Healthcare Predicitive Analytics for Risk Profiling in Chronic Care: A Bayesian Multitask Learning Approach
2 likes939 views
This document describes a study that developed a Bayesian multitask learning (BMTL) model to predict multiple chronic disease risks using electronic health record data. Specifically, the model aimed to predict risks of stroke, heart attack, and kidney failure in patients with diabetes. The study evaluated the BMTL model against single-task learning baselines and other multitask learning approaches, finding the BMTL model achieved better predictive performance. A counterfactual analysis also suggested the BMTL model could help identify more patients for preventive treatment interventions compared to current practice.
![Bayesian Analysis
5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 14
| =
,
=
|
| ∝ |
• is a random variable of interest; D is observed data
• We have a prior subjective belief about
[as a part of model specification]
• We update our prior belief with the data to form
posterior beliefs about |
[at a result of model fitting]
Bayes Rule:](https://image.slidesharecdn.com/bmtlmisqsocialmediaposting-170531151610/85/Healthcare-Predicitive-Analytics-for-Risk-Profiling-in-Chronic-Care-A-Bayesian-Multitask-Learning-Approach-14-320.jpg)
Healthcare Predicitive Analytics for Risk Profiling in Chronic Care: A Bayesian Multitask Learning Approach
- 1. Healthcare Predictive Analytics for Risk Profiling in Chronic Care: A Bayesian Multitask Learning Approach Yu-Kai Lin (Florida State University) Hsinchun Chen (University of Arizona) Randall A. Brown (Southern Arizona VA Health Care System) Shu-Hsing Li (National Taiwan University) Hung-Jen Yang (Stanford University) 5/27/2017 1Healthcare Predictive Analytics for Risk Profiling in Chronic Care
- 2. Background Healthcare Predictive Analytics for Risk Profiling in Chronic Care: A Bayesian Multitask Learning Approach 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 2
- 3. How to improve chronic care? 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 3 Chronic Care Model Bodenheimer et al. (2002) Chronic Disease Control Brownson and Bright (2004) “Technovigilance” Dixon-Woods et al. (2013) • Community Resources and Policies • Health Care Organization • Self-management Support • Delivery System Design • Decision Support • Clinical Information Systems Data and science-driven decision-making If one consistent message has emerged from the literature on improving quality and safety in health care, it is that high-quality intelligence is indispensable.
- 4. Healthcare analytics for clinical intelligence 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 4 Healthcare Analytics ⇔ Business Analytics in Healthcare • Marketers vs. Clinicians (Fichman et al. 2011) − Marketers: ◦ Consumer profiling for targeted marketing ◦ How likely a particular consumer will click an ad link, download an app, respond to a coupon, … − Clinicians: ◦ Patient profiling for personalized care ◦ How likely a particular patient will develop a complication, experience an adverse medical event, respond to a treatment, …
- 5. Health analytics using EHR data 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 5 • Healthcare predictive analytics using electronic health records (EHRs) is a promising IS research direction • Fichman, Kohli & Krishnan (ISR 2011: Healthcare IS) “Using digital technology to enable new kinds of mathematical healthcare modeling … and how they should be integrated with electronic health records warrants future research attention.” • Chen, Chiang & Storey (MISQ 2012: BI & Analytics) “Over the past decade, electronic health records have been widely adopted in hospitals and clinics worldwide. Significant clinical knowledge and a deeper understanding of patient disease patterns can be gleaned from such collections.”
- 6. Research motivation 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 6 • Chronic Diseases (CDs) in the US − Half of all adults (117 million people) had one or more CDs − 86% of the nation’s health care costs are for treating CDs − Seven of the top 10 causes of death in 2010 were CDs − Maps on the prevalence of diagnosed diabetes: 1994 2000 <4.5% 4.5%–5.9% 6.0%–7.4% 7.5%–8.9% >9.0% 2014 Data source: https://www.cdc.gov/chronicdisease/
- 7. Research motivation (cont.) 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 7 • Patients with chronic diseases are often at risk for multiple complications − Diabetes stroke, heart attack, kidney failure, eye problems, and so on • Surprisingly, almost all the existing clinical risk models are designed to focus on only a single outcome. − Diabetes stroke − Diabetes heart attack − Diabetes kidney diseases
- 8. Single-Task Learning (STL) 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 8 • In the usual approach, each event is modeled independently: Predict Event 1 Predict Event 2 Predict Event K logit = + logit = + … logit = +
- 9. STL is fine, but… 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 9 • If the outcomes are related, the models are perhaps related (that is, coefficients/parameters are related). • If the models are related, a model can perhaps “borrow” information from the other models. Spillover effect in model training
- 10. Multitask Learning (MTL) 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 10 Task 1 Task 2 Task K Data Data Data Training Training Training Trained Model Trained Model Trained Model … … … Task 1 Task 2 Task K Data Data Data Training Trained Model Trained Model Trained Model … … Single-Task Learning Multitask Learning
- 11. Research questions & relevance 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 11 • We are interested in studying: − What are a patient’s risks to an array of events? − How can we model multiple risks simultaneously? − Does simultaneous learning of multiple event risks improve overall predictive performance of each event risk? • Relevance to Information Systems (IS) research − Healthcare IS (Fichman et al. 2011; Bardhan et al. 2014) − Predictive analytics (Shmueli and Koppius 2011) − Design science (Hevner et al. 2004; Gregor and Hevner 2013)
- 12. Model Development Healthcare Predictive Analytics for Risk Profiling in Chronic Care: A Bayesian Multitask Learning Approach 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 12
- 13. Bayesian MultiTask Learning (BMTL) intuition 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 13 Ω1, … , ΩJ are K by K correlation matrices. Ω1 (1) Ω2 ΩJ (1) (1) (2) (2) (2) (K) (K) (K) ...Task 1 Task 2 Task K β1 β1 β1 β2 β2 β2 βJ βJ βJ ... ... ... ...
