OBJECTIVE: This study examined the use of machine learning (ML) and domain-specific enrichment in patient-generated health data, in the form of free-text meal logs, to classify meals on alignment with different nutritional goals. MATERIALS AND METHODS: We used a dataset of over 3000 meal records collected by 114 individuals from a diverse, low-income community in a major US city using a mobile app. Registered dietitians (RDs) provided expert judgment for meal-goal alignment, used as the "gold-standard" for evaluation. Using text embeddings (TF-IDF and BERT) and domain-specific enrichment information (ontologies, ingredient parsers, and macronutrient contents) as inputs, we evaluated the performance of logistic regression and multilayer perceptron classifiers using accuracy, precision, recall, and F1 score against the gold standard and the individual's self-assessment. RESULTS: On average, individuals who logged meals achieved 0.576 accuracy of meal-goal alignment self-assessments. Even without enrichment, ML outperformed individual's self-assessments, with accuracies within 0.726-0.841 for different goals. The best-performing combination of ML classifier with enrichment achieved even higher accuracies (0.814-0.902). In general, ML classifiers with enrichment of parsed ingredients, food entities, and macronutrients information performed well across multiple nutritional goals, but there was variability in the impact of enrichment and classification algorithm on accuracy of classification for different nutritional goals. CONCLUSION: ML can utilize unstructured free-text meal logs and reliably classify whether meals align with specific nutritional goals, exceeding individuals' self-assessments, especially when incorporating nutrition domain knowledge. Our findings highlight the potential of ML analysis of patient-generated health data to support patient-centered nutrition guidance in precision healthcare.

We introduce a novel stochastic variational inference method for Gaussian process ($\mathcal{GP}$) regression, by deriving a posterior over a learnable set of coresets: i.e., over pseudo-input/output, weighted pairs. Unlike former free-form variational families for stochastic inference, our coreset-based variational $\mathcal{GP}$ (CVGP) is defined in terms of the $\mathcal{GP}$ prior and the (weighted) data likelihood. This formulation naturally incorporates inductive biases of the prior, and ensures its kernel and likelihood dependencies are shared with the posterior. We derive a variational lower-bound on the log-marginal likelihood by marginalizing over the latent $\mathcal{GP}$ coreset variables, and show that CVGP’s lower-bound is amenable to stochastic optimization. CVGP reduces the dimensionality of the variational parameter search space to linear $\mathcal{O}(M)$ complexity, while ensuring numerical stability at $\mathcal{O}(M^3)$ time complexity and $\mathcal{O}(M^2)$ space complexity. Evaluations on real-world and simulated regression problems demonstrate that CVGP achieves superior inference and predictive performance than state-of-the-art, stochastic sparse $\mathcal{GP}$ approximation methods.

The primary practice of healthcare artificial intelligence (AI) starts with model development, often using state-of-the-art AI, retrospectively evaluated using metrics lifted from the AI literature like AUROC and DICE score. However, good performance on these metrics may not translate to improved clinical outcomes. Instead, we argue for a better development pipeline constructed by working backward from the end goal of positively impacting clinically relevant outcomes using AI, leading to considerations of causality in model development and validation, and subsequently a better development pipeline. Healthcare AI should be "actionable," and the change in actions induced by AI should improve outcomes. Quantifying the effect of changes in actions on outcomes is causal inference. The development, evaluation, and validation of healthcare AI should therefore account for the causal effect of intervening with the AI on clinically relevant outcomes. Using a causal lens, we make recommendations for key stakeholders at various stages of the healthcare AI pipeline. Our recommendations aim to increase the positive impact of AI on clinical outcomes.

