence of treatments in an evolving, time-varying system. Discovers individualized treatment regimens for cancer. Selects treatments that improve outcomes even when the relationship between treatments and outcomes is not fully known. Can evaluate treatments based on immediate and long-term effects. Applications: Behavioral disorders (Pineau et al., 2007). Sequential multiple assignment randomized trial (SMART) designs for developing dynamic treatment regimes (Murphy et al., 2007).

26 Clinical Reinforcement Trial Drafted Protocol 1. A finite, reasonably small set of decision times is identified. 2. For each decision time, a set of possible treatments to be randomized is identified. 3. A utility function is identified which can be assessed at each time point. 4. Patients are then recruited into the study and randomized to the treatment set under the protocol restrictions at each decision point. 5. The patient data is collected and Q-learning is applied, in combination with SVR applied, to estimate the optimal treatment rule as a function of patient variables and biomarkers, at each decision time.

27 Reinforcement Learning Trial Y D =0,δ =1,T 0 = T 1 t 1 T1 Y D =0,δ =0,T 0 = C t 1 C (t 2 + T M) T P T2 Y D =1,δ =1,T 0 = t 2 + T 2 t 1 t2 T 1 = T D t 2, Y D = I(T D C t 2 ) Y D =1,δ =0,T 0 = C (t 2 + T M) T P t 1 t2 T 2 = (T D t 2 )I(T D t 2 ) = (T D t 2 )I(T 1 = t 2 ) C 2 = (C t 2 )I(C t 2 ) T D = T 1 + T 2, T 0 = T D C = T 1 C + Y D (T 2 C 2 ) C

28 Learning Algorithm 1. At t 1, (w 1, m 1, d 1, T 1 C, ) n i=1 ; at t 2, (w 2, m 2, d 2, T M, T 2 C 2, ) n j=1, where n n. 2. Q 2, the expected reward at stage 2, is estimated using regression methods with outcome T 2 C Q 1, the expected optimal reward at stage 1, is estimated using regression methods with outcome T 1 C + I(T 1 = t 2 ) max d 2,T M Q In step 2 & 3, regression methods can include machine learning algorithms (SVR). 5. Given Q 1 (ˆθ 1 ) and Q 2 (ˆθ 2 ), compute D 1 and D 2 by maximizing Q 1 and Q 2, respectively.

29 Simulation Settings 1st-line treatment regimens: A 1 or A 2 2nd-line treatment regimens: A 3 or A 4 initiation time for 2nd-line treatment: random time between 2.8 and 4.8 months from 1st-line treatment prognostic factors: W t (quality of life), M t (tumor size) four groups are generated along with their optimal treatment regimens

30 Simulation scenario table Table 1: The scenarios studied in the simulation. Sample size = 100/group. Group State Variables Status Timing Optimal Regimen W 1 N(0.25, σ 2 ) 1 M 1 N(0.75, σ 2 ) W 1 M 1 A 1 A W 1 N(0.75, σ 2 ) M 1 N(0.75, σ 2 ) W 1 N(0.25, σ 2 ) M 1 N(0.25, σ 2 ) W 1 N(0.75, σ 2 ) M 1 N(0.25, σ 2 ) W 1 M 1 A 1 A 4 1 W 1 M 1 A 2 A 3 3 W 1 M 1 A 2 A 4 2

31 Performance of the optimal regimen A 1A 31 A 1A 32 A 1A 33 A 1A 41 A 1A 42 A 1A 43 A 2A 31 A 2A 32 A 2A 33 A 2A 41 A 2A 42 A 2A 43 optimal Overall Survival

32 Performance of the optimal regimen Table: Comparisons between true optimal regimens and estimated optimal regimens for overall survival (month). Each reinforcement trial is of size 100/group with 10 simulation runs. The confirmatory trial is of size 100/group. Optimal True Predicted survival Group regimen survival Min Mean Max 1 A 1 A A 1 A A 2 A A 2 A Average

33 A 1A 32 A 1A 41 A 2A 33 A 2A 42 Frequency Frequency Frequency Frequency Group 1 (92%, 100%) Group 2 (100%, 100%) Group 3 (100%, 100%) Group 4 (100%, 81%)

34 Sensitivity of the predicted survival to the sample size Overall Survival Sample Size for Each Group

35 ɛ-svr-c performance Predicted Overall Survival Predicted Overall Survival Predicted Overall Survival None 25% delete None 50% delete None 75% delete (a) (b) (c)

36 Summary Identified optimal treatment strategies tailored to proper subpopulation of NSCLC patients. Solved timing problem of initiating second-line therapy in NSCLC. Handled right censored data.

37 Design Limitations

38 Practical limitations of SMART Operation cost of administrating multiple stage studies and multiple treatments is high. The length of trial period is long (March et al. 2010). Study dropout or compliance is common even in regular RCTs: In the CATIE study, 705 of 1460 patients stayed for the entire 18 months of the study. In ExTENd, the drop-out rate was 17% (52 out of 302) in the first-stage treatment and an additional 13% (41 out of 302) during the second stage.