- 14. Bayesian Analysis 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 14 | = , = | | ∝ | • is a random variable of interest; D is observed data • We have a prior subjective belief about [as a part of model specification] • We update our prior belief with the data to form posterior beliefs about | [at a result of model fitting] Bayes Rule:
- 15. Model Spec. 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 15 Para- meter Distribution/Function Form and Supporting Reference Cauchy distribution (Gelman et al. 2008): ~Cauchy 0, 10 Multivariate normal (MVN) distribution with the horseshoe prior (Carvalho et al. 2010): = , , … , " ~ MVN $, % Σ % Horseshoe prior (Carvalho et al. 2010): % = ' ( ' , (~Half-Cauchy 0,1 Σ Covariance matrix (Barnard et al. 2000) Σ = diag + ∗ Ω ∗ diag + + = . , . , … , . " . Half-Cauchy distribution (Gelman et al. 2008): . ~ Half-Cauchy 0, 2.5 Ω Lewandowski, Kurowicka and Joe (LKJ, 2009): Ω ~LKJ 2, 1 Note for the table: The index j ranges from 1 to J (the total number of predictors), and the index k ranges from 1 to K (the total number of tasks). J K σ Ω Σ β τ rψ K θ N α K
- 16. Model fitting 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 16 • No-U-Turn Sampler (NUTS; Hoffman and Gelman 2014) − A variant of Hamiltonian Monte Carlo (HMC) − Adaptively sets the algorithmic parameters in HMC • 2 Markov chains, 1000 warm-up draws, 1000 sampling draws • For each parameter, convergence was assessed using Gelman and Rubin’s (1992) diagnostic test, a.k.a., 34 statistic, with the value less than 1.2.
- 17. Evaluations Healthcare Predictive Analytics for Risk Profiling in Chronic Care: A Bayesian Multitask Learning Approach 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 17
- 18. Illustration of experiment design 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 18 v1 v2 v3 v4 v5 v6 v7 v8 Time Step 1: Randomly sample a visit from the first half of the patient’s medical history. v1 v2 v3 v4 v5 v6 v7 v8 Time Step 2: Use information available in and before the sampled visit for learning or prediction. v1 v2 v3 v4 v5 v6 v7 v8 Time Step 3: Learn and predict if an event will happen in the next w years. Event occurrence Sampled visit, denoted by v0i
- 19. Summary of data 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 19 • 14,752 adult patients with type 2 diabetes • Events / Complications − stroke (henceforth denoted by STK), acute myocardial infarction (AMI; aka, heart attack), and acute renal failure (ARF) − Exclude patients with all three events occurred before v0i Before v0i During v0i and v0i + w years After v0i + 5 w = 1 w = 2 w = 3 w = 4 w = 5 STK 1507 354 560 685 793 828 47 AMI 485 75 146 178 210 225 20 ARF 410 217 399 488 536 571 37
- 20. Variables in our analysis 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 20 • 179 variables in total − Missing values are imputed using the column mean • Examples Category Example Variables Patient information Age, body weight, male, smoking Diagnoses Three digit ICD-9 codes, e.g., 401 for essential hypertension and 427 for cardiac dysrhythmias Treatments Aspirin, clopidogrel, insulin, isoket, metformin Labs and exams CT scan, low-density lipoprotein cholesterol, serum creatinine, systolic blood pressure Note: ICD-9=International Classification of Diseases, Ninth Revision
- 21. Three sets of evaluations 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 21 1. BMTL vs. single task learning approaches 2. BMTL vs. other multitask learning approaches 3. Counterfactual analysis of practical use
- 22. Evaluations 1 and 2 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 22 • 10-fold cross validation • Area Under the Curve (AUC) − Ranges from 0.5 (a worthless model) to 1 (a perfect model) − The DeLong test of AUC (DeLong et al. 1988) Testing data Training data Fold 1 Fold 2 Fold 3 Fold 10 …… OriginalData TruePositiveRate False Positive Rate
- 23. Evaluation 1 (AUC; 10-fold CV) BMTL vs. STL approaches 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 23 Window (w) Task (k) Models BMTL-Logit B-Logit Logit Logit-lasso 1 STK 0.747 0.725*** 0.723*** 0.735*** 1 AMI 0.778 0.744*** 0.729*** 0.758** 1 ARF 0.863 0.855* 0.847** 0.849*** 3 STK 0.742 0.724*** 0.722*** 0.728*** 3 AMI 0.736 0.703*** 0.699*** 0.704*** 3 ARF 0.833 0.823*** 0.819*** 0.823*** 5 STK 0.739 0.724*** 0.723*** 0.727*** 5 AMI 0.727 0.699*** 0.698*** 0.704*** 5 ARF 0.820 0.812*** 0.809*** 0.814*** Note. Bolded values highlight the best AUC result in a row. *** The AUC result is statistically significantly different from BMTL-Logit at α = 0.01. ** The AUC result is statistically significantly different from BMTL-Logit at α = 0.05. * The AUC result is statistically significantly different from BMTL-Logit at α = 0.1.