Intelligent systems for self-management can help patients and improve quality of life. However, designing AI-based systems is challenging because designers need to account not only for user needs, but also for capabilities and practical constraints of underlying algorithms. We propose and implement a human-centered AI framework to align human and technological requirements and constraints that can guide design of intelligent systems for personal health. We use concepts from a machine learning technique, reinforcement learning, to elicit user needs, through directed content analysis of user interviews, and uncover practical data constraints, through analysis of "in the wild" user engagement logs from a self-monitoring app. We gather and triangulate human-machine-data requirements for a self-management tool for individuals with endometriosis - a poorly understood, complex chronic condition with no reliable treatment. We present recommendations for developing a system that aligns with needs, capabilities, and constraints from human user, data, and machine learning perspectives.

OBJECTIVE: The study sought to build predictive models of next menstrual cycle start date based on mobile health self-tracked cycle data. Because app users may skip tracking, disentangling physiological patterns of menstruation from tracking behaviors is necessary for the development of predictive models. MATERIALS AND METHODS: We use data from a popular menstrual tracker (186 000 menstruators with over 2 million tracked cycles) to learn a predictive model, which (1) accounts explicitly for self-tracking adherence; (2) updates predictions as a given cycle evolves, allowing for interpretable insight into how these predictions change over time; and (3) enables modeling of an individual's cycle length history while incorporating population-level information. RESULTS: Compared with 5 baselines (mean, median, convolutional neural network, recurrent neural network, and long short-term memory network), the model yields better predictions and consistently outperforms them as the cycle evolves. The model also provides predictions of skipped tracking probabilities. DISCUSSION: Mobile health apps such as menstrual trackers provide a rich source of self-tracked observations, but these data have questionable reliability, as they hinge on user adherence to the app. By taking a machine learning approach to modeling self-tracked cycle lengths, we can separate true cycle behavior from user adherence, allowing for more informed predictions and insights into the underlying observed data structure. CONCLUSIONS: Disentangling physiological patterns of menstruation from adherence allows for accurate and informative predictions of menstrual cycle start date and is necessary for mobile tracking apps. The proposed predictive model can support app users in being more aware of their self-tracking behavior and in better understanding their cycle dynamics.

In many biomedical, science, and engineering problems, one must sequentially decide which action to take next so as to maximize rewards. One general class of algorithms for optimizing interactions with the world, while simultaneously learning how the world operates, is the multi-armed bandit setting and, in particular, the contextual bandit case. In this setting, for each executed action, one observes rewards that are dependent on a given ’context’, available at each interaction with the world. The Thompson sampling algorithm has recently been shown to enjoy provable optimality properties for this set of problems, and to perform well in real-world settings. It facilitates generative and interpretable modeling of the problem at hand. Nevertheless, the design and complexity of the model limit its application, since one must both sample from the distributions modeled and calculate their expected rewards. We here show how these limitations can be overcome using variational inference to approximate complex models, applying to the reinforcement learning case advances developed for the inference case in the machine learning community over the past two decades. We consider contextual multi-armed bandit applications where the true reward distribution is unknown and complex, which we approximate with a mixture model whose parameters are inferred via variational inference. We show how the proposed variational Thompson sampling approach is accurate in approximating the true distribution, and attains reduced regrets even with complex reward distributions. The proposed algorithm is valuable for practical scenarios where restrictive modeling assumptions are undesirable.

We present a novel Rao-Blackwellized multiple particle filtering method for inference of correlated latent states observed via nonlinear functions. We adopt a state-space framework and model the dynamic correlated states using a mixing matrix, embedded in white Gaussian noise. The critical challenges in practice are the lack of knowledge about the mixing parameters and the possibly large dimensionality of the state. We address these issues by implementing Rao-Blackwellization of the unknown parameters and adopting a divide-and-conquer approach. The former strategy amounts to marginalizing out some of the variables; the latter breaks the space of the system in subsystems, and runs a separate particle filter for each of them. The resulting Rao-Blackwellized multiple particle filtering accurately estimates the correlated latent states, as shown by the provided simulation results.

