SlideShare a Scribd company logo
1 of 25
Download to read offline
Identifying Precision Treatment for Rheumatoid
Arthritis with Reinforcement Learning
Chixiang Chen1
Ashley Gannon2
Duwani Katumullage3
Miaoqi Li4
Mengfei Liu5
Rebecca North6
Jialu Wang5
1Pennsylvania State University, Division of
Biostatistics and Bioinformatics
2Florida State University, Department of
Scientific Computing
3Sam Houston State University, Department of
Statistics
4University of Cincinnati, Department of
Mathematical Sciences
5The George Washington University,
Department of Statistics
6North Carolina State University, Department of
Statistics
July 25, 2018C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 1 / 24
Outline
1 Introduction
Background
The Problem
2 Methodology
3 Results
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 2 / 24
What is RA?
Rheumatoid Arthritis (RA)
A heterogeneous, chronic inflammatory disease that affects the lining
of the joints, causing pain and inflammatory flare ups
Other symptoms:
Issues of the heart, lungs, eyes and skin
60% increase in the risk of heart attack
Twice as likely as the average person to become depressed.
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 3 / 24
Current Treatments
Drug Therapies:
Disease-modifying
antirheumatic drugs
(DMARDs)
Synthetic or biologic
Glucocorticoids
NSAIDs or opioids
Other Therapies:
Physical and occupational
therapy
Surgical procedures
Assistive devices,
supportive products,
adjustments to home
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 4 / 24
The Problem
Motivation: The American College of Rheumatology (ACR) does not
provide an optimal treatment regime to follow after diagnosis.1
Research Goal: Use precision medicine techniques to develop a
dynamic treatment regime for the treatment of RA.
1
C.A. Wijbrandts and P.P. Tak. “Prediction of Response to Targeted Treatment in
Rheumatoid Arthritis”. In: Mayo Clinic Proceedings 92.7 (July 2017), pp. 1129–1143.
ISSN: 00256196. DOI: 10.1016/j.mayocp.2017.05.009. URL:
http://linkinghub.elsevier.com/retrieve/pii/S0025619617303658
(visited on 07/24/2018).
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 5 / 24
The Data
For each of 6,846 RA patients:
At least 6 months of weekly claims data prior to diagnosis
At least 12 months of weekly claims data after diagnosis
Comorbidity indicators prior to/after diagnosis & week of
development with respect to diagnosis
Age, gender, insurance subscriber indicator
For each week:
Counts for hospital (IP, OP, ER) and DR visits, and Rx scripts
Indicators of DMARD, NSAID, opioid, and glucocorticoid use
Treatment group
______ ______ ______ ______ ______1 10 0 0DMARD MTX nTNF TNF Tofa
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 6 / 24
The Data
For our analyses:
Only weeks after treatment initiation were included
We collapsed the data to be on a monthly scale
Summed hospital visit and Rx script counts
Indicator for NSAID or opioid use during the month
Indicator for glucocorticoid use during the month
Treatment group reassigned as maximum of each drug indicator
across the month
Comorbidity baseline adjusted to be at treatment initiation
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 7 / 24
Outline
1 Introduction
Background
The Problem
2 Methodology
3 Results
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 8 / 24
Q-Learning
Let’s narrow our focus to two stage decision-making problem.
Y: second stage outcome.
Ai: Treatment assignment at ith stage, i = 1, 2.
di(Xi): optimal decision rule at ith stage, i = 1, 2.
(d1(X1), d2(X2)) = argmin
a1,a2
E{Y|A1 = a1, X1, A2 = a2, X2}
Key questions:
How to model E{Y|A1 = a1, X1, A2 = a2, X2}?
How to predict optimize decision rule given a new observation?
How to evaluate the effect of optimized decision rule?
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 9 / 24
Q-Learning
Let’s deal with the first question under more rigorous formula:
Key formula
E{Y|A1, A2} = E E E{Y|A1, X1, A2, X2}|X1, A1 |A1, A2
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 10 / 24
Q-Learning
More specific, the pseudo algorithm we propose before is the idea of
Q-Learning2:
step 1: model E{Y|A1, X1, A2, X2} by some specified function
f2(A1, X1, A2, X2) and apply logistic regression to fit
step 2: predict the pseudo-outcome Y based on the fitted model
in step 1 under the optimized decision rules.
step 3 model E Y |X1, A1 by some function f1(A1, X1) based on
the pseudo-outcome Y and apply ordinary regression to fit
2
Rui Song et al. “Penalized Q-learning for dynamic treatment regimens”. In:
Statistica Sinica (2015). ISSN: 10170405. DOI: 10.5705/ss.2012.364. URL:
http:
//www3.stat.sinica.edu.tw/statistica/J25N3/J25N35/J25N35.html
(visited on 07/24/2018).
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 11 / 24
Q-Learning
Next, suppose we have a new patient stop by at the first month, then
optimized decision rule at first month will be given as:
ˆd1(X1) = argmin
a1
ˆf1(A1 = a1, X1)
If later we get the information about that patient at the second month,
then the corresponding optimal decision will be:
ˆd2(X1) = argmin
a2
ˆf2(A1 = ˆd1(X1), X1, A2 = a2, X2)
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 12 / 24
Estimation for expected outcome
Finally, a measure to evaluate how well our optimized decision rule
works:
E{Y|A1 = d1(X1), A2 = d1(X2)} ≤ E{Y|A1 = a1, A2 = a2}
Three Estimators for E{Y|A1, A2}:
naive estimation:
˜Y =
n
i Yi1{A1i}1{A2i}
n
i 1{A1i}1{A2i}
.
It might be biased under observational study
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 13 / 24
Estimation for expected outcome
IPTW estimator:
˜YIPTW =
n
i
Yi 1{A1i }1{A2i }
ˆP(A1i |x1i )ˆP(A2i |x2i )
n
i
1{A1i }1{A2i }
ˆP(A1i |x1i )ˆP(A2i |x2i )
P(A|X) is called propensity score modeled by multinomial logistic
regression.
We assume A1|X1 is independent of A2|X2
˜YIPTW is unbiased estimator only when propensity score P(A|X) is
correctedly specified
3
3
Erica E. M. Moodie, Bibhas Chakraborty, and Michael S. Kramer. “Q-learning for
estimating optimal dynamic treatment rules from observational data”. In: Canadian
Journal of Statistics 40.4 (Dec. 2012), pp. 629–645. ISSN: 03195724. DOI:
10.1002/cjs.11162. URL: http://doi.wiley.com/10.1002/cjs.11162
(visited on 07/24/2018).
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 14 / 24
Estimation for expected outcome
Double robust IPTW estimator:
˜YIPTW-rob =
1
n
n
i
Yi1(A1i)1(A2i)
ˆP(A1i|x1i)ˆP(A2i|x2i)
−
1(A1i)1(A2i) − ˆP(A1i|x1i)ˆP(A2i|x2i)
ˆP(A1i|x1i)ˆP(A2i|x2i)
× ˆf1(ˆd1(x1i), x1i)
same assumption and notation as before
˜YIPTW-rob is unbiased estimator if either one of ˆP(A|x) or
ˆf1(ˆd1(x1), x1), not both, is mis-specified (more recommended).
4
4
Baqun Zhang et al. “A Robust Method for Estimating Optimal Treatment
Regimes”. In: Biometrics 68.4 (Dec. 2012), pp. 1010–1018. ISSN: 0006341X. DOI:
10.1111/j.1541-0420.2012.01763.x. URL:
http://doi.wiley.com/10.1111/j.1541-0420.2012.01763.x (visited onC. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 15 / 24
Outline
1 Introduction
Background
The Problem
2 Methodology
3 Results
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 16 / 24
Results for Two-Stage Analysis
Table: Important Covariates for Two-Stage Analysis
Table: Stage 1
Age at diagnosis
Treatment group (month 1)
COPD
Hypertension
Angina
Skin ulcers
Gender
Table: Stage 2
Treatment group(month 2)
Glucocorticoid or painkiller(month 1)
Age at diagnosis
Prescriptions in month 1
Dr Visits in month 1
OP visits in month 1
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 17 / 24
Results for Two-Stage Analysis
Table: Important Covariates for Two-Stage Analysis
Table: Stage 1
Age at diagnosis
Treatment group (month 1)
COPD
Hypertension
Angina
Skin ulcers
Gender
Table: Stage 2
Treatment group(month 2)
Glucocorticoid or painkiller(month 1)
Age at diagnosis
Prescriptions in month 1
Dr Visits in month 1
OP visits in month 1
Under Optimal
Estimator Treatment Regime Under Observed Data
˜YIPTW 0.040 0.122
˜YIPTW-rob 0.059 0.112
Table: Results under Various Criteria
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 17 / 24
Visualization 1
factor(opt_trt)
1 TNF
1 DMARD
Visualization of the Categorical Response Variable and TreatmentID
When Angina = 1 and COPD = 1
Female
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 18 / 24
Visualization 2
factor(opt_trt)
1 TNF
1 MTX
Visualization of the Categorical Response Variable and TreatmentID
When Angina = 1 and COPD = 1
Male
1 DMARD
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 19 / 24
Visualization 3
factor(opt_trt)
1 TNF
1 MTX
Visualization of the Categorical Response Variable and TreatmentID
When Angina = 0 and COPD = 0
Female
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 20 / 24
Visualization 4
factor(opt_trt)
1 TNF
1 MTX
Visualization of the Categorical Response Variable and TreatmentID
When Angina = 0 and COPD = 0
Male
1 DMARD
1 DMARD &
1 MTX
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 21 / 24
Future Study
1 Making multiple decisions across half year can be extended.
2 Model selection procedure in multiple stages will be conducted.
3 We will obtain the confidence interval for predicted ˜Y.
4 Semi-parametric or non-parametric model will be implemented to
fit the data.
5 Methods dealing with not randomly censored data will be
developed.
6 Model global effect instead of the last stage effect will be
considered.
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 22 / 24
Acknowledgements
Grant Weller, Victoria Mansfield, Yinglong Guo
UnitedHealth Group, Research & Development
Daniel Luckett
University of North Carolina, Department of Biostatistics
SAMSI
Department of Mathematics, NC State University
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 23 / 24
Questions
Thank you for your attention.
Questions?
C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 24 / 24