Lecture 1: Introduction to Personalized Medicine. Donglin Zeng, Department of Biostatistics, University of North Carolina

Lecture 1: Introduction to Personalized Medicine. Donglin Zeng, Department of Biostatistics, University of North Carolina Lecture 1: Introduction to Personalized Medicine Personalized Medicine A Quick View Personalized Medicine is a general medical paradigm referring to systematic use of individual patient information to

More information

BIOSTATISTICAL METHODS

BIOSTATISTICAL METHODS BIOSTATISTICAL METHODS FOR TRANSLATIONAL & CLINICAL RESEARCH ADAPTIVE DESIGN: SMART TRIALS ADAPTIVE CLINICAL TRIALS An adaptive clinical trial is a clinical trial that evaluates a treatment (or treatments)

More information

SMART Clinical Trial Designs for Dynamic Treatment Regimes

SMART Clinical Trial Designs for Dynamic Treatment Regimes SMART Clinical Trial Designs for Dynamic Treatment Regimes Bibhas Chakraborty Centre for Quantitative Medicine, Duke-NUS Graduate Medical School, Singapore bibhas.chakraborty@duke-nus.edu.sg MCP Conference

More information

University of North Carolina at Chapel Hill

University of North Carolina at Chapel Hill University of North Carolina at Chapel Hill The University of North Carolina at Chapel Hill Department of Biostatistics Technical Report Series Year 2009 Paper 13 Reinforcement Learning Strategies for

More information

Introduction to Design and Analysis of SMARTs

Introduction to Design and Analysis of SMARTs Introduction to Design and Analysis of SMARTs Michael R. Kosorok September 2013 Outline What are SMARTs and Why Do We Care? Necessary Non-Standard Analytical Tools Some Illustrative Examples Overview of

More information

Automatic learning of adaptive treatment strategies. using kernel-based reinforcement learning

Automatic learning of adaptive treatment strategies. using kernel-based reinforcement learning using kernel-based reinforcement learning Presented by:, School of Computer Science, McGill University Joint work with: Marc G. Bellemare, Adrian Ghizaru, Zaid Zawaideh, Susan Murphy (Univ. Michigan),

More information

Maintenance Therapy for Advanced NSCLC: When, What, Why & What s Left After Post-Maintenance Relapse?

Maintenance Therapy for Advanced NSCLC: When, What, Why & What s Left After Post-Maintenance Relapse? Maintenance Therapy for Advanced NSCLC: When, What, Why & What s Left After Post-Maintenance Relapse? Mark A. Socinski, MD Professor of Medicine Multidisciplinary Thoracic Oncology Program Lineberger Comprehensive

More information

Lecture 13: Finding optimal treatment policies

Lecture 13: Finding optimal treatment policies MACHINE LEARNING FOR HEALTHCARE 6.S897, HST.S53 Lecture 13: Finding optimal treatment policies Prof. David Sontag MIT EECS, CSAIL, IMES (Thanks to Peter Bodik for slides on reinforcement learning) Outline

More information

Lecture 10: Learning Optimal Personalized Treatment Rules Under Risk Constraint

Lecture 10: Learning Optimal Personalized Treatment Rules Under Risk Constraint Lecture 10: Learning Optimal Personalized Treatment Rules Under Risk Constraint Introduction Consider Both Efficacy and Safety Outcomes Clinician: Complete picture of treatment decision making involves

More information

Considerations in Optimizing Personalized Treatments: Estimation and Evaluation in Light of Benefit and Risk

Considerations in Optimizing Personalized Treatments: Estimation and Evaluation in Light of Benefit and Risk Considerations in Optimizing Personalized Treatments: Estimation and Evaluation in Light of Benefit and Risk Yuanjia Wang, Ph.D. Department of Biostatistics, Mailman School of Public Health & Division

More information

Overall survival in non-small cell lung cancer what is clinically meaningful?

Overall survival in non-small cell lung cancer what is clinically meaningful? Editorial Overall survival in non-small cell lung cancer what is clinically meaningful? Klaus Fenchel 1, Ludger Sellmann 2, Wolfram C. M. Dempke 3 1 Medical School Hamburg (MSH), Hamburg, Germany; 2 University

More information

Management Guidelines and Targeted Therapies in Metastatic Non-Small Cell Lung Cancer: An Oncologist s Perspective

Management Guidelines and Targeted Therapies in Metastatic Non-Small Cell Lung Cancer: An Oncologist s Perspective Management Guidelines and Targeted Therapies in Metastatic Non-Small Cell Lung Cancer: An Oncologist s Perspective Julie R. Brahmer, M.D. Associate Professor of Oncology The Sidney Kimmel Comprehensive

More information

Erlotinib for the first-line treatment of EGFR-TK mutation positive non-small cell lung cancer

Erlotinib for the first-line treatment of EGFR-TK mutation positive non-small cell lung cancer ERRATUM Erlotinib for the first-line treatment of EGFR-TK mutation positive non-small cell lung cancer This report was commissioned by the NIHR HTA Programme as project number 11/08 Completed 6 th January

More information

Adaptive Treatment of Epilepsy via Batch Mode Reinforcement Learning

Adaptive Treatment of Epilepsy via Batch Mode Reinforcement Learning Adaptive Treatment of Epilepsy via Batch Mode Reinforcement Learning Arthur Guez, Robert D. Vincent and Joelle Pineau School of Computer Science, McGill University Massimo Avoli Montreal Neurological Institute

More information

Adaptive Interventions Treatment Modelling and Regimen Optimization Using Sequential Multiple Assignment Randomized Trials (Smart) and Q- Learning

Adaptive Interventions Treatment Modelling and Regimen Optimization Using Sequential Multiple Assignment Randomized Trials (Smart) and Q- Learning South Dakota State University Open PRAIRIE: Open Public Research Access Institutional Repository and Information Exchange Electronic Theses and Dissertations 2018 Adaptive Interventions Treatment Modelling

