The Cox Proportional Hazards (CPH) model has long been the preferred survival model for its explainability. However, to increase its predictive power beyond its linear log-risk, it was extended to utilize deep neural networks, sacrificing its explainability. In this work, we explore the potential of self-explaining neural networks (SENN) for survival analysis. We propose a new locally explainable Cox proportional hazards model, named CoxSE, by estimating a locally-linear log-hazard function using the SENN. We also propose a modification to the Neural additive (NAM) model, hybrid with SENN, named CoxSENAM, which enables the control of the stability and consistency of the generated explanations. Several experiments using synthetic and real datasets are presented, benchmarking CoxSE and CoxSENAM against a NAM-based model, a DeepSurv model explained with SHAP, and a linear CPH model. The results show that, unlike the NAM-based model, the SENN-based model can provide more stable and consistent explanations while maintaining the predictive power of the black-box model. The results also show that, due to their structural design, NAM-based models demonstrate better robustness to non-informative features. Among the models, the hybrid model exhibits the best robustness. Full implementation is available on GitHub. 1

Abstract We propose a new sequential decision-making setting, combining key aspects of two established online learning problems with bandit feedback. The optimal action to play at any given moment is contingent on an underlying changing state that is not directly observable by the agent. Each state is associated with a context distribution, possibly corrupted, allowing the agent to identify the state. Furthermore, states evolve in a Markovian fashion, providing useful information to estimate the current state via state history. In the proposed problem setting, we tackle the challenge of deciding on which of the two sources of information the agent should base its action selection. We present an algorithm that uses a referee to dynamically combine the policies of a contextual bandit and a multi-armed bandit. We capture the time-correlation of states through iteratively learning the action-reward transition model, allowing for efficient exploration of actions. Our setting is motivated by adaptive mobile health (mHealth) interventions. Users transition through different, time-correlated, but only partially observable internal states, determining their current needs. The side information associated with each internal state might not always be reliable, and standard approaches solely rely on the context risk of incurring high regret. Similarly, some users might exhibit weaker correlations between subsequent states, leading to approaches that solely rely on state transitions risking the same. We analyze our setting and algorithm in terms of regret lower bound and upper bounds and evaluate our method on simulated medication adherence intervention data and several real-world data sets, showing improved empirical performance compared to several popular algorithms.

The Concordance Index (C-index) is a commonly used metric in Survival Analysis for evaluating the performance of a prediction model. In this paper, we propose a decomposition of the C-index into a weighted harmonic mean of two quantities: one for ranking observed events versus other observed events, and the other for ranking observed events versus censored cases. This decomposition enables a finer-grained analysis of the relative strengths and weaknesses between different survival prediction methods. The usefulness of this decomposition is demonstrated through benchmark comparisons against classical models and state-of-the-art methods, together with the new variational generative neural-network-based method (SurVED) proposed in this paper. The performance of the models is assessed using four publicly available datasets with varying levels of censoring. Using the C-index decomposition and synthetic censoring, the analysis shows that deep learning models utilize the observed events more effectively than other models. This allows them to keep a stable C-index in different censoring levels. In contrast to such deep learning methods, classical machine learning models deteriorate when the censoring level decreases due to their inability to improve on ranking the events versus other events.

Deep convolutional neural networks (CNNs) have been shown to predict poverty and development indicators from satellite images with surprising accuracy. This paper presents a first attempt at analyzing the CNNs responses in detail and explaining the basis for the predictions. The CNN model, while trained on relatively low resolution day- and night-time satellite images, is able to outperform human subjects who look at high-resolution images in ranking the Wealth Index categories. Multiple explainability experiments performed on the model indicate the importance of the sizes of the objects, pixel colors in the image, and provide a visualization of the importance of different structures in input images. A visualization is also provided of type images that maximize the network prediction of Wealth Index, which provides clues on what the CNN prediction is based on.

BACKGROUND: Electronic health records (EHRs) are generated at an ever-increasing rate. EHR trajectories, the temporal aspect of health records, facilitate predicting patients' future health-related risks. It enables healthcare systems to increase the quality of care through early identification and primary prevention. Deep learning techniques have shown great capacity for analyzing complex data and have been successful for prediction tasks using complex EHR trajectories. This systematic review aims to analyze recent studies to identify challenges, knowledge gaps, and ongoing research directions. METHODS: For this systematic review, we searched Scopus, PubMed, IEEE Xplore, and ACM databases from Jan 2016 to April 2022 using search terms centered around EHR, deep learning, and trajectories. Then the selected papers were analyzed according to publication characteristics, objectives, and their solutions regarding existing challenges, such as the model's capacity to deal with intricate data dependencies, data insufficiency, and explainability. RESULTS: After removing duplicates and out-of-scope papers, 63 papers were selected, which showed rapid growth in the number of research in recent years. Predicting all diseases in the next visit and the onset of cardiovascular diseases were the most common targets. Different contextual and non-contextual representation learning methods are employed to retrieve important information from the sequence of EHR trajectories. Recurrent neural networks and the time-aware attention mechanism for modeling long-term dependencies, self-attentions, convolutional neural networks, graphs for representing inner visit relations, and attention scores for explainability were frequently used among the reviewed publications. CONCLUSIONS: This systematic review demonstrated how recent breakthroughs in deep learning methods have facilitated the modeling of EHR trajectories. Research on improving the ability of graph neural networks, attention mechanisms, and cross-modal learning to analyze intricate dependencies among EHRs has shown good progress. There is a need to increase the number of publicly available EHR trajectory datasets to allow for easier comparison among different models. Also, very few developed models can handle all aspects of EHR trajectory data.