More Related Content

What's hot

The Rothamsted school meets Lord's paradox
The Rothamsted school meets Lord's paradoxThe Rothamsted school meets Lord's paradox
The Rothamsted school meets Lord's paradoxStephen Senn
 
First in man tokyo
First in man tokyoFirst in man tokyo
First in man tokyoStephen Senn
 
In search of the lost loss function
In search of the lost loss function In search of the lost loss function
In search of the lost loss function Stephen Senn
 
The challenge of small data
The challenge of small dataThe challenge of small data
The challenge of small dataStephen Senn
 
Choosing Regression Models
Choosing Regression ModelsChoosing Regression Models
Choosing Regression ModelsStephen Senn
 
Vaccine trials in the age of COVID-19
Vaccine trials in the age of COVID-19Vaccine trials in the age of COVID-19
Vaccine trials in the age of COVID-19Stephen Senn
 
Clinical trials: quo vadis in the age of covid?
Clinical trials: quo vadis in the age of covid?Clinical trials: quo vadis in the age of covid?
Clinical trials: quo vadis in the age of covid?Stephen Senn
 
To infinity and beyond
To infinity and beyond To infinity and beyond
To infinity and beyond Stephen Senn
 
Does transfer to intensive care units reduce mortality for deteriorating ward...
Does transfer to intensive care units reduce mortality for deteriorating ward...Does transfer to intensive care units reduce mortality for deteriorating ward...
Does transfer to intensive care units reduce mortality for deteriorating ward...cheweb1
 
Sample size calculation - a brief overview
Sample size calculation - a brief overviewSample size calculation - a brief overview
Sample size calculation - a brief overviewAzmi Mohd Tamil
 
Personalised medicine a sceptical view
Personalised medicine a sceptical viewPersonalised medicine a sceptical view
Personalised medicine a sceptical viewStephen Senn
 
The Seven Habits of Highly Effective Statisticians
The Seven Habits of Highly Effective StatisticiansThe Seven Habits of Highly Effective Statisticians
The Seven Habits of Highly Effective StatisticiansStephen Senn
 
Big Data Analytics for Healthcare
Big Data Analytics for HealthcareBig Data Analytics for Healthcare
Big Data Analytics for HealthcareChandan Reddy
 
To infinity and beyond v2
To infinity and beyond v2To infinity and beyond v2
To infinity and beyond v2Stephen Senn
 
Clinical_Decision_Support_For_Heart_Disease
Clinical_Decision_Support_For_Heart_DiseaseClinical_Decision_Support_For_Heart_Disease
Clinical_Decision_Support_For_Heart_DiseaseSunil Kakade
 