More information

Reinforcement Learning

Reinforcement Learning Reinforcement Learning Michèle Sebag ; TP : Herilalaina Rakotoarison TAO, CNRS INRIA Université Paris-Sud Nov. 9h, 28 Credit for slides: Richard Sutton, Freek Stulp, Olivier Pietquin / 44 Introduction

More information

Erlotinib (Tarceva) for non small cell lung cancer advanced or metastatic maintenance monotherapy

Erlotinib (Tarceva) for non small cell lung cancer advanced or metastatic maintenance monotherapy Erlotinib (Tarceva) for non small cell lung cancer advanced or metastatic maintenance monotherapy September 2008 This technology summary is based on information available at the time of research and a

More information

Developing Adaptive Health Interventions

Developing Adaptive Health Interventions Developing Adaptive Health Interventions Getting SMART Daniel Almirall 1,2 Susan A Murphy 1,2,3 1 Institute for Social Research, Survey Research Center, University of Michigan 2 The Methodology Center,

More information

Bayesian Nonparametric Methods for Precision Medicine

Bayesian Nonparametric Methods for Precision Medicine Bayesian Nonparametric Methods for Precision Medicine Brian Reich, NC State Collaborators: Qian Guan (NCSU), Eric Laber (NCSU) and Dipankar Bandyopadhyay (VCU) University of Illinois at Urbana-Champaign

More information

An Introduction to Dynamic Treatment Regimes

An Introduction to Dynamic Treatment Regimes An Introduction to Dynamic Treatment Regimes Marie Davidian Department of Statistics North Carolina State University http://www4.stat.ncsu.edu/davidian 1/64 Dynamic Treatment Regimes Webinar Outline What

More information

Maintenance paradigm in non-squamous NSCLC

Maintenance paradigm in non-squamous NSCLC Maintenance paradigm in non-squamous NSCLC L. Paz-Ares Hospital Universitario Virgen del Rocío Sevilla Agenda Theoretical basis The data The comparisons Agenda Theoretical basis The data The comparisons

More information

First-line treatment of patients with advanced or metastatic squamous non-small cell lung cancer: systematic review and network meta-analysis

First-line treatment of patients with advanced or metastatic squamous non-small cell lung cancer: systematic review and network meta-analysis Original Article First-line treatment of patients with advanced or metastatic squamous non-small cell lung cancer: systematic review and network meta-analysis Lisa M. Hess, Amy M. DeLozier, Fanni Natanegara,

More information

Comparing Dynamic Treatment Regimes Using Repeated-Measures Outcomes: Modeling Considerations in SMART Studies

Comparing Dynamic Treatment Regimes Using Repeated-Measures Outcomes: Modeling Considerations in SMART Studies Comparing Dynamic Treatment Regimes Using Repeated-Measures Outcomes: Modeling Considerations in SMART Studies Xi Lu The University of Michigan Connie Kasari University of California Los Angeles Inbal

More information

Thoracic and head/neck oncology new developments

Thoracic and head/neck oncology new developments Thoracic and head/neck oncology new developments Goh Boon Cher Department of Hematology-Oncology National University Cancer Institute of Singapore Research Clinical Care Education Scope Lung cancer Screening

More information

Heather Wakelee, M.D.

Heather Wakelee, M.D. Heather Wakelee, M.D. Assistant Professor of Medicine, Oncology Stanford University Sponsored by Educational Grant Support from Adjuvant (Post-Operative) Lung Cancer Chemotherapy Heather Wakelee, M.D.

More information

Artificial Intelligence Lecture 7

Artificial Intelligence Lecture 7 Artificial Intelligence Lecture 7 Lecture plan AI in general (ch. 1) Search based AI (ch. 4) search, games, planning, optimization Agents (ch. 8) applied AI techniques in robots, software agents,... Knowledge

More information

Personalized maintenance therapy in advanced non-small cell lung cancer

Personalized maintenance therapy in advanced non-small cell lung cancer China Lung Cancer Research Highlight Personalized maintenance therapy in advanced non-small cell lung cancer Kazuhiro Asami, Kyoichi Okishio, Tomoya Kawaguchi, Shinji Atagi Department of Clinical Oncology,

More information

1st-line Chemotherapy for Advanced disease

1st-line Chemotherapy for Advanced disease SESSION 3: ADVANCED NSCLC 1st-line Chemotherapy for Advanced disease JY DOUILLARD MD PhD Professor Emeritus in Medical Oncology Chief Medical Officer (CMO) ESMO Lugano CH Percent Survival HISTORICAL BASIS

More information

Artificial Intelligence. Intelligent Agents

Artificial Intelligence. Intelligent Agents Artificial Intelligence Intelligent Agents Agent Agent is anything that perceives its environment through sensors and acts upon that environment through effectors. Another definition later (Minsky) Humans

More information

CS 771 Artificial Intelligence. Intelligent Agents

CS 771 Artificial Intelligence. Intelligent Agents CS 771 Artificial Intelligence Intelligent Agents What is AI? Views of AI fall into four categories 1. Thinking humanly 2. Acting humanly 3. Thinking rationally 4. Acting rationally Acting/Thinking Humanly/Rationally