Survival Analysis models usually output functions (survival or hazard functions) rather than point predictions like regression and classification models. This makes the explanations of such models a challenging task, especially using the Shapley values. We propose SurvSHAP, a new model-agnostic algorithm to explain survival models that predict survival curves. The algorithm is based on discovering patterns in the predicted survival curves that would identify significantly different survival behaviors, and utilizing a proxy model and SHAP method to explain these distinct survival behaviors. Experiments on synthetic and real datasets demonstrate that the SurvSHAP is able to capture the underlying factors of the survival patterns. Moreover, SurvSHAP results on the Cox Proportional Hazard model are compared with the weights of the model to show that we provide faithful overall explanations, with more fine-grained explanations of the sub-populations. We also illustrate the wrong model and explanations learned by a Cox model when applied to heterogeneous sub-populations. We show that a non-linear machine learning survival model with SurvSHAP can better model the data and provide better explanations than linear models.

Molecular mechanisms employed by individual multipotent cells at the point of lineage commitment remain largely uncharacterized. Current paradigms span from instructive to noise-driven mechanisms. Of considerable interest is also whether commitment involves a limited set of genes or the entire transcriptional program, and to what extent gene expression configures multiple trajectories into commitment. Importantly, the transient nature of the commitment transition confounds the experimental capture of committing cells. We develop a computational framework that simulates stochastic commitment events, and affords mechanistic exploration of the fate transition. We use a combined modeling approach guided by gene expression classifier methods that infers a time-series of stochastic commitment events from experimental growth characteristics and gene expression profiling of individual hematopoietic cells captured immediately before and after commitment. We define putative regulators of commitment and probabilistic rules of transition through machine learning methods, and employ clustering and correlation analyses to interrogate gene regulatory interactions in multipotent cells. Against this background, we develop a Monte Carlo time-series stochastic model of transcription where the parameters governing promoter status, mRNA production and mRNA decay in multipotent cells are fitted to experimental static gene expression distributions. Monte Carlo time is converted to physical time using cell culture kinetic data. Probability of commitment in time is a function of gene expression as defined by a logistic regression model obtained from experimental single-cell expression data. Our approach should be applicable to similar differentiating systems where single cell data is available. Within our system, we identify robust model solutions for the multipotent population within physiologically reasonable values and explore model predictions with regard to molecular scenarios of entry into commitment. The model suggests distinct dependencies of different commitment-associated genes on mRNA dynamics and promoter activity, which globally influence the probability of lineage commitment.

A quantitative analysis of metastatic bone involvement can be an important prognostic indicator of survival or a tool in monitoring treatment response in patients with cancer. The purpose of this study was to develop a completely automated decision support system for whole-body bone scans using image analysis and artificial neural networks. The study population consisted of 795 whole-body bone scans. The decision support system first detects and classifies individual hotspots as being metastatic or not. A second prediction model then classifies the scan regarding metastatic disease on a patient level. The test set sensitivity and specificity was 95% and 64% respectively, corresponding to 95% area under the receiver operating characteristics curve.

MOTIVATION: The pairwise alignment of biological sequences obtained from an algorithm will in general contain both correct and incorrect parts. Hence, to allow for a valid interpretation of the alignment, the local trustworthiness of the alignment has to be quantified. RESULTS: We present a novel approach that attributes a reliability index to every pair of residues, including gapped regions, in the optimal alignment of two protein sequences. The method is based on a fuzzy recast of the dynamic programming algorithm for sequence alignment in terms of mean field annealing. An extensive evaluation with structural reference alignments not only shows that the probability for a pair of residues to be correctly aligned grows consistently with increasing reliability index, but moreover demonstrates that the value of the reliability index can directly be translated into an estimate of the probability for a correct alignment.

A strategy for finding approximate solutions to discrete optimization problems with inequality constraints using mean field neural networks is presented. The constraints x ≤ 0 are encoded by x⊖(x) terms in the energy function. A careful treatment of the mean field approximation for the self-coupling parts of the energy is crucial, and results in an essentially parameter-free algorithm. This methodology is extensively tested on the knapsack problem of size up to 103items. The algorithm scales like NM for problems with N items and M constraints. Comparisons are made with an exact branch and bound algorithm when this is computationally possible (N ≤ 30). The quality of the neural network solutions consistently lies above 95% of the optimal ones at a significantly lower CPU expense. For the larger problem sizes the algorithm is compared with simulated annealing and a modified linear programming approach. For "nonhomogeneous" problems these produce good solutions, whereas for the more difficult "homogeneous" problems the neural approach is a winner with respect to solution quality and/or CPU time consumption. The approach is of course also applicable to other problems of similar structure, like set covering.