Ct lecture 8. comparing two groups categorical data
Ct lecture 8. comparing two groups   categorical dataCt lecture 8. comparing two groups   categorical data
Ct lecture 8. comparing two groups categorical dataHau Pham
 
Network meta-analysis with integrated nested Laplace approximations
Network meta-analysis with integrated nested Laplace approximationsNetwork meta-analysis with integrated nested Laplace approximations
Network meta-analysis with integrated nested Laplace approximationsBurak Kürsad Günhan
 

What's hot (20)

PMED Transition Workshop - Dynamic Treatment Regimes via Reward Ignorant Mode...
PMED Transition Workshop - Dynamic Treatment Regimes via Reward Ignorant Mode...PMED Transition Workshop - Dynamic Treatment Regimes via Reward Ignorant Mode...
PMED Transition Workshop - Dynamic Treatment Regimes via Reward Ignorant Mode...
 
The Rothamsted school meets Lord's paradox
The Rothamsted school meets Lord's paradoxThe Rothamsted school meets Lord's paradox
The Rothamsted school meets Lord's paradox
 
First in man tokyo
First in man tokyoFirst in man tokyo
First in man tokyo
 
Yates and cochran
Yates and cochranYates and cochran
Yates and cochran
 
In search of the lost loss function
In search of the lost loss function In search of the lost loss function
In search of the lost loss function
 
The challenge of small data
The challenge of small dataThe challenge of small data
The challenge of small data
 
Choosing Regression Models
Choosing Regression ModelsChoosing Regression Models
Choosing Regression Models
 
Vaccine trials in the age of COVID-19
Vaccine trials in the age of COVID-19Vaccine trials in the age of COVID-19
Vaccine trials in the age of COVID-19
 
Clinical trials: quo vadis in the age of covid?
Clinical trials: quo vadis in the age of covid?Clinical trials: quo vadis in the age of covid?
Clinical trials: quo vadis in the age of covid?
 
Clinical data analytics
Clinical data analyticsClinical data analytics
Clinical data analytics
 
To infinity and beyond
To infinity and beyond To infinity and beyond
To infinity and beyond
 
Does transfer to intensive care units reduce mortality for deteriorating ward...
Does transfer to intensive care units reduce mortality for deteriorating ward...Does transfer to intensive care units reduce mortality for deteriorating ward...
Does transfer to intensive care units reduce mortality for deteriorating ward...
 
Sample size calculation - a brief overview
Sample size calculation - a brief overviewSample size calculation - a brief overview
Sample size calculation - a brief overview
 
Personalised medicine a sceptical view
Personalised medicine a sceptical viewPersonalised medicine a sceptical view
Personalised medicine a sceptical view
 
The Seven Habits of Highly Effective Statisticians
The Seven Habits of Highly Effective StatisticiansThe Seven Habits of Highly Effective Statisticians
The Seven Habits of Highly Effective Statisticians
 
Big Data Analytics for Healthcare
Big Data Analytics for HealthcareBig Data Analytics for Healthcare
Big Data Analytics for Healthcare
 
To infinity and beyond v2
To infinity and beyond v2To infinity and beyond v2
To infinity and beyond v2
 
Clinical_Decision_Support_For_Heart_Disease
Clinical_Decision_Support_For_Heart_DiseaseClinical_Decision_Support_For_Heart_Disease
Clinical_Decision_Support_For_Heart_Disease
 
Ct lecture 8. comparing two groups categorical data
Ct lecture 8. comparing two groups   categorical dataCt lecture 8. comparing two groups   categorical data
Ct lecture 8. comparing two groups categorical data
 
Network meta-analysis with integrated nested Laplace approximations
Network meta-analysis with integrated nested Laplace approximationsNetwork meta-analysis with integrated nested Laplace approximations
Network meta-analysis with integrated nested Laplace approximations
 

Similar to 2018 IMSM: Identifying Precision Treatment for Rheumatoid Arthritis with Reinforcement Learning - Savvysherpa Working Group, July 25, 2018

25_Anderson_Biostatistics_and_Epidemiology.ppt
25_Anderson_Biostatistics_and_Epidemiology.ppt25_Anderson_Biostatistics_and_Epidemiology.ppt
25_Anderson_Biostatistics_and_Epidemiology.pptPriyankaSharma89719
 
Subgroup identification for precision medicine. a comparative review of 13 me...
Subgroup identification for precision medicine. a comparative review of 13 me...Subgroup identification for precision medicine. a comparative review of 13 me...
Subgroup identification for precision medicine. a comparative review of 13 me...SuciAidaDahhar
 
Ct lecture 20. survival analysis (part 2)
Ct lecture 20. survival analysis (part 2)Ct lecture 20. survival analysis (part 2)
Ct lecture 20. survival analysis (part 2)Hau Pham
 
臨床疫学研究における傾向スコア分析の使い⽅ 〜観察研究における治療効果研究〜
臨床疫学研究における傾向スコア分析の使い⽅ 〜観察研究における治療効果研究〜臨床疫学研究における傾向スコア分析の使い⽅ 〜観察研究における治療効果研究〜
臨床疫学研究における傾向スコア分析の使い⽅ 〜観察研究における治療効果研究〜Yasuyuki Okumura
 
Global Journal of Perioperative Medicine
Global Journal of Perioperative MedicineGlobal Journal of Perioperative Medicine
Global Journal of Perioperative Medicinepeertechzpublication
 
Apprasial of a thearpy paper sep.2018
Apprasial of a thearpy paper sep.2018Apprasial of a thearpy paper sep.2018
Apprasial of a thearpy paper sep.2018Amna Khairy
 
Effectiveness of Aerobic Exercise on Ambulatory Blood P.docx
     Effectiveness of Aerobic Exercise on Ambulatory Blood P.docx     Effectiveness of Aerobic Exercise on Ambulatory Blood P.docx
Effectiveness of Aerobic Exercise on Ambulatory Blood P.docxrobert345678
 
Clinical data based optimal STI strategies for HIV: a reinforcement learning ...
Clinical data based optimal STI strategies for HIV: a reinforcement learning ...Clinical data based optimal STI strategies for HIV: a reinforcement learning ...
Clinical data based optimal STI strategies for HIV: a reinforcement learning ...Université de Liège (ULg)
 
