Abstract
Using electronic medical records to learn personalized risk trajectories poses significant challenges because often very few samples are available in a patient’s history, and, when available, their information content is highly diverse. In this article, we consider how to integrate sparsely sampled longitudinal data, missing measurements informative of the underlying health status, and static information to estimate (dynamically, as new information becomes available) personalized survival distributions. We achieve this by developing a nonparametric probabilistic model that generates survival trajectories, and corresponding uncertainty estimates, from an ensemble of Bayesian trees in which time is incorporated explicitly to learn variable interactions over time, without needing to specify the longitudinal process beforehand. As such, the changing influence on survival of variables over time is inferred from the data directly, which we analyze with post-processing statistics derived from our model.
- Kartik Ahuja, William Zame, and Mihaela van der Schaar. 2017. DPSCREEN: Dynamic personalized screening. In Advances in Neural Information Processing Systems 30 (NIPS’17). 1321--1332.Google Scholar
- Ahmed M. Alaa, Scott Hu, and Mihaela van der Schaar. 2017. Learning from clinical judgments: Semi-Markov-modulated marked Hawkes processes for risk prognosis. In Proceedings of theInternational Conference of Machine Learning.Google Scholar
- James H. Albert and Siddhartha Chib. 1993. Bayesian analysis of binary and polychotomous response data. Journal of the American Statistical Association 88, 422 (1993), 669--679.Google Scholar
Cross Ref
- Eleni-Rosalina Andrinopoulou, Dimitris Rizopoulos, Ruyun Jin, Ad J. J. C. Bogers, Emmanuel Lesaffre, and Johanna J. M. Takkenberg. 2012. An introduction to mixed models and joint modeling: Analysis of valve function over time. Annals of Thoracic Surgery 93, 6 (2012), 1765--1772.Google Scholar
- Peter C. Austin. 2012. Generating survival times to simulate Cox proportional hazards models with time-varying covariates. Statistics in Medicine 31, 29 (2012), 3946--3958.Google Scholar
- Edmon Begoli, Tanmoy Bhattacharya, and Dimitri Kusnezov. 2019. The need for uncertainty quantification in machine-assisted medical decision making. Nature Machine Intelligence 1, 1 (2019), 20.Google Scholar
Cross Ref
- Melanie L. Bell, Mallorie Fiero, Nicholas J. Horton, and Chiu-Hsieh Hsu. 2014. Handling missing data in RCTs; a review of the top medical journals. BMC Medical Research Methodology 14 (2014), 118.Google Scholar
Cross Ref
- Alexis Bellot and Mihaela van der Schaar. 2018. Boosted trees for risk prognosis. In Proceedings of the Machine Learning for Healthcare Conference. 2--16.Google Scholar
- Alexis Bellot and Mihaela van der Schaar. 2018. Multitask boosting for survival analysis with competing risks. In Advances in Neural Information Processing Systems 31 (NIPS’18). 1390--1399.Google Scholar
- Krishnan Bhaskaran, Harriet J. Forbes, Ian Douglas, David A. Leon, and Liam Smeeth. 2013. Representativeness and optimal use of body mass index (BMI) in the UK clinical practice research datalink (CPRD). BMJ Open 3, 9 (2013), e003389.Google Scholar
- Hugh A. Chipman, Edward I. George, and Robert E. McCulloch. 2010. BART: Bayesian additive regression trees. Annals of Applied Statistics 4, 1 (2010), 266--298.Google Scholar
Cross Ref
- Edward Choi, Mohammad Taha Bahadori, Andy Schuetz, Walter F. Stewart, and Jimeng Sun. 2016. Doctor AI: Predicting clinical events via recurrent neural networks. In Proceedings of the Machine Learning for Healthcare Conference. 301--318.Google Scholar
- Tamara Fernández, Nicolás Rivera, and Yee Whye Teh. 2016. Gaussian processes for survival analysis. In Advances in Neural Information Processing Systems. 5021--5029.Google Scholar
- Loïc Ferrer, Hein Putter, and Cécile Proust-Lima. 2017. Individual dynamic predictions using landmarking and joint modelling: Validation of estimators and robustness assessment. arXiv:1707.03706.Google Scholar
- Jerome H. Friedman. 2001. Greedy function approximation: A gradient boosting machine. Annals of Statistics 29, 5 (2001), 1189--1232.Google Scholar
Cross Ref
- Thomas A. Gerds, Michael W. Kattan, Martin Schumacher, and Changhong Yu. 2013. Estimating a time-dependent concordance index for survival prediction models with covariate dependent censoring. Statistics in Medicine 32, 13 (2013), 2173--2184.Google Scholar
- Yanzhang Gong, S. L. Klingenberg, and C. Gluud. 2006. Systematic review and meta-analysis: D-Penicillamine vs. placebo/no intervention in patients with primary biliary cirrhosis—Cochrane Hepato-Biliary Group. Alimentary Pharmacology 8 Therapeutics 24, 11--12 (2006), 1535--1544.Google Scholar
- Trevor Hastie and Robert Tibshirani. 2000. Bayesian backfitting. Statistical Science 15, 3 (2000), 196--223.Google Scholar
Cross Ref
- Robin Henderson, Peter Diggle, and Angela Dobson. 2000. Joint modelling of longitudinal measurements and event time data. Biostatistics 1, 4 (2000), 465--480.Google Scholar
