Subscribe to RSS
Patient-Specific Explanations for Predictions of Clinical OutcomesFunding The research reported in this publication was supported by the National Library of Medicine of the National Institutes of Health under award number R01LM012095. The content of the paper is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the University of Pittsburgh.
31 March 2018
07 August 2019
10 November 2019 (online)
Background Machine learning models that are used for predicting clinical outcomes can be made more useful by augmenting predictions with simple and reliable patient-specific explanations for each prediction.
Objectives This article evaluates the quality of explanations of predictions using physician reviewers. The predictions are obtained from a machine learning model that is developed to predict dire outcomes (severe complications including death) in patients with community acquired pneumonia (CAP).
Methods Using a dataset of patients diagnosed with CAP, we developed a predictive model to predict dire outcomes. On a set of 40 patients, who were predicted to be either at very high risk or at very low risk of developing a dire outcome, we applied an explanation method to generate patient-specific explanations. Three physician reviewers independently evaluated each explanatory feature in the context of the patient's data and were instructed to disagree with a feature if they did not agree with the magnitude of support, the direction of support (supportive versus contradictory), or both.
Results The model used for generating predictions achieved a F1 score of 0.43 and area under the receiver operating characteristic curve (AUROC) of 0.84 (95% confidence interval [CI]: 0.81–0.87). Interreviewer agreement between two reviewers was strong (Cohen's kappa coefficient = 0.87) and fair to moderate between the third reviewer and others (Cohen's kappa coefficient = 0.49 and 0.33). Agreement rates between reviewers and generated explanations—defined as the proportion of explanatory features with which majority of reviewers agreed—were 0.78 for actual explanations and 0.52 for fabricated explanations, and the difference between the two agreement rates was statistically significant (Chi-square = 19.76, p-value < 0.01).
Conclusion There was good agreement among physician reviewers on patient-specific explanations that were generated to augment predictions of clinical outcomes. Such explanations can be useful in interpreting predictions of clinical outcomes.
Keywordspredictive model - patient-specific explanation - machine learning - clinical decision support system
Protection of Human and Animal Subjects
All research activities reported in this publication were reviewed and approved by the University of Pittsburgh’s Institutional Review Board.
- 1 Goldstein BA, Navar AM, Pencina MJ, Ioannidis JPA. Opportunities and challenges in developing risk prediction models with electronic health records data: a systematic review. J Am Med Inform Assoc 2017; 24 (01) 198-208
- 2 Rothman B, Leonard JC, Vigoda MM. Future of electronic health records: implications for decision support. Mt Sinai J Med 2012; 79 (06) 757-768
- 3 Steyerberg EW. Clinical Prediction Models: A Practical Approach to Development, Validation, and Updating. New York, NY: Springer; 2009
- 4 Miotto R, Wang F, Wang S, Jiang X, Dudley JT. Deep learning for healthcare: review, opportunities and challenges. Brief Bioinform 2017; 1: 11
- 5 Liu S, Liu S, Cai W, Pujol S, Kikinis R, Feng D. Early diagnosis of Alzheimer's disease with deep learning. IEEE 11th International Symposium on Biomedical Imaging (ISBI) 2014; 1015-1018
- 6 Miotto R, Li L, Kidd BA, Dudley JT. Deep patient: an unsupervised representation to predict the future of patients from the electronic health records. Sci Rep 2016; 6 (01) 26094
- 7 Avati A, Jung K, Harman S, Downing L, Ng A, Shah NH. Improving palliative care with deep learning. IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2017; 6: 311-316
- 8 Razavian N, Marcus J, Sontag D. Multi-task prediction of disease onsets from longitudinal lab tests. Machine Learning for Healthcare Conference. 2016: 73-100
- 9 Choi E, Bahadori MT, Schuetz A, Stewart WF, Sun J. Doctor AI: predicting clinical events via recurrent neural networks. Machine Learning for Healthcare Conference. 2016: 301-318
- 10 Caruana R, Kangarloo H, David J, Dionisio N, Sinha U, Johnson D. Case-based explanation of non-case-based learning methods. Proc AMIA Symp 1999; 212-215
- 11 Reggia JA, Perricone BT. Answer justification in medical decision support systems based on Bayesian classification. Comput Biol Med 1985; 15 (04) 161-167
- 12 Caruana R, Lou Y, Gehrke J, Koch P, Sturm M, Elhadad N. Intelligible models for healthcare: predicting pneumonia risk and hospital 30-day readmission. ACM Digital Library 2015; 1721-1730
- 13 Fine MJ, Auble TE, Yealy DM. , et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997; 336 (04) 243-250
- 14 Lipton ZC. The mythos of model interpretability. Available at: https://arxiv.org/pdf/1606.03490.pdf . Accessed August 30 2019.