More information

Sequential, Multiple Assignment, Randomized Trials

Sequential, Multiple Assignment, Randomized Trials Sequential, Multiple Assignment, Randomized Trials Module 2 Experimental Design and Analysis Methods for Developing Adaptive Interventions: Getting SMART Daniel Almirall, Ahnalee Brincks, Billie Nahum-Shani

More information

NSCLC: Terapia medica nella fase avanzata. Paolo Bidoli S.C. Oncologia Medica H S. Gerardo Monza

NSCLC: Terapia medica nella fase avanzata. Paolo Bidoli S.C. Oncologia Medica H S. Gerardo Monza NSCLC: Terapia medica nella fase avanzata Paolo Bidoli S.C. Oncologia Medica H S. Gerardo Monza First-line Second-line Third-line Not approved CT AND SILENT APPROVAL Docetaxel 1999 Paclitaxel Gemcitabine

More information

Exploring Personalized Therapy for First Line Treatment of Advanced Non-Small Cell Lung Cancer (NSCLC)

Exploring Personalized Therapy for First Line Treatment of Advanced Non-Small Cell Lung Cancer (NSCLC) Exploring Personalized Therapy for First Line Treatment of Advanced Non-Small Cell Lung Cancer (NSCLC) Suresh S. Ramalingam, MD Director of Thoracic Oncology Associate Professor Emory University Atlanta,

More information

NIH Public Access Author Manuscript J Am Stat Assoc. Author manuscript.

NIH Public Access Author Manuscript J Am Stat Assoc. Author manuscript. NIH Public Access Author Manuscript SMART Design Issues and the Consideration of Opposing Outcomes: Discussion of Evaluation of Viable Dynamic Treatment Regimes in a Sequentially Randomized Trial of Advanced

More information

1 st line chemotherapy and contribution of targeted agents in non-driver addicted NSCLC

1 st line chemotherapy and contribution of targeted agents in non-driver addicted NSCLC 1 st line chemotherapy and contribution of targeted agents in non-driver addicted NSCLC Dr Ross Soo, FRACP National University Cancer Institute, Singapore National University Health System Cancer Science

More information

COMBINING SINGLE CASE DESIGN AND GROUP EXPERIMENTAL DESIGN RESEARCH

COMBINING SINGLE CASE DESIGN AND GROUP EXPERIMENTAL DESIGN RESEARCH COMBINING SINGLE CASE DESIGN AND GROUP EXPERIMENTAL DESIGN RESEARCH Ann P. Kaiser, PhD Vanderbilt University IES SCD June 2018 1 Today s Talk Overview of research on Enhanced Milieu Teaching (EMT) Illustrate

More information

Benefit Risk Analysis Of Decision-Making: Oncology

Benefit Risk Analysis Of Decision-Making: Oncology Benefit Risk Analysis Of Decision-Making: Oncology G.K. Raju, Ph.D. December 13 th 2016 Outline Background Approach Application to Oncology Non-Small Cell Lung Cancer Learnings & Ongoing Work Acknowledgements

More information

On Adaptive Interventions and SMART

On Adaptive Interventions and SMART On Adaptive Interventions and SMART Daniel Almirall; Inbal (Billie) Nahum- Shani; Shawna Smith School of Nursing University of Michigan February 3, 2016 1 Outline Adaptive Intervention (AIs) What are AIs?

More information

Maintenance therapies in advanced non-small-cell lung cancer

Maintenance therapies in advanced non-small-cell lung cancer Review Maintenance therapies in advanced non-small-cell lung cancer Advanced non-small-cell lung cancer is treated with upfront platinum doublet chemotherapy, which produces moderate survival improvements.

More information

National Horizon Scanning Centre. Erlotinib (Tarceva) in combination with bevacizumab for advanced or metastatic non-small cell lung cancer

National Horizon Scanning Centre. Erlotinib (Tarceva) in combination with bevacizumab for advanced or metastatic non-small cell lung cancer Erlotinib (Tarceva) in combination with bevacizumab for advanced or metastatic non-small cell lung cancer This technology summary is based on information available at the time of research and a limited

More information

Sequential Decision Making

Sequential Decision Making Sequential Decision Making Sequential decisions Many (most) real world problems cannot be solved with a single action. Need a longer horizon Ex: Sequential decision problems We start at START and want

More information

Early Interventions for ASD: State of the Science

Early Interventions for ASD: State of the Science Early Interventions for ASD: State of the Science Connie Kasari, Ph.D. University of California, Los Angeles Malaysian Regional Conference April 22-23 Today What is the evidence for current early interventions?

More information

Getting SMART about Adapting Interventions. S.A. Murphy Early Childhood Interventions Inaugural Conference 04/21/12

Getting SMART about Adapting Interventions. S.A. Murphy Early Childhood Interventions Inaugural Conference 04/21/12 Getting SMART about Adapting Interventions S.A. Murphy Early Childhood Interventions Inaugural Conference 04/21/12 Adaptive Interventions are individually tailored sequences of interventions, with treatment

More information

Lecture II: Difference in Difference. Causality is difficult to Show from cross

Lecture II: Difference in Difference. Causality is difficult to Show from cross Review Lecture II: Regression Discontinuity and Difference in Difference From Lecture I Causality is difficult to Show from cross sectional observational studies What caused what? X caused Y, Y caused

More information

Systemic Treatment for Patients with Advanced Non-Small Cell Lung Cancer P.M. Ellis, E.T. Vella, Y.C. Ung and the Lung Cancer Disease Site Group

