Towards an Explainable Mortality Prediction Model

Influence functions are analytical tools from robust statistics that can help interpret the decisions of black-box machine learning models. Influence functions can be used to attribute changes in the loss function due to small perturbations in the input features. The current work on using influence functions is limited to the features available before the last layer of deep neural networks (DNNs). We extend the influence function approximation to DNNs by computing gradients in an end-to-end manner and relate changes in the loss function to individual input features using an efficient algorithm. We propose an accurate mortality prediction neural network and show the effectiveness of extended influence functions on the eICU dataset. The features chosen by proposed extended influence functions were more like those selected by human experts than those chosen by other traditional methods.

[1]  Christopher P. Bonafide,et al.  Performance of a Clinical Decision Support Tool to Identify PICU Patients at High Risk for Clinical Deterioration* , 2020, Pediatric critical care medicine : a journal of the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies.

[2]  Alistair E. W. Johnson,et al.  The eICU Collaborative Research Database, a freely available multi-center database for critical care research , 2018, Scientific Data.

[3]  Carlos Guestrin,et al.  Anchors: High-Precision Model-Agnostic Explanations , 2018, AAAI.

[4]  S. Weisberg,et al.  Characterizations of an Empirical Influence Function for Detecting Influential Cases in Regression , 1980 .

[5]  Carlos Guestrin,et al.  "Why Should I Trust You?": Explaining the Predictions of Any Classifier , 2016, ArXiv.

[6]  Naman Agarwal,et al.  Second-Order Stochastic Optimization for Machine Learning in Linear Time , 2016, J. Mach. Learn. Res..

[7]  Scott Lundberg,et al.  A Unified Approach to Interpreting Model Predictions , 2017, NIPS.

[8]  Alexander Binder,et al.  On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation , 2015, PloS one.

[9]  Anna Goldenberg,et al.  What Clinicians Want: Contextualizing Explainable Machine Learning for Clinical End Use , 2019, MLHC.

[10]  Sepp Hochreiter,et al.  Self-Normalizing Neural Networks , 2017, NIPS.

[11]  Ankur Taly,et al.  Axiomatic Attribution for Deep Networks , 2017, ICML.

[12]  Gaël Varoquaux,et al.  Scikit-learn: Machine Learning in Python , 2011, J. Mach. Learn. Res..

[13]  J. Zimmerman,et al.  Acute Physiology and Chronic Health Evaluation (APACHE) IV: Hospital mortality assessment for today’s critically ill patients* , 2006, Critical care medicine.

[14]  Martin Wattenberg,et al.  SmoothGrad: removing noise by adding noise , 2017, ArXiv.

[15]  Lalana Kagal,et al.  Explaining Explanations: An Overview of Interpretability of Machine Learning , 2018, 2018 IEEE 5th International Conference on Data Science and Advanced Analytics (DSAA).

[16]  Percy Liang,et al.  Understanding Black-box Predictions via Influence Functions , 2017, ICML.

[17]  Gordon H. Guyatt,et al.  Variability in Physician Opinion on Limiting Pediatric Life Support , 1999, Pediatrics.

[18]  Avanti Shrikumar,et al.  Learning Important Features Through Propagating Activation Differences , 2017, ICML.

[19]  Thomas Brox,et al.  Striving for Simplicity: The All Convolutional Net , 2014, ICLR.

[20]  Michael Draugelis,et al.  Clinician Perception of a Machine Learning-Based Early Warning System Designed to Predict Severe Sepsis and Septic Shock. , 2019, Critical care medicine.

[21]  Peter Bauer,et al.  SAPS 3—From evaluation of the patient to evaluation of the intensive care unit. Part 2: Development of a prognostic model for hospital mortality at ICU admission , 2005, Intensive Care Medicine.

[22]  Gun Lee,et al.  Performance on the APACHE II, SAPS II, SOFA and the OHCA score of post-cardiac arrest patients treated with therapeutic hypothermia , 2018, PloS one.

[23]  Anitha S. Pillai,et al.  Big Data solutions in Healthcare: Problems and perspectives , 2015, 2015 International Conference on Innovations in Information, Embedded and Communication Systems (ICIIECS).

[24]  Fabio Roli,et al.  Towards Poisoning of Deep Learning Algorithms with Back-gradient Optimization , 2017, AISec@CCS.

[25]  Nitesh V. Chawla,et al.  SMOTE: Synthetic Minority Over-sampling Technique , 2002, J. Artif. Intell. Res..

[26]  Andrew Zisserman,et al.  Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps , 2013, ICLR.