Quantitative Statistical Analysis Work Sample From Statswork
Quantitative Statistical Analysis Work Sample From StatsworkQuantitative Statistical Analysis Work Sample From Statswork
Quantitative Statistical Analysis Work Sample From StatsworkStats Statswork
 
Heart Disease Prediction Using Data Mining
Heart Disease Prediction Using Data MiningHeart Disease Prediction Using Data Mining
Heart Disease Prediction Using Data MiningIRJET Journal
 
Analytic Methods and Issues in CER from Observational Data
Analytic Methods and Issues in CER from Observational DataAnalytic Methods and Issues in CER from Observational Data
Analytic Methods and Issues in CER from Observational DataCTSI at UCSF
 
Machine Learning for Survival Analysis
Machine Learning for Survival AnalysisMachine Learning for Survival Analysis
Machine Learning for Survival AnalysisChandan Reddy
 
IRJET -Improving the Accuracy of the Heart Disease Prediction using Hybrid Ma...
IRJET -Improving the Accuracy of the Heart Disease Prediction using Hybrid Ma...IRJET -Improving the Accuracy of the Heart Disease Prediction using Hybrid Ma...
IRJET -Improving the Accuracy of the Heart Disease Prediction using Hybrid Ma...IRJET Journal
 
Heart disease prediction by using novel optimization algorithm_ A supervised ...
Heart disease prediction by using novel optimization algorithm_ A supervised ...Heart disease prediction by using novel optimization algorithm_ A supervised ...
Heart disease prediction by using novel optimization algorithm_ A supervised ...BASMAJUMAASALEHALMOH
 
Walmart HR head tells how to save money
Walmart HR head tells how to save moneyWalmart HR head tells how to save money
Walmart HR head tells how to save moneyNelson Hendler
 

Similar to 2018 IMSM: Identifying Precision Treatment for Rheumatoid Arthritis with Reinforcement Learning - Savvysherpa Working Group, July 25, 2018 (20)

25_Anderson_Biostatistics_and_Epidemiology.ppt
25_Anderson_Biostatistics_and_Epidemiology.ppt25_Anderson_Biostatistics_and_Epidemiology.ppt
25_Anderson_Biostatistics_and_Epidemiology.ppt
 
Subgroup identification for precision medicine. a comparative review of 13 me...
Subgroup identification for precision medicine. a comparative review of 13 me...Subgroup identification for precision medicine. a comparative review of 13 me...
Subgroup identification for precision medicine. a comparative review of 13 me...
 
Lg ph d_slides_vfinal
Lg ph d_slides_vfinalLg ph d_slides_vfinal
Lg ph d_slides_vfinal
 
Ct lecture 20. survival analysis (part 2)
Ct lecture 20. survival analysis (part 2)Ct lecture 20. survival analysis (part 2)
Ct lecture 20. survival analysis (part 2)
 
Quantitative Synthesis I
Quantitative Synthesis IQuantitative Synthesis I
Quantitative Synthesis I
 
臨床疫学研究における傾向スコア分析の使い⽅ 〜観察研究における治療効果研究〜
臨床疫学研究における傾向スコア分析の使い⽅ 〜観察研究における治療効果研究〜臨床疫学研究における傾向スコア分析の使い⽅ 〜観察研究における治療効果研究〜
臨床疫学研究における傾向スコア分析の使い⽅ 〜観察研究における治療効果研究〜
 
Global Journal of Perioperative Medicine
Global Journal of Perioperative MedicineGlobal Journal of Perioperative Medicine
Global Journal of Perioperative Medicine
 
Apprasial of a thearpy paper sep.2018
Apprasial of a thearpy paper sep.2018Apprasial of a thearpy paper sep.2018
Apprasial of a thearpy paper sep.2018
 
Effectiveness of Aerobic Exercise on Ambulatory Blood P.docx
     Effectiveness of Aerobic Exercise on Ambulatory Blood P.docx     Effectiveness of Aerobic Exercise on Ambulatory Blood P.docx
Effectiveness of Aerobic Exercise on Ambulatory Blood P.docx
 
Clinical data based optimal STI strategies for HIV: a reinforcement learning ...
Clinical data based optimal STI strategies for HIV: a reinforcement learning ...Clinical data based optimal STI strategies for HIV: a reinforcement learning ...
Clinical data based optimal STI strategies for HIV: a reinforcement learning ...
 
Quantitative Statistical Analysis Work Sample From Statswork
Quantitative Statistical Analysis Work Sample From StatsworkQuantitative Statistical Analysis Work Sample From Statswork
Quantitative Statistical Analysis Work Sample From Statswork
 
Heart Disease Prediction Using Data Mining
Heart Disease Prediction Using Data MiningHeart Disease Prediction Using Data Mining
Heart Disease Prediction Using Data Mining
 
Analytic Methods and Issues in CER from Observational Data
Analytic Methods and Issues in CER from Observational DataAnalytic Methods and Issues in CER from Observational Data
Analytic Methods and Issues in CER from Observational Data
 
Machine Learning for Survival Analysis
Machine Learning for Survival AnalysisMachine Learning for Survival Analysis
Machine Learning for Survival Analysis
 
Causal Inference Opening Workshop - Some Applications of Reinforcement Learni...
Causal Inference Opening Workshop - Some Applications of Reinforcement Learni...Causal Inference Opening Workshop - Some Applications of Reinforcement Learni...
Causal Inference Opening Workshop - Some Applications of Reinforcement Learni...
 
IRJET -Improving the Accuracy of the Heart Disease Prediction using Hybrid Ma...
IRJET -Improving the Accuracy of the Heart Disease Prediction using Hybrid Ma...IRJET -Improving the Accuracy of the Heart Disease Prediction using Hybrid Ma...
IRJET -Improving the Accuracy of the Heart Disease Prediction using Hybrid Ma...
 
JSM2012-HybridPMS-Liao-v3
JSM2012-HybridPMS-Liao-v3JSM2012-HybridPMS-Liao-v3
JSM2012-HybridPMS-Liao-v3
 
Heart disease prediction by using novel optimization algorithm_ A supervised ...
Heart disease prediction by using novel optimization algorithm_ A supervised ...Heart disease prediction by using novel optimization algorithm_ A supervised ...
Heart disease prediction by using novel optimization algorithm_ A supervised ...
 