Systemic Treatment for Patients with Advanced Non-Small Cell Lung Cancer P.M. Ellis, E.T. Vella, Y.C. Ung and the Lung Cancer Disease Site Group A Quality Initiative of the Program in Evidence-Based Care (PEBC), Cancer Care Ontario (CCO) Systemic Treatment for Patients with Advanced Non-Small Cell Lung Cancer P.M. Ellis, E.T. Vella, Y.C. Ung and

More information

Technology appraisal guidance Published: 29 June 2011 nice.org.uk/guidance/ta227

Technology appraisal guidance Published: 29 June 2011 nice.org.uk/guidance/ta227 Erlotinib monotherapy for maintenance treatment of non-small-cell lung cancer Technology appraisal guidance Published: 29 June 2011 nice.org.uk/guidance/ta227 NICE 2018. All rights reserved. Subject to

More information

Targeted Agents as Maintenance Therapy. Karen Kelly, MD Professor of Medicine UC Davis Cancer Center

Targeted Agents as Maintenance Therapy. Karen Kelly, MD Professor of Medicine UC Davis Cancer Center Targeted Agents as Maintenance Therapy Karen Kelly, MD Professor of Medicine UC Davis Cancer Center Disclosures Genentech Advisory Board Maintenance Therapy Defined Treatment Non-Progressing Patients Drug

More information

To help doctors give their patients the best possible care, the American. What to Know

To help doctors give their patients the best possible care, the American. What to Know Patient Information Resources from ASCO What to Know ASCO s Guideline on Chemotherapy for Stage IV Non-Small Cell Lung Cancer SEPTEMBER 2011 KEY MESSAGES Chemotherapy for stage IV non-small cell lung cancer

More information

Getting SMART about Adaptive Interventions: A Conceptual Introduction Rona L. Levy With many thanks to Daniel Almirall

Getting SMART about Adaptive Interventions: A Conceptual Introduction Rona L. Levy With many thanks to Daniel Almirall Getting MART about Adaptive Interventions: A Conceptual Introduction Rona L. Levy rlevy@uw.edu With many thanks to Daniel Almirall Institute for ocial Research, University of Michigan Adaptive Interventions

More information

Maintenance therapy in advanced non-small cell lung cancer. Egbert F. Smit MD PhD Dept Thoracic Oncology Netherlands Cancer Institute

Maintenance therapy in advanced non-small cell lung cancer. Egbert F. Smit MD PhD Dept Thoracic Oncology Netherlands Cancer Institute Maintenance therapy in advanced non-small cell lung cancer. Egbert F. Smit MD PhD Dept Thoracic Oncology Netherlands Cancer Institute e.smit@nki.nl Evolution of front line therapy in NSCLC unselected pts

More information

Osimertinib (lung cancer)

Osimertinib (lung cancer) IQWiG Reports Commission No. A16-14 Osimertinib (lung cancer) Benefit assessment according to 35a Social Code Book V 1 Extract 1 Translation of Sections 2.1 to 2.6 of the dossier assessment Osimertinib

More information

ERA: Architectures for Inference

ERA: Architectures for Inference ERA: Architectures for Inference Dan Hammerstrom Electrical And Computer Engineering 7/28/09 1 Intelligent Computing In spite of the transistor bounty of Moore s law, there is a large class of problems

More information

Design for Targeted Therapies: Statistical Considerations

Design for Targeted Therapies: Statistical Considerations Design for Targeted Therapies: Statistical Considerations J. Jack Lee, Ph.D. Department of Biostatistics University of Texas M. D. Anderson Cancer Center Outline Premise General Review of Statistical Designs

More information

A Case Review: Treatment-Naïve Patient with Advanced NSCLC: Smoker with Metastatic Squamous Cell Tumor

A Case Review: Treatment-Naïve Patient with Advanced NSCLC: Smoker with Metastatic Squamous Cell Tumor Transcript Details This is a transcript of a continuing medical education (CME) activity accessible on the ReachMD network. Additional media formats for the activity and full activity details (including

More information

MAINTENANCE TREATMENT CHEMO MAINTENANCE OR TARGETED OF BOTH? Martin Reck Department of Thoracic Oncology LungenClinic Grosshansdorf

MAINTENANCE TREATMENT CHEMO MAINTENANCE OR TARGETED OF BOTH? Martin Reck Department of Thoracic Oncology LungenClinic Grosshansdorf MAINTENANCE TREATMENT CHEMO MAINTENANCE OR TARGETED OF BOTH? Martin Reck Department of Thoracic Oncology LungenClinic Grosshansdorf OUTLINE Background and Concept Switch Maintenance Continuation Maintenance

More information

CS324-Artificial Intelligence

CS324-Artificial Intelligence CS324-Artificial Intelligence Lecture 3: Intelligent Agents Waheed Noor Computer Science and Information Technology, University of Balochistan, Quetta, Pakistan Waheed Noor (CS&IT, UoB, Quetta) CS324-Artificial

More information

Management of advanced non small cell lung cancer

Management of advanced non small cell lung cancer Management of advanced non small cell lung cancer Jean-Paul Sculier Intensive Care & Thoracic Oncology Institut Jules Bordet Université Libre de Bruxelles (ULB) www.pneumocancero.com Declaration No conflict

More information

Background 1. Comparative effectiveness of nintedanib

Background 1. Comparative effectiveness of nintedanib NCPE report on the cost effectiveness of nintedanib (Vargatef ) in combination with docetaxel for the treatment of adult patients with locally advanced, metastatic or locally recurrent non-small cell lung

