Mental health informatics is the use of data science tools to improve mental health outcomes. Binge drinking is tied to several negative mental health outcomes including increased severity of anxiety and depression symptoms. Many existing methods of identifying binge drinking risk factors fail to account for data heterogeneity within a population. To address this limitation, we formulate a subpopulation binge drinking behavioral risk factors ranking problem, under the framework of multi-task learning (MTL). This outputs a ranked list of binge drinking behavioral risk factors for each subpopulation (task) concurrently while incorporating shared data across tasks. We then analyzed the ranked lists of risk factors and compared the top ranking factors for each MTL setting. We conclude that MTL can be effectively used to obtain risk factors for binge drinking and suggest future research directions.

We review HCI history from both the perspective of its 1980s split with human factors and its nature as a discipline. We then revisit human augmentation as an alternative to user friendliness that seems particularly relevant in the areas of inclusive design and artificial intelligence. Viewing human-AI interaction as a kind of human augmentation raises issues such as how to promote trust and situation awareness. We also pose the question: Can HCI and human factors engineering work together to solve the increasingly urgent challenges of human-AI technology? In an initial look at this question, we contrast the different approaches of HCI and human factors on emerging AI research. This article concludes by considering other potentially promising paths for HCI. We propose more collaboration between HCI and human factors, or related disciplines, in the future to address the massive challenges posed by the rapid growth in data science and artificial intelligence.

Major depressive disorder (MDD) is the most common mental health disorder and is one of the leading preventable causes of death in the United States (U.S.). It is also recognized as a global problem by the World Health organization (WHO). The persistence of MDD leads to many negative consequences including suicide and disability. The Hamilton Rating Scale for Depression (HAM-D) evaluates depression severity based on 17 risk factors (symptoms) of depression. Risk factor analysis is a process to identify and understand the risk factors contributing to a particular disease, and is an essential component in the development of efficient and effective prevention and intervention efforts. Most existing methods use a one-size-fits-all model to identify the risk factors at the population-level. However, this type of method fails to account for data heterogeneity within a population. To overcome this limitation, we formulate a subpopulation specific MDD risk factors (symptoms) ranking problem, under the framework of multi-task learning (MTL), to identify a ranked list of MDD risk factors for each subpopulation (task) simultaneously while utilizing appropriate shared information across tasks. By synchronously learning multiple related tasks, MTL provides a paradigm to rank risk factors both at the subpopulation and population-level. To the best of our knowledge, this is the first study to investigate HAM-D using MTL.

The Hamilton Rating Scale for Depression (HAM-D) evaluates depression severity based on 17 symptoms of major depressive disorder (MDD). Understanding the clustering of symptoms within MDD is helpful for the identification of depression subgroups that might differ in how they respond to various treatment interventions. In the research reported in this paper, we employed hierarchical clustering to generate a symptom clustering hierarchy. We compare our findings with previously identified factor analytic HAM-D symptom groups. Our findings are comparatively assessed with previously identified HAM-D symptom groups using factor analysis.

Being able to accurately predict the time to event of interest, commonly known as survival analysis, is extremely beneficial in many real-world applications. Traditional commonly used statistical survival analysis methods, e.g., Cox proportional hazards model and parametric censored regressions, are based on strong and sometimes impractical assumptions and can only handle linearity relationship between features and target. Recently, deep learning based formulations have been proposed for survival analysis to handle non-linearity. However, these existing deep learning methods either inherit strong assumptions from their corresponding base models or tailor discrete-time survival analysis. To overcome the limitations within these existing models in the literature, we propose an objective function to guide the training of a deep learning model for continuous-time survival analysis. The objective function combines both ranking based and point-wise regression based losses. The ranking based loss measures the goodness of the orders of the predicted survival time for all instances. The point-wise based loss measures the difference between the predicted survival time and the true survival time for the right censored time-to-event data. More specifically, we derive two versions of the ranking based loss from the smoothed concordance index, and two versions of point-wise based loss based on the normalized mean squared error (MSE) and mean absolute error (MAE). Thus, the proposed formulation is capable of dealing with the continuous-time survival analysis from both global and local perspectives. We conduct experimental analysis over several large-scale real-world time-to-event datasets, and the results demonstrate that our model outperforms the state-of-the-art survival analysis methods. The codes and data used in the experiments are available in the link1.1https://github.com/yanlirock/local_global_survival