Nursing research
Nursing researchNursing research
Nursing research
 
Walmart HR head tells how to save money
Walmart HR head tells how to save moneyWalmart HR head tells how to save money
Walmart HR head tells how to save money
 

More from The Statistical and Applied Mathematical Sciences Institute

More from The Statistical and Applied Mathematical Sciences Institute (20)

Causal Inference Opening Workshop - Latent Variable Models, Causal Inference,...
Causal Inference Opening Workshop - Latent Variable Models, Causal Inference,...Causal Inference Opening Workshop - Latent Variable Models, Causal Inference,...
Causal Inference Opening Workshop - Latent Variable Models, Causal Inference,...
 
2019 Fall Series: Special Guest Lecture - 0-1 Phase Transitions in High Dimen...
2019 Fall Series: Special Guest Lecture - 0-1 Phase Transitions in High Dimen...2019 Fall Series: Special Guest Lecture - 0-1 Phase Transitions in High Dimen...
2019 Fall Series: Special Guest Lecture - 0-1 Phase Transitions in High Dimen...
 
Causal Inference Opening Workshop - Causal Discovery in Neuroimaging Data - F...
Causal Inference Opening Workshop - Causal Discovery in Neuroimaging Data - F...Causal Inference Opening Workshop - Causal Discovery in Neuroimaging Data - F...
Causal Inference Opening Workshop - Causal Discovery in Neuroimaging Data - F...
 
Causal Inference Opening Workshop - Smooth Extensions to BART for Heterogeneo...
Causal Inference Opening Workshop - Smooth Extensions to BART for Heterogeneo...Causal Inference Opening Workshop - Smooth Extensions to BART for Heterogeneo...
Causal Inference Opening Workshop - Smooth Extensions to BART for Heterogeneo...
 
Causal Inference Opening Workshop - A Bracketing Relationship between Differe...
Causal Inference Opening Workshop - A Bracketing Relationship between Differe...Causal Inference Opening Workshop - A Bracketing Relationship between Differe...
Causal Inference Opening Workshop - A Bracketing Relationship between Differe...
 
Causal Inference Opening Workshop - Testing Weak Nulls in Matched Observation...
Causal Inference Opening Workshop - Testing Weak Nulls in Matched Observation...Causal Inference Opening Workshop - Testing Weak Nulls in Matched Observation...
Causal Inference Opening Workshop - Testing Weak Nulls in Matched Observation...
 
Causal Inference Opening Workshop - Difference-in-differences: more than meet...
Causal Inference Opening Workshop - Difference-in-differences: more than meet...Causal Inference Opening Workshop - Difference-in-differences: more than meet...
Causal Inference Opening Workshop - Difference-in-differences: more than meet...
 
Causal Inference Opening Workshop - New Statistical Learning Methods for Esti...
Causal Inference Opening Workshop - New Statistical Learning Methods for Esti...Causal Inference Opening Workshop - New Statistical Learning Methods for Esti...
Causal Inference Opening Workshop - New Statistical Learning Methods for Esti...
 
Causal Inference Opening Workshop - Bipartite Causal Inference with Interfere...
Causal Inference Opening Workshop - Bipartite Causal Inference with Interfere...Causal Inference Opening Workshop - Bipartite Causal Inference with Interfere...
Causal Inference Opening Workshop - Bipartite Causal Inference with Interfere...
 
Causal Inference Opening Workshop - Bridging the Gap Between Causal Literatur...
Causal Inference Opening Workshop - Bridging the Gap Between Causal Literatur...Causal Inference Opening Workshop - Bridging the Gap Between Causal Literatur...
Causal Inference Opening Workshop - Bridging the Gap Between Causal Literatur...
 
Causal Inference Opening Workshop - Bracketing Bounds for Differences-in-Diff...
Causal Inference Opening Workshop - Bracketing Bounds for Differences-in-Diff...Causal Inference Opening Workshop - Bracketing Bounds for Differences-in-Diff...
Causal Inference Opening Workshop - Bracketing Bounds for Differences-in-Diff...
 
Causal Inference Opening Workshop - Assisting the Impact of State Polcies: Br...
Causal Inference Opening Workshop - Assisting the Impact of State Polcies: Br...Causal Inference Opening Workshop - Assisting the Impact of State Polcies: Br...
Causal Inference Opening Workshop - Assisting the Impact of State Polcies: Br...
 
Causal Inference Opening Workshop - Experimenting in Equilibrium - Stefan Wag...
Causal Inference Opening Workshop - Experimenting in Equilibrium - Stefan Wag...Causal Inference Opening Workshop - Experimenting in Equilibrium - Stefan Wag...
Causal Inference Opening Workshop - Experimenting in Equilibrium - Stefan Wag...
 
Causal Inference Opening Workshop - Targeted Learning for Causal Inference Ba...
Causal Inference Opening Workshop - Targeted Learning for Causal Inference Ba...Causal Inference Opening Workshop - Targeted Learning for Causal Inference Ba...
Causal Inference Opening Workshop - Targeted Learning for Causal Inference Ba...
 
Causal Inference Opening Workshop - Bayesian Nonparametric Models for Treatme...
Causal Inference Opening Workshop - Bayesian Nonparametric Models for Treatme...Causal Inference Opening Workshop - Bayesian Nonparametric Models for Treatme...
Causal Inference Opening Workshop - Bayesian Nonparametric Models for Treatme...
 
2019 Fall Series: Special Guest Lecture - Adversarial Risk Analysis of the Ge...
2019 Fall Series: Special Guest Lecture - Adversarial Risk Analysis of the Ge...2019 Fall Series: Special Guest Lecture - Adversarial Risk Analysis of the Ge...
2019 Fall Series: Special Guest Lecture - Adversarial Risk Analysis of the Ge...
 
2019 Fall Series: Professional Development, Writing Academic Papers…What Work...
2019 Fall Series: Professional Development, Writing Academic Papers…What Work...2019 Fall Series: Professional Development, Writing Academic Papers…What Work...
2019 Fall Series: Professional Development, Writing Academic Papers…What Work...
 
2019 GDRR: Blockchain Data Analytics - Machine Learning in/for Blockchain: Fu...
2019 GDRR: Blockchain Data Analytics - Machine Learning in/for Blockchain: Fu...2019 GDRR: Blockchain Data Analytics - Machine Learning in/for Blockchain: Fu...
2019 GDRR: Blockchain Data Analytics - Machine Learning in/for Blockchain: Fu...
 
2019 GDRR: Blockchain Data Analytics - QuTrack: Model Life Cycle Management f...
2019 GDRR: Blockchain Data Analytics - QuTrack: Model Life Cycle Management f...2019 GDRR: Blockchain Data Analytics - QuTrack: Model Life Cycle Management f...
2019 GDRR: Blockchain Data Analytics - QuTrack: Model Life Cycle Management f...
 