More information

Chapter 2: Intelligent Agents

Chapter 2: Intelligent Agents Chapter 2: Intelligent Agents Outline Last class, introduced AI and rational agent Today s class, focus on intelligent agents Agent and environments Nature of environments influences agent design Basic

More information

Comparisons of Dynamic Treatment Regimes using Observational Data

Comparisons of Dynamic Treatment Regimes using Observational Data Comparisons of Dynamic Treatment Regimes using Observational Data Bryan Blette University of North Carolina at Chapel Hill 4/19/18 Blette (UNC) BIOS 740 Final Presentation 4/19/18 1 / 15 Overview 1 Motivation

More information

Challenges in Developing Learning Algorithms to Personalize mhealth Treatments

Challenges in Developing Learning Algorithms to Personalize mhealth Treatments Challenges in Developing Learning Algorithms to Personalize mhealth Treatments JOOLHEALTH Bar-Fit Susan A Murphy 01.16.18 HeartSteps SARA Sense 2 Stop Continually Learning Mobile Health Intervention 1)

More information

Learning Optimal Individualized Treatment Rules from Electronic Health Record Data

Learning Optimal Individualized Treatment Rules from Electronic Health Record Data 2016 IEEE International Conference on Healthcare Informatics Learning Optimal Individualized Treatment Rules from Electronic Health Record Data Yuanjia Wang, Peng Wu, Ying Liu Department of Biostatistics

More information

60 minutes. This is the 4 th module of a 6 module Seminar on experimental designs for building optimal adaptive health interventions.

60 minutes. This is the 4 th module of a 6 module Seminar on experimental designs for building optimal adaptive health interventions. 60 minutes This is the 4 th module of a 6 module Seminar on experimental designs for building optimal adaptive health interventions. By now, you know what an ATS is. You have discussed why they are important

More information

CANCER TREATMENT REGIMENS

CANCER TREATMENT REGIMENS CANCER TREATMENT S Lung Cancer The selection, dosing, and administration of anticancer agents and the management of associated toxicities are complex. Drug dose modifications and schedule and initiation

More information

Strategies for handling missing data in randomised trials

Strategies for handling missing data in randomised trials Strategies for handling missing data in randomised trials NIHR statistical meeting London, 13th February 2012 Ian White MRC Biostatistics Unit, Cambridge, UK Plan 1. Why do missing data matter? 2. Popular

More information

Maintenance Therapy for Advanced NSCLC: Which Patients, Which Approach?

Maintenance Therapy for Advanced NSCLC: Which Patients, Which Approach? Maintenance Therapy for Advanced NSCLC: Which Patients, Which Approach? Mark A. Socinski, MD Visiting Professor of Medicine and Thoracic Surgery Director, Lung Cancer Section, Division of Hematology/Oncology

More information

Combined Modality Therapy State of the Art. Everett E. Vokes The University of Chicago

Combined Modality Therapy State of the Art. Everett E. Vokes The University of Chicago Combined Modality Therapy State of the Art Everett E. Vokes The University of Chicago What we Know Some patients are cured (20%) Induction and concurrent chemoradiotherapy are each superior to radiotherapy

More information

doi: /theoncologist originally published online February 3, 2009

doi: /theoncologist originally published online February 3, 2009 Potential Treatment Options After First-Line Chemotherapy for Advanced NSCLC: Maintenance Treatment or Early Second-Line? Cesare Gridelli, Paolo Maione, Antonio Rossi, Marianna Luciana Ferrara, Maria Anna

More information

Background Comparative effectiveness of nivolumab

Background Comparative effectiveness of nivolumab NCPE report on the cost effectiveness of nivolumab (Opdivo ) for the treatment of locally advanced or metastatic squamous non-small cell lung cancer after prior chemotherapy in adults. The NCPE has issued

More information

A Vision-based Affective Computing System. Jieyu Zhao Ningbo University, China

A Vision-based Affective Computing System. Jieyu Zhao Ningbo University, China A Vision-based Affective Computing System Jieyu Zhao Ningbo University, China Outline Affective Computing A Dynamic 3D Morphable Model Facial Expression Recognition Probabilistic Graphical Models Some

More information

Lung Cancer Epidemiology. AJCC Staging 6 th edition

Lung Cancer Epidemiology. AJCC Staging 6 th edition Surgery for stage IIIA NSCLC? Sometimes! Anne S. Tsao, M.D. Associate Professor Director, Mesothelioma Program Director, Thoracic Chemo-Radiation Program May 7, 2011 The University of Texas MD ANDERSON

More information

Câncer de Pulmão Não Pequenas Células

Câncer de Pulmão Não Pequenas Células Câncer de Pulmão Não Pequenas Células Carboplatina + Paclitaxel Paclitaxel: 200mg/m 2 IV D1 Carboplatina: AUC 6 IV D1 a cada 21 dias X 4 ciclos Ref. (1) Vinorelbina + Cisplatina Vinorelbina: 25mg/m 2 IV

More information

60 minutes Give examples of SMARTs that are completed or in the field o ASD, child ADHD, women who are pregnant and abuse substances, adult alcohol

60 minutes Give examples of SMARTs that are completed or in the field o ASD, child ADHD, women who are pregnant and abuse substances, adult alcohol 60 minutes Give examples of SMARTs that are completed or in the field o ASD, child ADHD, women who are pregnant and abuse substances, adult alcohol use, depression Discuss the variety of rationales underlying