- 24. Evaluation 2 (AUC; 10-fold CV) BMTL vs. other MTL approaches 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 24 Window (w) Task (k) Models BMTL-Logit (this study) MTL-Logit (Huang et al. 2012) MTL-Tree (Simm et al. 2014) MTL-ANN (Caruana 1997) 1 STK 0.747 0.746 0.717** 0.660*** 1 AMI 0.778 0.767* 0.737** 0.686** 1 ARF 0.863 0.849* 0.831*** 0.650*** 3 STK 0.742 0.730** 0.702*** 0.677*** 3 AMI 0.736 0.693*** 0.727* 0.680*** 3 ARF 0.833 0.816*** 0.787*** 0.763*** 5 STK 0.739 0.719*** 0.686*** 0.670*** 5 AMI 0.727 0.705** 0.692** 0.653*** 5 ARF 0.820 0.809*** 0.770*** 0.703*** Note. Bolded values highlight the best AUC result in a row. *** The AUC result is statistically significantly different from BMTL-Logit at α = 0.01. ** The AUC result is statistically significantly different from BMTL-Logit at α = 0.05. * The AUC result is statistically significantly different from BMTL-Logit at α = 0.1.
- 25. Evaluation 3 Counterfactual analysis 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 25 • Prediction of risk is not enough—we need evidence that prediction can lead to actions that reduce risk beyond what would occur without the prediction rule. (Grady and Berkowitz 2011) − How to assess the practical value of a predictive model without actual use? − Assumption for our counterfactual analysis: Physicians will always provide guideline-recommended preventive interventions if they believe a patient has a high risk of STK/AMI/ARF. − Among the positive cases (patients with the STK/AMI/ARF events between v0i and v0i + 5 years), what happened to them and what could happen to them given a prediction rule.
- 26. Evaluation 3 Counterfactual analysis 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 26 • Guideline-recommended preventive treatments − Source: “Diabetes Comprehensive Care Plan Guidelines” from the American Association of Clinical Endocrinologists Comorbidity Preventive Treatment STK • Antihypertensive agents • Antiplatelet therapy AMI • Antihypertensive agents • Antiplatelet therapy • Lipid lowering therapy ARF • Antihypertensive agents • Angiotensin receptor blockers • Angiotensin-converting-enzyme inhibitors
- 27. Evaluation 3 Counterfactual analysis 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 27 • From the positive cases, we are interested in the proportions who − actually received preventive interventions at or before v0i? − potentially could receive preventive interventions at v0i, given model predictions? • Practically useful models: small c and large d Predicted Risk (from some model) Low High Preventive treatment prescribed at/before v0i Yes a b No c d High/low risk cutoff level: 2% per year 10% over 5 yrs (Dhamoon and Elkind 2010)
- 28. Evaluation 3 Summary of results 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 28 A lower c (making fewer mistakes) is better A higher d (supporting physicians) is better
- 29. Conclusions Healthcare Predictive Analytics for Risk Profiling in Chronic Care: A Bayesian Multitask Learning Approach 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 29
- 30. Conclusions 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 30 • The proposed BMTL approach outperforms the alternative models in risk profiling, and could support physicians to identify high risk patients. • Multitask learning improves overall learning performance by sharing information across models − Evidence for the spillover effect in model training • Beyond healthcare
- 31. Practical implications Risk profiling in chronic care 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 31 • Hospitals: new healthcare delivery models − Accountable care organizations; bundled payments • Physicians: decision support at the point of care − To err is human • Patients: healthcare spending and # of conditions − Medical Expenditure Panel Survey
- 32. To error is human 5/27/2017 Healthcare Predictive Analytics for Risk Profiling in Chronic Care 32 • Building a better health system with IT and analytics
- 33. Thank you Please send comments to Yu-Kai Lin ylin@business.fsu.edu 5/27/2017 33Healthcare Predictive Analytics for Risk Profiling in Chronic Care