2019 GDRR: Blockchain Data Analytics - Modeling Cryptocurrency Markets with T...
2019 GDRR: Blockchain Data Analytics - Modeling Cryptocurrency Markets with T...2019 GDRR: Blockchain Data Analytics - Modeling Cryptocurrency Markets with T...
2019 GDRR: Blockchain Data Analytics - Modeling Cryptocurrency Markets with T...
 

Recently uploaded

Basic Civil Engineering first year Notes- Chapter 4 Building.pptx
Basic Civil Engineering first year Notes- Chapter 4 Building.pptxBasic Civil Engineering first year Notes- Chapter 4 Building.pptx
Basic Civil Engineering first year Notes- Chapter 4 Building.pptxDenish Jangid
 
1029-Danh muc Sach Giao Khoa khoi 6.pdf
1029-Danh muc Sach Giao Khoa khoi  6.pdf1029-Danh muc Sach Giao Khoa khoi  6.pdf
1029-Danh muc Sach Giao Khoa khoi 6.pdfQucHHunhnh
 
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptxMaritesTamaniVerdade
 
Mixin Classes in Odoo 17 How to Extend Models Using Mixin Classes
Mixin Classes in Odoo 17  How to Extend Models Using Mixin ClassesMixin Classes in Odoo 17  How to Extend Models Using Mixin Classes
Mixin Classes in Odoo 17 How to Extend Models Using Mixin ClassesCeline George
 
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...christianmathematics
 
Introduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The BasicsIntroduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The BasicsTechSoup
 
General Principles of Intellectual Property: Concepts of Intellectual Proper...
General Principles of Intellectual Property: Concepts of Intellectual  Proper...General Principles of Intellectual Property: Concepts of Intellectual  Proper...
General Principles of Intellectual Property: Concepts of Intellectual Proper...Poonam Aher Patil
 
Making communications land - Are they received and understood as intended? we...
Making communications land - Are they received and understood as intended? we...Making communications land - Are they received and understood as intended? we...
Making communications land - Are they received and understood as intended? we...Association for Project Management
 
HMCS Max Bernays Pre-Deployment Brief (May 2024).pptx
HMCS Max Bernays Pre-Deployment Brief (May 2024).pptxHMCS Max Bernays Pre-Deployment Brief (May 2024).pptx
HMCS Max Bernays Pre-Deployment Brief (May 2024).pptxEsquimalt MFRC
 
This PowerPoint helps students to consider the concept of infinity.
This PowerPoint helps students to consider the concept of infinity.This PowerPoint helps students to consider the concept of infinity.
This PowerPoint helps students to consider the concept of infinity.christianmathematics
 
ComPTIA Overview | Comptia Security+ Book SY0-701
ComPTIA Overview | Comptia Security+ Book SY0-701ComPTIA Overview | Comptia Security+ Book SY0-701
ComPTIA Overview | Comptia Security+ Book SY0-701bronxfugly43
 
Kodo Millet PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
Kodo Millet  PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...Kodo Millet  PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
Kodo Millet PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...pradhanghanshyam7136
 
Activity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdfActivity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdfciinovamais
 
Sociology 101 Demonstration of Learning Exhibit
Sociology 101 Demonstration of Learning ExhibitSociology 101 Demonstration of Learning Exhibit
Sociology 101 Demonstration of Learning Exhibitjbellavia9
 
Google Gemini An AI Revolution in Education.pptx
Google Gemini An AI Revolution in Education.pptxGoogle Gemini An AI Revolution in Education.pptx
Google Gemini An AI Revolution in Education.pptxDr. Sarita Anand
 
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptxSKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptxAmanpreet Kaur
 
Unit-IV- Pharma. Marketing Channels.pptx
Unit-IV- Pharma. Marketing Channels.pptxUnit-IV- Pharma. Marketing Channels.pptx
Unit-IV- Pharma. Marketing Channels.pptxVishalSingh1417
 
ICT role in 21st century education and it's challenges.
ICT role in 21st century education and it's challenges.ICT role in 21st century education and it's challenges.
ICT role in 21st century education and it's challenges.MaryamAhmad92
 
Dyslexia AI Workshop for Slideshare.pptx
Dyslexia AI Workshop for Slideshare.pptxDyslexia AI Workshop for Slideshare.pptx
Dyslexia AI Workshop for Slideshare.pptxcallscotland1987
 
On National Teacher Day, meet the 2024-25 Kenan Fellows
On National Teacher Day, meet the 2024-25 Kenan FellowsOn National Teacher Day, meet the 2024-25 Kenan Fellows
On National Teacher Day, meet the 2024-25 Kenan FellowsMebane Rash
 

Recently uploaded (20)

Basic Civil Engineering first year Notes- Chapter 4 Building.pptx
Basic Civil Engineering first year Notes- Chapter 4 Building.pptxBasic Civil Engineering first year Notes- Chapter 4 Building.pptx
Basic Civil Engineering first year Notes- Chapter 4 Building.pptx
 
1029-Danh muc Sach Giao Khoa khoi 6.pdf
1029-Danh muc Sach Giao Khoa khoi  6.pdf1029-Danh muc Sach Giao Khoa khoi  6.pdf
1029-Danh muc Sach Giao Khoa khoi 6.pdf
 
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
2024-NATIONAL-LEARNING-CAMP-AND-OTHER.pptx
 
Mixin Classes in Odoo 17 How to Extend Models Using Mixin Classes
Mixin Classes in Odoo 17  How to Extend Models Using Mixin ClassesMixin Classes in Odoo 17  How to Extend Models Using Mixin Classes
Mixin Classes in Odoo 17 How to Extend Models Using Mixin Classes
 
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
 
Introduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The BasicsIntroduction to Nonprofit Accounting: The Basics
Introduction to Nonprofit Accounting: The Basics
 
General Principles of Intellectual Property: Concepts of Intellectual Proper...
General Principles of Intellectual Property: Concepts of Intellectual  Proper...General Principles of Intellectual Property: Concepts of Intellectual  Proper...
General Principles of Intellectual Property: Concepts of Intellectual Proper...
 
Making communications land - Are they received and understood as intended? we...
Making communications land - Are they received and understood as intended? we...Making communications land - Are they received and understood as intended? we...
Making communications land - Are they received and understood as intended? we...
 