More information

Intelligent Agents. Chapter 2 ICS 171, Fall 2009

Intelligent Agents. Chapter 2 ICS 171, Fall 2009 Intelligent Agents Chapter 2 ICS 171, Fall 2009 Discussion \\Why is the Chinese room argument impractical and how would we have to change the Turing test so that it is not subject to this criticism? Godel

More information

Introduction: Statistics and Engineering

Introduction: Statistics and Engineering Introduction: Statistics and Engineering STAT:2020 Probability and Statistics for Engineering and Physical Sciences Week 1 - Lecture 1 Book Sections 1.1-1.2.4, 1.3: Introduction 1 / 13 Where do engineering

More information

PERIOPERATIVE TREATMENT OF NON SMALL CELL LUNG CANCER. Virginie Westeel Chest Disease Department University Hospital Besançon, France

PERIOPERATIVE TREATMENT OF NON SMALL CELL LUNG CANCER. Virginie Westeel Chest Disease Department University Hospital Besançon, France PERIOPERATIVE TREATMENT OF NON SMALL CELL LUNG CANCER Virginie Westeel Chest Disease Department University Hospital Besançon, France LEARNING OBJECTIVES 1. To understand the potential of perioperative

More information

BRAIN MISSION Understand - fix - enhance

BRAIN MISSION Understand - fix - enhance THE EUROPEAN BRAIN COUNCIL PRESENTS BRAIN MISSION Understand - fix - enhance THE space race OF the 21st century A major societal challenge The brain is the most complex human organ. It provides and controls

More information

Clinical Trials. Ovarian Cancer

Clinical Trials. Ovarian Cancer 1.0 0.8 0.6 0.4 0.2 0.0 < 65 years old 65 years old Events Censored Total 128 56 184 73 35 108 0 12 24 36 48 60 72 84 27-10-2012 Ovarian Cancer Stuart M. Lichtman, MD Attending Physician 65+ Clinical Geriatric

More information

Intelligent Agents. Soleymani. Artificial Intelligence: A Modern Approach, Chapter 2

Intelligent Agents. Soleymani. Artificial Intelligence: A Modern Approach, Chapter 2 Intelligent Agents CE417: Introduction to Artificial Intelligence Sharif University of Technology Spring 2016 Soleymani Artificial Intelligence: A Modern Approach, Chapter 2 Outline Agents and environments

More information

Slide 1. Slide 2 Maintenance Therapy Options. Slide 3. Maintenance Therapy in the Management of Non-Small Cell Lung Cancer. Maintenance Chemotherapy

Slide 1. Slide 2 Maintenance Therapy Options. Slide 3. Maintenance Therapy in the Management of Non-Small Cell Lung Cancer. Maintenance Chemotherapy Slide 1 Maintenance Therapy in the Management of Non-Small Cell Lung Cancer Frances A Shepherd, MD FRCPC Scott Taylor Chair in Lung Cancer Research Princess Margaret Hospital, Professor of Medicine, University

More information

Introduction to Machine Learning. Katherine Heller Deep Learning Summer School 2018

Introduction to Machine Learning. Katherine Heller Deep Learning Summer School 2018 Introduction to Machine Learning Katherine Heller Deep Learning Summer School 2018 Outline Kinds of machine learning Linear regression Regularization Bayesian methods Logistic Regression Why we do this

More information

1st line chemotherapy and contribution of targeted agents

1st line chemotherapy and contribution of targeted agents ESMO PRECEPTORSHIP PROGRAMME NON-SM ALL-CELL LUNG CANCER 1st line chemotherapy and contribution of targeted agents Yi-Long Wu Guangdong Lung Cancer Institute Guangdong General Hospital Guangdong Academy

More information

Adapting Early Communication Intervention to the Phenotypic Characteristics of Young Children with Language Impairment Part II

Adapting Early Communication Intervention to the Phenotypic Characteristics of Young Children with Language Impairment Part II Adapting Early Communication Intervention to the Phenotypic Characteristics of Young Children with Language Impairment Part II Ann P. Kaiser Vanderbilt University 1 Today s Talk Building a new generation

More information

LUNG CANCER TREATMENT: AN OVERVIEW

LUNG CANCER TREATMENT: AN OVERVIEW LUNG CANCER TREATMENT: AN OVERVIEW KONSTANTINOS N. SYRIGOS, M.D., Ph.D. Αναπλ. Καθηγητής Παθολογίας-Ογκολογίας, Ιατρικής Σχολής Αθηνών. Διευθυντής Ογκολογικής Μονάδας, Νοσ. «Η Σωτηρία». Visiting Professor

More information

ASCO Highlights Lung Cancer

ASCO Highlights Lung Cancer ASCO Highlights Lung Cancer Anne S. Tsao, M.D. Director, Mesothelioma Program Assistant Professor July 11, 2009 The University of Texas MD ANDERSON CANCER CENTER Department of Thoracic/Head & Neck Medical

More information

Adaptive Interventions: What are they? Why do we need them? and How can we study them?