Accurately predicting the time to an event of interest is an important problem in a wide range of real-world applications. However, prediction is often difficult because many medical datasets have a large number of unlabeled (“censored”) instances because labeling is costly and time consuming. Survival analysis focuses on labeled data to predict the time to an event of interest, such as time of death, or conversion to a different stage in a progressive disease. Grouping structure, which naturally exists in medical datasets, can be exploited to improve generalization performance by learning multiple related survival prediction tasks for subgroups collaboratively. Thus a multi-task learning framework can connect multiple survival prediction tasks (for different subgroups) and learn them simultaneously. In order to take into account both censored information, as well as discover the grouping structure, we propose a novel cluster-boosted multitask learning framework for survival analysis that boosts survival prediction performance. We develop an efficient algorithm and demonstrate the performance of the proposed cluster-boosted multi-task survival analysis method on The Cancer Genome Atlas (TCGA) dataset. Our results show that the proposed approach can significantly improve prediction performance in survival analysis while also identifying different subgroups of cancer patients.

Data exfiltration is a serious threat to organizations. Such exfiltrations cause breach events that can lead to millions of dollars of loss. Perimeter defense is not enough by itself since successful exploits from insiders can also be very damaging. Internal network user activities need to be monitored to detect malicious actions. Automatic machine learning methods can be applied for network anomaly detection, but they create a lot of false alarms. Domain experts can identify malicious users, but they are unable to process large volumes of data. Interactive machine learning (iML) deals with this tradeoff by creating an efficient collaboration between domain experts and machine learning algorithms. Previous research in iML has focused mainly on collaboration with non-experts. The design and requirements for expertise-driven iML have yet to be delineated for cybersecurity applications. In this research, we proposed an Active Learning (AL) model trained with outputs from a liberal (outputting many false alarms as well as possible hits) anomaly detection (AD) criterion to study expert-iML collaboration in anomaly detection. The results showed that: iML in this context can prune false alarms and minimize misses; the performance/compatibility tradeoff that typically occurs in conventional machine learning updates may be less salient in iML. We suggest that compatibility between experts and algorithms can be improved by presenting information about feature relevance during the training process.

Purpose. This study evaluates the predictive ability of patient types (clusters of similar patients) in identifying cancer patients at high risk for emergency department (ED) visits within one year (365 days) following their index date. A descriptive and retrospective cohort study of 45,356 unique cancer patients with only one primary cancer type and at least one ED visit was done using linked administrative sources of health care data. Methods. Three outcomes were investigated in this study. First, the time of ED visit following an index date was predicted using multiple linear regression. Second, those patients who visited an ED within seven days of their index date were detected using logistic regression. In addition to predicting an emergency department visit, vital status of patients was also predicted using logistic regression. We implemented the linear/logistic regression first on unclustered raw data and then on clustered data. The results of these two analyses were then compared. Conclusion. Clustering was found to contribute to a modest improvement in prediction accuracy for all three outcome variables. The results are discussed in terms of the predictive ability of patient types in development of clinical support tools with respect to privacy of patients and their implications for better allocation of resources to cancer patients.

There has been considerable research on developing symptom clusters for cancer patients but finding a consistent and replicable typology of patients in terms of the kinds of symptoms they experience remains elusive (e.g., [1]). One reason for lack of consensus may be that analyses are unfocused, and use different combinations of clustering variables. This paper identifies subgroups of similar cancer patients within a sample of 18,535 patients, based on the Edmonton Symptom Assessment System (ESAS) and also a set of inpatient and outpatient variables using linked administrative sources of Ontario healthcare data. K-means cluster analysis was performed on cancer patients having only one primary cancer type, and who had at least one emergency department (ED) visit. Only variables that had a Pearson correlation of at least 0.1 with the number of days until visiting an ED, after an assessment, were included in the cluster analysis. While information about next emergency department visit was not included as a clustering variable, the number of patient types was chosen so as to minimize the mean absolute error of predictions of next emergency department visits by cancer patients. The next emergency department visit, after the last assessment date of each patient, was predicted for each cluster solution. The cluster solution with maximum accuracy/minimum MAE was used to derive the final set of patient types. With the help of physicians, and guided by the results of these analyses, a description of each patient type was created. Based on our results we grouped cancer patients into four types that differ in terms of type of cancer, stage of cancer, and symptomology. Implications for symptom management and reduction of ED visits are discussed.