HMCS Max Bernays Pre-Deployment Brief (May 2024).pptx
HMCS Max Bernays Pre-Deployment Brief (May 2024).pptxHMCS Max Bernays Pre-Deployment Brief (May 2024).pptx
HMCS Max Bernays Pre-Deployment Brief (May 2024).pptx
 
This PowerPoint helps students to consider the concept of infinity.
This PowerPoint helps students to consider the concept of infinity.This PowerPoint helps students to consider the concept of infinity.
This PowerPoint helps students to consider the concept of infinity.
 
ComPTIA Overview | Comptia Security+ Book SY0-701
ComPTIA Overview | Comptia Security+ Book SY0-701ComPTIA Overview | Comptia Security+ Book SY0-701
ComPTIA Overview | Comptia Security+ Book SY0-701
 
Kodo Millet PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
Kodo Millet  PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...Kodo Millet  PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
Kodo Millet PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
 
Activity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdfActivity 01 - Artificial Culture (1).pdf
Activity 01 - Artificial Culture (1).pdf
 
Sociology 101 Demonstration of Learning Exhibit
Sociology 101 Demonstration of Learning ExhibitSociology 101 Demonstration of Learning Exhibit
Sociology 101 Demonstration of Learning Exhibit
 
Google Gemini An AI Revolution in Education.pptx
Google Gemini An AI Revolution in Education.pptxGoogle Gemini An AI Revolution in Education.pptx
Google Gemini An AI Revolution in Education.pptx
 
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptxSKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
 
Unit-IV- Pharma. Marketing Channels.pptx
Unit-IV- Pharma. Marketing Channels.pptxUnit-IV- Pharma. Marketing Channels.pptx
Unit-IV- Pharma. Marketing Channels.pptx
 
ICT role in 21st century education and it's challenges.
ICT role in 21st century education and it's challenges.ICT role in 21st century education and it's challenges.
ICT role in 21st century education and it's challenges.
 
Dyslexia AI Workshop for Slideshare.pptx
Dyslexia AI Workshop for Slideshare.pptxDyslexia AI Workshop for Slideshare.pptx
Dyslexia AI Workshop for Slideshare.pptx
 
On National Teacher Day, meet the 2024-25 Kenan Fellows
On National Teacher Day, meet the 2024-25 Kenan FellowsOn National Teacher Day, meet the 2024-25 Kenan Fellows
On National Teacher Day, meet the 2024-25 Kenan Fellows
 

2018 IMSM: Identifying Precision Treatment for Rheumatoid Arthritis with Reinforcement Learning - Savvysherpa Working Group, July 25, 2018