Adaptive Interventions: What are they? Why do we need them? and How can we study them? Adaptive Interventions: What are they? Why do we need them? and How can we study them? Daniel Almirall, Inbal Nahum-Shani, Susan A. Murphy Survey Research Center, Institute for Social Research University

More information

Antiangiogenici in combinazione a chemioterapia in prima linea: bevacizumab

Antiangiogenici in combinazione a chemioterapia in prima linea: bevacizumab Micro-ambiente tumorale. Antiangiogenici e immunoterapia: miti e realtà Milano, 11 Ottobre 2016 Antiangiogenici in combinazione a chemioterapia in prima linea: bevacizumab Francesco Grossi U.O.S. Tumori

More information

Dr. Mustafa Jarrar. Chapter 2 Intelligent Agents. Sina Institute, University of Birzeit

Dr. Mustafa Jarrar. Chapter 2 Intelligent Agents. Sina Institute, University of Birzeit Lecture Notes, Advanced Artificial Intelligence (SCOM7341) Sina Institute, University of Birzeit 2 nd Semester, 2012 Advanced Artificial Intelligence (SCOM7341) Chapter 2 Intelligent Agents Dr. Mustafa

More information

MSc Psychological Research Methods/ MPsych Advanced Psychology Module Catalogue / 2018

MSc Psychological Research Methods/ MPsych Advanced Psychology Module Catalogue / 2018 MSc Psychological Research Methods/ MPsych Advanced Psychology Module Catalogue - 2017 / 2018 PSY555: Communication of Research for Psychology (Semester 2) 10 credits core PRM, option MPsych This module

More information

Oncologist. The. Lung Cancer. Bevacizumab Treatment to Progression After Chemotherapy: Outcomes from a U.S. Community Practice Network

Oncologist. The. Lung Cancer. Bevacizumab Treatment to Progression After Chemotherapy: Outcomes from a U.S. Community Practice Network The Oncologist Lung Cancer Bevacizumab Treatment to Progression After Chemotherapy: Outcomes from a U.S. Community Practice Network ERIC NADLER, a ELAINE YU, b ARLIENE RAVELO, b AMY SING, b MICHAEL FORSYTH,

More information

SUPPLEMENTARY MATERIAL

SUPPLEMENTARY MATERIAL SUPPLEMENTARY MATERIAL Supplementary Figure 1. Recursive partitioning using PFS data in patients with advanced NSCLC with non-squamous histology treated in the placebo pemetrexed arm of LUME-Lung 2. (A)

More information

Maintenance Treatment for Advanced NSCLC. Yvonne Summers PhD, FRCP ESMO Preceptorship Programme March 2017

Maintenance Treatment for Advanced NSCLC. Yvonne Summers PhD, FRCP ESMO Preceptorship Programme March 2017 Maintenance Treatment for Advanced NSCLC Yvonne Summers PhD, FRCP ESMO Preceptorship Programme March 2017 Milestones in the Palliative Systemic Treatment of NSCLC 1990 2000 2010 2015 Platinum based Chemotherapy

More information

Intelligent Machines That Act Rationally. Hang Li Toutiao AI Lab

Intelligent Machines That Act Rationally. Hang Li Toutiao AI Lab Intelligent Machines That Act Rationally Hang Li Toutiao AI Lab Four Definitions of Artificial Intelligence Building intelligent machines (i.e., intelligent computers) Thinking humanly Acting humanly Thinking

More information

Getting SMART about Combating Autism with Adaptive Interventions: Novel Treatment and Research Methods

Getting SMART about Combating Autism with Adaptive Interventions: Novel Treatment and Research Methods Getting SMART about Combating Autism with Adaptive Interventions: Novel Treatment and Research Methods 1. Introduction to Sequential Multiple Assignment Randomized Trials and Adaptive Interventions: Two

More information

PROGNOSTIC AND PREDICTIVE BIOMARKERS IN NSCLC. Federico Cappuzzo Istituto Toscano Tumori Ospedale Civile-Livorno Italy

PROGNOSTIC AND PREDICTIVE BIOMARKERS IN NSCLC. Federico Cappuzzo Istituto Toscano Tumori Ospedale Civile-Livorno Italy PROGNOSTIC AND PREDICTIVE BIOMARKERS IN NSCLC Federico Cappuzzo Istituto Toscano Tumori Ospedale Civile-Livorno Italy Prognostic versus predictive Prognostic: In presence of the biomarker patient outcome

More information

Real-time computational attention model for dynamic scenes analysis

Real-time computational attention model for dynamic scenes analysis Computer Science Image and Interaction Laboratory Real-time computational attention model for dynamic scenes analysis Matthieu Perreira Da Silva Vincent Courboulay 19/04/2012 Photonics Europe 2012 Symposium,

More information

A Reinforcement Learning Approach Involving a Shortest Path Finding Algorithm

A Reinforcement Learning Approach Involving a Shortest Path Finding Algorithm Proceedings of the 003 IEEE/RSJ Intl. Conference on Intelligent Robots and Systems Proceedings Las Vegas, of Nevada the 003 October IEEE/RSJ 003 Intl. Conference on Intelligent Robots and Systems Las Vegas,

More information

A HMM-based Pre-training Approach for Sequential Data

A HMM-based Pre-training Approach for Sequential Data A HMM-based Pre-training Approach for Sequential Data Luca Pasa 1, Alberto Testolin 2, Alessandro Sperduti 1 1- Department of Mathematics 2- Department of Developmental Psychology and Socialisation University

More information

Reinforcement Learning and Artificial Intelligence

Reinforcement Learning and Artificial Intelligence Reinforcement Learning and Artificial Intelligence PIs: Rich Sutton Michael Bowling Dale Schuurmans Vadim Bulitko plus: Dale Schuurmans Vadim Bulitko Lori Troop Mark Lee Reinforcement learning is learning

More information