Obesity has become a significant threat to health. Identifying and understanding the underlying obesity risk factors (ORFs) are crucial for optimizing prevention, intervention and treatment for obesity. Most existing methodological approaches to risk factor analysis are employed within the single task learning (STL) framework to learn a ranked list of ORFs for a whole population. However, obesity is a multi-faced health outcome. Some ORFs are highly specific to a certain subpopulation and others are universal to the entire population. Multi-task learning (MTL) framework offers a solution to connect multiple related tasks. Within the MTL framework, we implement two tailor-made models, i.e., multi-task feature learning (MTFL) and clustered multi-task learning (CMTL), to conduct ORFs analysis. The former is capable of finding the universal ORFs for all subpopulations without sacrificing the uniqueness of each subpopulation. The latter uncovers the grouping structure and conducts multi-level ORFs analysis simultaneously. Experiments on a public behavioral dataset demonstrate a superior performance of our methods in prioritizing multi-level ORFs.

Collecting labeling information of time-to-event analysis is naturally very time consuming, i.e., one has to wait for the occurrence of the event of interest, which may not always be observed for every instance. By taking advantage of censored instances, survival analysis methods internally consider more samples than standard regression methods, which partially alleviates this data insufficiency problem. Whereas most existing survival analysis models merely focus on a single survival prediction task, when there are multiple related survival prediction tasks, we may benefit from the tasks relatedness. Simultaneously learning multiple related tasks, multi-task learning (MTL) provides a paradigm to alleviate data insufficiency by bridging data from all tasks and improves generalization performance of all tasks involved. Even though MTL has been extensively studied, there is no existing work investigating MTL for survival analysis. In this paper, we propose a novel multi-task survival analysis framework that takes advantage of both censored instances and task relatedness. Specifically, based on two common used task relatedness assumptions, i.e., low-rank assumption and cluster structure assumption, we formulate two concrete models, COX-TRACE and COX-cCMTL, under the proposed framework, respectively. We develop efficient algorithms and demonstrate the performance of the proposed multi-task survival analysis models on the The Cancer Genome Atlas (TCGA) dataset. Our results show that the proposed approaches can significantly improve the prediction performance in survival analysis and can also discover some inherent relationships among different cancer types.

Over-dispersed network data mining has emerged as a central theme in data science, evident by a sharp increase in the volume of real-world network data with imbalanced clusters.While most of existing clustering methods are designed for discovering the number of clusters and class specific connectivity patterns, few methods are available to uncover the imbalanced clusters,commonly existing in network communities and image segments.In this paper, we propose a generalized probabilistic modeling framework,SizeConnectivity, to estimate over-dispersed cluster size distribution together with class specific connectivity patterns from network data.We performed extensive synthetic and real-world experiments on clustering social network data and image data for detecting network communities and image segments.Our results demonstrate a superior performance of our SizeConnectivity clustering method in recovering the hidden structure of network data via modeling over-dispersion.

In survival analysis, the primary goal is to monitor several entities and model the occurrence of a particular event of interest. In such applications, it is quite often the case that the event of interest may not always be observed during the study period and this gives rise to the problem of censoring which cannot be easily handled in the standard regression approaches. In addition, obtaining sufficient labeled training instances for learning a robust prediction model is a very time consuming process and can be extremely difficult in practice. In this paper, we propose a transfer learning based Cox method, called Transfer-Cox, which uses auxiliary data to augment learning when there are insufficient amount of training examples. The proposed method aims to extract "useful" knowledge from the source domain and transfer it to the target domain, thus potentially improving the prediction performance in such time-to-event data. The proposed method uses the l_2,1-norm penalty to encourage multiple predictors to share similar sparsity patterns, thus learns a shared representation across source and target domains, potentially improving the model performance on the target task. To speedup the computation, we apply the screening approach and extend the strong rule to sparse survival analysis models in multiple high-dimensional censored datasets. We demonstrate the performance of the proposed transfer learning method using several synthetic and high-dimensional microarray gene expression benchmark datasets and compare with other related competing state-of-the-art methods. Our results show that the proposed screening approach significantly improves the computational efficiency of the proposed algorithm without compromising the prediction performance. We also demonstrate the scalability of the proposed approach and show that the time taken to obtain the results is linear with respect to both the number of instances and features.