In this paper, we consider state-space models where the latent processes represent correlated mixtures of fractional Gaussian processes embedded in white Gaussian noises. The observed data are nonlinear functions of the latent states. The fractional Gaussian processes have interesting properties including long-memory, self-similarity and scale-invariance, and thus, are of interest for building models in finance and econometrics. We propose sequential Monte Carlo (SMC) methods for inference of the latent processes where each method is based on different assumptions about the parameters of the state-space model. The methods are extensively evaluated via simulations of the popular stochastic volatility model.

This paper considers inference on the widely used state-space models described by hidden ARMA state processes of unknown order observed via non-linear functions of the states. We propose a particle filtering method for sequentially inferring the unknown ARMA time-series by Rao-Blackwellization of all the static unknowns. Our method does not rely either on any assumption on the model order or on the static ARMA and state innovation parameters. Consequently, when the ARMA model order is unknown, it can be used without a follow-up model selection procedure. Extensive simulation results validate the proposed method across different ARMA models.

There are many practical signal processing settings where a state-space model consists of a state described by an ARMA process that is observed via non-linear functions of the state. In this paper, we propose a particle filtering method for sequentially estimating the ARMA process in the presence of unknown parameters. In the considered problem, we have static and dynamic unknowns, and we show how to handle the static parameters so that the estimation of the state process does not degrade with time. We propose a new particle filter that approximates the posterior of all the unknowns by a Gaussian distribution, in combination with a Monte Carlo approach to the Rao-Blackwellization of the static parameters. We demonstrate the performance of the proposed method by extensive computer simulations.

In this paper, we propose a novel approach for decomposing hedge fund returns onto observable risk factors. We utilize a vector stochastic-volatility model to extract the time-varying exposure of low frequency hedge fund returns on high frequency market data. We implement the estimation by using particle filtering and the concept of Rao-Blackwellization. With the latter, we remove all the static parameters of the model and thereby reduce the dimension of the parameter space for particle generation. Thus, we are able to obtain accurate estimates of the posterior distributions of the model states. For our model, this reduction is significant because the number of static parameters is large. We use the proposed model to analyze hedge fund performance and to optimally replicate hedge fund strategies economically. We demonstrate the validity and effectiveness of the method by computer simulations.

Both energy efficiency and high data relevancy are crucial for wireless sensor network applications. Network clustering and event-driven protocols are two main approaches typically used to fulfill those requirements. Existing techniques are either focused only on performance of clusterheads or too restricted to specific applications; however, few of previous work took advantage of the combination of both approaches. In this paper, we formalize the combined problem of network clustering in an event-driven manner, show its NP-Completeness and present an innovative distributed heuristic solution. We hereby propose AWARE, an event-driven and energy-aware network clustering technique. AWARE groups highly active nodes together and thus make clusterheads report more efficiently with only relevant data. This approach guarantees the benefit of both reduced energy consumption and aggregated highly relevant data. Furthermore, AWARE is not only able to recover from node failures for in-network communication, but also provides a communication structure for high level services. We theoretically prove that AWARE indeed provides the solution to the problem, as it completes in constant time, forms connected clusters and can recover from node failures. Finally, our extensive performance studies also validate the effectiveness and efficiency of AWARE.

Increased interest in Wireless Sensor Networks (WSNs) by scientists and engineers is forcing WSN research to focus on application requirements. Data is available as never before in many fields of study; practitioners are now burdened with the challenge of doing data-rich research rather than being data-starved. In-situ sensors can be prone to errors, links between nodes are often unreliable, and nodes may become unresponsive in harsh environments, leaving to researchers the onerous task of deciphering often anomalous data. Presented here is the REDFLAG fault detection service for WSN applications, a Run-timE, Distributed, Flexible, detector of faults, that is also Lightweight And Generic. REDFLAG addresses the two most worrisome issues in data-driven wireless sensor applications: abnormal data and missing data. REDFLAG exposes faults as they occur by using distributed algorithms in order to conserve energy. Simulation results show that REDFLAG is lightweight both in terms of footprint and required power resources while ensuring satisfactory detection and diagnosis accuracy. Because REDFLAG is unrestrictive, it is generically available to a myriad of applications and scenarios.