  • 1. Identifying Precision Treatment for Rheumatoid Arthritis with Reinforcement Learning Chixiang Chen1 Ashley Gannon2 Duwani Katumullage3 Miaoqi Li4 Mengfei Liu5 Rebecca North6 Jialu Wang5 1Pennsylvania State University, Division of Biostatistics and Bioinformatics 2Florida State University, Department of Scientific Computing 3Sam Houston State University, Department of Statistics 4University of Cincinnati, Department of Mathematical Sciences 5The George Washington University, Department of Statistics 6North Carolina State University, Department of Statistics July 25, 2018C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 1 / 24
  • 2. Outline 1 Introduction Background The Problem 2 Methodology 3 Results C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 2 / 24
  • 3. What is RA? Rheumatoid Arthritis (RA) A heterogeneous, chronic inflammatory disease that affects the lining of the joints, causing pain and inflammatory flare ups Other symptoms: Issues of the heart, lungs, eyes and skin 60% increase in the risk of heart attack Twice as likely as the average person to become depressed. C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 3 / 24
  • 4. Current Treatments Drug Therapies: Disease-modifying antirheumatic drugs (DMARDs) Synthetic or biologic Glucocorticoids NSAIDs or opioids Other Therapies: Physical and occupational therapy Surgical procedures Assistive devices, supportive products, adjustments to home C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 4 / 24
  • 5. The Problem Motivation: The American College of Rheumatology (ACR) does not provide an optimal treatment regime to follow after diagnosis.1 Research Goal: Use precision medicine techniques to develop a dynamic treatment regime for the treatment of RA. 1 C.A. Wijbrandts and P.P. Tak. “Prediction of Response to Targeted Treatment in Rheumatoid Arthritis”. In: Mayo Clinic Proceedings 92.7 (July 2017), pp. 1129–1143. ISSN: 00256196. DOI: 10.1016/j.mayocp.2017.05.009. URL: http://linkinghub.elsevier.com/retrieve/pii/S0025619617303658 (visited on 07/24/2018). C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 5 / 24
  • 6. The Data For each of 6,846 RA patients: At least 6 months of weekly claims data prior to diagnosis At least 12 months of weekly claims data after diagnosis Comorbidity indicators prior to/after diagnosis & week of development with respect to diagnosis Age, gender, insurance subscriber indicator For each week: Counts for hospital (IP, OP, ER) and DR visits, and Rx scripts Indicators of DMARD, NSAID, opioid, and glucocorticoid use Treatment group ______ ______ ______ ______ ______1 10 0 0DMARD MTX nTNF TNF Tofa C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 6 / 24
  • 7. The Data For our analyses: Only weeks after treatment initiation were included We collapsed the data to be on a monthly scale Summed hospital visit and Rx script counts Indicator for NSAID or opioid use during the month Indicator for glucocorticoid use during the month Treatment group reassigned as maximum of each drug indicator across the month Comorbidity baseline adjusted to be at treatment initiation C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 7 / 24
  • 8. Outline 1 Introduction Background The Problem 2 Methodology 3 Results C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 8 / 24
  • 9. Q-Learning Let’s narrow our focus to two stage decision-making problem. Y: second stage outcome. Ai: Treatment assignment at ith stage, i = 1, 2. di(Xi): optimal decision rule at ith stage, i = 1, 2. (d1(X1), d2(X2)) = argmin a1,a2 E{Y|A1 = a1, X1, A2 = a2, X2} Key questions: How to model E{Y|A1 = a1, X1, A2 = a2, X2}? How to predict optimize decision rule given a new observation? How to evaluate the effect of optimized decision rule? C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 9 / 24
  • 10. Q-Learning Let’s deal with the first question under more rigorous formula: Key formula E{Y|A1, A2} = E E E{Y|A1, X1, A2, X2}|X1, A1 |A1, A2 C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 10 / 24
  • 11. Q-Learning More specific, the pseudo algorithm we propose before is the idea of Q-Learning2: step 1: model E{Y|A1, X1, A2, X2} by some specified function f2(A1, X1, A2, X2) and apply logistic regression to fit step 2: predict the pseudo-outcome Y based on the fitted model in step 1 under the optimized decision rules. step 3 model E Y |X1, A1 by some function f1(A1, X1) based on the pseudo-outcome Y and apply ordinary regression to fit 2 Rui Song et al. “Penalized Q-learning for dynamic treatment regimens”. In: Statistica Sinica (2015). ISSN: 10170405. DOI: 10.5705/ss.2012.364. URL: http: //www3.stat.sinica.edu.tw/statistica/J25N3/J25N35/J25N35.html (visited on 07/24/2018). C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 11 / 24
  • 12. Q-Learning Next, suppose we have a new patient stop by at the first month, then optimized decision rule at first month will be given as: ˆd1(X1) = argmin a1 ˆf1(A1 = a1, X1) If later we get the information about that patient at the second month, then the corresponding optimal decision will be: ˆd2(X1) = argmin a2 ˆf2(A1 = ˆd1(X1), X1, A2 = a2, X2) C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 12 / 24
  • 13. Estimation for expected outcome Finally, a measure to evaluate how well our optimized decision rule works: E{Y|A1 = d1(X1), A2 = d1(X2)} ≤ E{Y|A1 = a1, A2 = a2} Three Estimators for E{Y|A1, A2}: naive estimation: ˜Y = n i Yi1{A1i}1{A2i} n i 1{A1i}1{A2i} . It might be biased under observational study C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 13 / 24
  • 14. Estimation for expected outcome IPTW estimator: ˜YIPTW = n i Yi 1{A1i }1{A2i } ˆP(A1i |x1i )ˆP(A2i |x2i ) n i 1{A1i }1{A2i } ˆP(A1i |x1i )ˆP(A2i |x2i ) P(A|X) is called propensity score modeled by multinomial logistic regression. We assume A1|X1 is independent of A2|X2 ˜YIPTW is unbiased estimator only when propensity score P(A|X) is correctedly specified 3 3 Erica E. M. Moodie, Bibhas Chakraborty, and Michael S. Kramer. “Q-learning for estimating optimal dynamic treatment rules from observational data”. In: Canadian Journal of Statistics 40.4 (Dec. 2012), pp. 629–645. ISSN: 03195724. DOI: 10.1002/cjs.11162. URL: http://doi.wiley.com/10.1002/cjs.11162 (visited on 07/24/2018). C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 14 / 24
  • 15. Estimation for expected outcome Double robust IPTW estimator: ˜YIPTW-rob = 1 n n i Yi1(A1i)1(A2i) ˆP(A1i|x1i)ˆP(A2i|x2i) − 1(A1i)1(A2i) − ˆP(A1i|x1i)ˆP(A2i|x2i) ˆP(A1i|x1i)ˆP(A2i|x2i) × ˆf1(ˆd1(x1i), x1i) same assumption and notation as before ˜YIPTW-rob is unbiased estimator if either one of ˆP(A|x) or ˆf1(ˆd1(x1), x1), not both, is mis-specified (more recommended). 4 4 Baqun Zhang et al. “A Robust Method for Estimating Optimal Treatment Regimes”. In: Biometrics 68.4 (Dec. 2012), pp. 1010–1018. ISSN: 0006341X. DOI: 10.1111/j.1541-0420.2012.01763.x. URL: http://doi.wiley.com/10.1111/j.1541-0420.2012.01763.x (visited onC. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 15 / 24
  • 16. Outline 1 Introduction Background The Problem 2 Methodology 3 Results C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 16 / 24
  • 17. Results for Two-Stage Analysis Table: Important Covariates for Two-Stage Analysis Table: Stage 1 Age at diagnosis Treatment group (month 1) COPD Hypertension Angina Skin ulcers Gender Table: Stage 2 Treatment group(month 2) Glucocorticoid or painkiller(month 1) Age at diagnosis Prescriptions in month 1 Dr Visits in month 1 OP visits in month 1 C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 17 / 24
  • 18. Results for Two-Stage Analysis Table: Important Covariates for Two-Stage Analysis Table: Stage 1 Age at diagnosis Treatment group (month 1) COPD Hypertension Angina Skin ulcers Gender Table: Stage 2 Treatment group(month 2) Glucocorticoid or painkiller(month 1) Age at diagnosis Prescriptions in month 1 Dr Visits in month 1 OP visits in month 1 Under Optimal Estimator Treatment Regime Under Observed Data ˜YIPTW 0.040 0.122 ˜YIPTW-rob 0.059 0.112 Table: Results under Various Criteria C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 17 / 24
  • 19. Visualization 1 factor(opt_trt) 1 TNF 1 DMARD Visualization of the Categorical Response Variable and TreatmentID When Angina = 1 and COPD = 1 Female C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 18 / 24
  • 20. Visualization 2 factor(opt_trt) 1 TNF 1 MTX Visualization of the Categorical Response Variable and TreatmentID When Angina = 1 and COPD = 1 Male 1 DMARD C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 19 / 24
  • 21. Visualization 3 factor(opt_trt) 1 TNF 1 MTX Visualization of the Categorical Response Variable and TreatmentID When Angina = 0 and COPD = 0 Female C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 20 / 24
  • 22. Visualization 4 factor(opt_trt) 1 TNF 1 MTX Visualization of the Categorical Response Variable and TreatmentID When Angina = 0 and COPD = 0 Male 1 DMARD 1 DMARD & 1 MTX C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 21 / 24
  • 23. Future Study 1 Making multiple decisions across half year can be extended. 2 Model selection procedure in multiple stages will be conducted. 3 We will obtain the confidence interval for predicted ˜Y. 4 Semi-parametric or non-parametric model will be implemented to fit the data. 5 Methods dealing with not randomly censored data will be developed. 6 Model global effect instead of the last stage effect will be considered. C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 22 / 24
  • 24. Acknowledgements Grant Weller, Victoria Mansfield, Yinglong Guo UnitedHealth Group, Research & Development Daniel Luckett University of North Carolina, Department of Biostatistics SAMSI Department of Mathematics, NC State University C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 23 / 24
  • 25. Questions Thank you for your attention. Questions? C. Chen et al. (SAMSI IMSM 2018) Precision Treatment for RA July 25, 2018 24 / 24