2007
DOI: 10.1016/j.drugalcdep.2007.01.005
|View full text |Cite
|
Sign up to set email alerts
|

Constructing evidence-based treatment strategies using methods from computer science

Abstract: This paper details a new methodology, instance-based reinforcement learning, for constructing adaptive treatment strategies from randomized trials. Adaptive treatment strategies are operationalized clinical guidelines which recommend the next best treatment for an individual based on his/her personal characteristics and response to earlier treatments. The instance-based reinforcement learning methodology comes from the computer science literature, where it was developed to optimize sequences of actions in an e… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
24
0

Year Published

2009
2009
2018
2018

Publication Types

Select...
8
1
1

Relationship

1
9

Authors

Journals

citations
Cited by 41 publications
(24 citation statements)
references
References 29 publications
0
24
0
Order By: Relevance
“…Two common learning approaches are Q -learning (Watkins, 1989; Sutton and Barto, 1998) and A -learning (Murphy, 2003; Blatt et al, 2004), where ‘ Q ’ denotes ‘quality’ and ‘ A ’ denotes ‘advantage’. The Q -learning algorithm, originally proposed in the computer science literature, has become a powerful tool to discover optimal DTRs in the clinical research arena (Murphy et al, 2007; Pineau et al, 2007; Zhao et al, 2009; Nahum-Shani et al, 2012). Q -learning is an approximate dynamic programming procedure that estimates the optimal DTR by first estimating the conditional expectation of the sum of current and future rewards given the current patient history and assuming that optimal decisions are made at all future decision points.…”
Section: Introductionmentioning
confidence: 99%
“…Two common learning approaches are Q -learning (Watkins, 1989; Sutton and Barto, 1998) and A -learning (Murphy, 2003; Blatt et al, 2004), where ‘ Q ’ denotes ‘quality’ and ‘ A ’ denotes ‘advantage’. The Q -learning algorithm, originally proposed in the computer science literature, has become a powerful tool to discover optimal DTRs in the clinical research arena (Murphy et al, 2007; Pineau et al, 2007; Zhao et al, 2009; Nahum-Shani et al, 2012). Q -learning is an approximate dynamic programming procedure that estimates the optimal DTR by first estimating the conditional expectation of the sum of current and future rewards given the current patient history and assuming that optimal decisions are made at all future decision points.…”
Section: Introductionmentioning
confidence: 99%
“…The left panel of Figure 1 shows the raw QIDS score for 50 randomly sampled subjects, we have highlighted two subjects to emphasize heterogeneity in the observed patterns. Based on a survey of clinical papers and previous analyses on STAR*D (see Pineau et al, 2007; Fava et al, 2008; Young et al, 2009; Chakraborty et al, 2013; Schulte et al, 2014; Novick et al, 2015, and references therein) we include as covariates: mean Clinical Global Impression (CGI) score taken over pre-randomization visits, sex, number of comorbidities at baseline, education level (indicator of some college or above), and the log number of major depressive episodes per year between diagnoses and enrollment. To be consistent with our paradigm of maximizing a desirable outcome we use 27 minus the average QIDS score over the course of post-randomization follow-up in the first stage as our outcome.…”
Section: Application To Star*dmentioning
confidence: 99%
“…19,32 In this section, we describe how reinforcement learning can be used to directly optimize stimulation patterns of a closed-loop stimulation device, without necessarily requiring accurate seizure prediction.…”
Section: Reinforcement Learningmentioning
confidence: 99%