- 15 Ribeiro MT, Singh S, Guestrin C. Why should I trust you?: explaining the predictions of any classifier. ACM Digital Library 2016; 1135-1144
- 16 Kim B. Interactive and interpretable machine learning models for human machine collaboration. PhD dissertation. Massachusetts Institute of Technology, 2015
- 17 Turner R. A model explanation system. IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2016:1–6
- 18 Luo G. Automatically explaining machine learning prediction results: a demonstration on type 2 diabetes risk prediction. Health Inf Sci Syst 2016; 4 (01) 2
- 19 Štrumbelj E, Bosnić Z, Kononenko I, Zakotnik B, Kuhar CG. Explanation and reliability of prediction models: the case of breast cancer recurrence. Knowl Inf Syst 2010; 24 (02) 305-324
- 20 Kapoor WN. Assessment of the Variantion and Outcomes of Pneumonia: Pneumonia Patient Outcomes Research Team (PORT) Final Report. Washington DC: Agency for Health Policy and Research (AHCPR); 1996
- 21 Cooper GF, Abraham V, Aliferis CF. , et al. Predicting dire outcomes of patients with community acquired pneumonia. J Biomed Inform 2005; 38 (05) 347-366
- 22 Caruana R. Iterated k-nearest neighbor method and article of manufacture for filling in missing values. United States Patent 6,047,287. May 5, 2000
- 23 Pedregosa F, Varoquaux G, Gramfort A. , et al. Scikit-learn: machine learning in python. J Mach Learn Res 2011; 12 (Oct): 2825-2830
- 24 Van Rijsbergen CJ. Information Retrieval. 2nd ed. Newton, MA, USA: Butterworth-Heinemann; 1979
- 25 Cohen J. A coefficient of agreeement for nominal scales. Educ Psychol Meas 1960; 20: 37-46
- 26 Fleiss JL. Measuring nominal scale agreement among many raters. Psychol Bull 1971; 76 (05) 378-382
- 27 Pearson K. On the criterion that a given system of deviations from the probable in the case of a correlated system of variables is such that it can be reasonably supposed to have arisen from random sampling. The Philosophical Magazine: A Journal of Theoretical Experimental and Applied Physics 1990; 50 (302) 151-175
- 28 McHugh ML. Interrater reliability: the kappa statistic. Biochem Med (Zagreb) 2012; 22 (03) 276-282
- 29 Lim WS, van der Eerden MM, Laing R. , et al. Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study. Thorax 2003; 58 (05) 377-382
- 30 Lundberg SM, Lee SI. A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems 2017:4765–4774
- 31 Krause J, Perer A, Ng K. Interacting with predictions: visual inspection of black-box machine learning models. ACM Conference on Human Factors in Computing Systems. 2016 :5686–5697
- 32 Baehrens D, Schroeter T, Harmeling S, Kawanabe M, Hansen K, Mueller K-R. How to explain individual classification decisions. J Mach Learn Res 2009; 11: 1803-1831
- 33 Sikonja MR, Kononenko I. Explaining classifications for individual instances. IEEE Trans Knowl Data Eng 2008; 20: 589-600
- 34 Štrumbelj E, Kononenko I. Towards a model independent method for explaining classification for individual instances. International Conference on Data Warehousing and Knowledge Discovery. 2008: 273 282
- 35 Lemaire V, Féraud R, Voisine N. Contact personalization using a score understanding method. Proceedings of the International Joint Conference on Neural Networks. 2008: 649-654
- 36 Poulin B, Eisner R, Szafron D. , et al. Visual explanation of evidence in additive classifiers. Proc Conference on Innovative Applications of Artificial Intelligence (IAAI06). 2006: 1822-1829
- 37 Szafron D, Greiner R, Lu P, Wishart D, MacDonell C, Anvik J. , et al. Explaining naïve Bayes classifications. Technical Report. Department of Computing Science, University of Alberta. 2003
- 38 DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44 (03) 837-845
- 39 Robin X, Turck N, Hainard A. , et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics 2011; 12 (01) 77
- 40 Wilson EB. Probable inference, the law of succession, and statistical inference. J Am Stat Assoc 1927; 22 (158) 209-212