Cross Ref
- Graeme L. Hickey, Pete Philipson, Andrea Jorgensen, and Ruwanthi Kolamunnage-Dona. 2016. Joint modelling of time-to-event and multivariate longitudinal outcomes: Recent developments and issues. BMC Medical Research Methodology 16, 1 (2016), 117.Google Scholar
- Hemant Ishwaran, Udaya B. Kogalur, Eugene H. Blackstone, and Michael S. Lauer. 2008. Random survival forests. Annals of Applied Statistics 2, 3 (2008), 841--860.Google Scholar
Cross Ref
- Adam Kapelner and Justin Bleich. 2013. bartMachine: Machine learning with Bayesian additive regression trees. arXiv:1312.2171.Google Scholar
- Silvan Licher, Alis Heshmatollah, Kimberly D. van der Willik, Bruno H. Ch Stricker, Rikje Ruiter, Emmely W. de Roos, Lies Lahousse. 2019. Lifetime risk and multimorbidity of non-communicable diseases and disease-free life expectancy in the general population: A population-based cohort study. PLoS Medicine 16, 2 (2019), e1002741.Google Scholar
- Bryan Lim and Mihaela van der Schaar. 2018. Disease-Atlas: Navigating disease trajectories with deep learning. In Proceedings of the Machine Learning for Healthcare Conference.Google Scholar
- Zachary C. Lipton. 2016. The mythos of model interpretability. arXiv:1606.03490.Google Scholar
- Zachary C. Lipton. 2017. The doctor just won’t accept that! arXiv:1711.08037.Google Scholar
- Roderick J. A. Little and Donald B. Rubin. 2014. Statistical Analysis with Missing Data. Vol. 333. John Wiley 8 Sons.Google Scholar
- Matthew Powney, Paula Williamson, Jamie Kirkham, and Ruwanthi Kolamunnage-Dona. 2014. A review of the handling of missing longitudinal outcome data in clinical trials. Trials 15, 1 (2014), 237.Google Scholar
- Rajesh Ranganath, Adler Perotte, Noémie Elhadad, and David Blei. 2016. Deep survival analysis. In Proceedings of the Machine Learning for Healthcare Conference. 101--114.Google Scholar
- Dimitris Rizopoulos. 2011. Dynamic predictions and prospective accuracy in joint models for longitudinal and time-to-event data. Biometrics 67, 3 (2011), 819--829.Google Scholar
Cross Ref
- Patrick Royston. 2004. Multiple imputation of missing values. Stata Journal 4, 3 (2004), 227--41.Google Scholar
Cross Ref
- Judith D. Singer and John B. Willett. 1993. It’s about time: Using discrete-time survival analysis to study duration and the timing of events. Journal of Educational Statistics 18, 2 (1993), 155--195.Google Scholar
- Hossein Soleimani, James Hensman, and Suchi Saria. 2017. Scalable joint models for reliable uncertainty-aware event prediction. IEEE Transactions on Pattern Analysis and Machine Intelligence 40, 8 (2017), 1948--1963.Google Scholar
- Rodney A. Sparapani, Brent R. Logan, Robert E. McCulloch, and Purushottam W. Laud. 2016. Nonparametric survival analysis using Bayesian additive regression trees (BART). Statistics in Medicine 35, 16 (2016), 2741--2753.Google Scholar
- Damian C. Stanziano, Michael Whitehurst, Patricia Graham, and Bernard A. Roos. 2010. A review of selected longitudinal studies on aging: Past findings and future directions. Journal of the American Geriatrics Society 58 (2010), S292--S297.Google Scholar
- Terry M. Therneau and P. M. Grambsch. 2000. Modeling Survival Data: Extending the Cox Model, P. Bickel, P. Diggle, S. Fienberg, et al. (Eds.). Statistics for Biology and Health. Springer.Google Scholar
- Beth Twala, M. C. Jones, and David J. Hand. 2008. Good methods for coping with missing data in decision trees. Pattern Recognition Letters 29, 7 (2008), 950--956.Google Scholar
Digital Library
- Hans C. Van Houwelingen. 2007. Dynamic prediction by landmarking in event history analysis. Scandinavian Journal of Statistics 34, 1 (2007), 70--85.Google Scholar
- Andrew K. Wills, Debbie A. Lawlor, Fiona E. Matthews, Avan Aihie Sayer, Eleni Bakra, Yoav Ben-Shlomo, Michaela Benzeval, et al. 2011. Life course trajectories of systolic blood pressure using longitudinal data from eight UK cohorts. PLoS Medicine 8, 6 (2011), e1000440.Google Scholar
Index Terms
Flexible Modelling of Longitudinal Medical Data: A Bayesian Nonparametric Approach
Recommendations
Semiparametric lack-of-fit tests in an additive hazard regression model
In the semiparametric additive hazard regression model of McKeague and Sasieni (Biometrika 81: 501–514), the hazard contributions of some covariates are allowed to change over time, without parametric restrictions (Aalen model), while the contributions ...
Model specification test in a semiparametric regression model for longitudinal data
We propose a model specification test for whether or not a postulated parametric model (null hypothesis) fits longitudinal data as well as a semiparametric model (alternative hypothesis) does. In the semiparametric model, we suppose that a baseline ...
Full Bayesian inference with hazard mixture models
Bayesian nonparametric inferential procedures based on Markov chain Monte Carlo marginal methods typically yield point estimates in the form of posterior expectations. Though very useful and easy to implement in a variety of statistical problems, these ...






Comments