The application of unsupervised deep learning in predictive models using electronic health records.

BACKGROUND: The main goal of this study is to explore the use of features representing patient-level electronic health record (EHR) data, generated by the unsupervised deep learning algorithm autoencoder, in predictive modeling. Since autoencoder features are unsupervised, this paper focuses on their general lower-dimensional representation of EHR information in a wide variety of predictive tasks. METHODS: We compare the model with autoencoder features to traditional models: logistic model with least absolute shrinkage and selection operator (LASSO) and Random Forest algorithm. In addition, we include a predictive model using a small subset of response-specific variables (Simple Reg) and a model combining these variables with features from autoencoder (Enhanced Reg). We performed the study first on simulated data that mimics real world EHR data and then on actual EHR data from eight Advocate hospitals. RESULTS: On simulated data with incorrect categories and missing data, the precision for autoencoder is 24.16% when fixing recall at 0.7, which is higher than Random Forest (23.61%) and lower than LASSO (25.32%). The precision is 20.92% in Simple Reg and improves to 24.89% in Enhanced Reg. When using real EHR data to predict the 30-day readmission rate, the precision of autoencoder is 19.04%, which again is higher than Random Forest (18.48%) and lower than LASSO (19.70%). The precisions for Simple Reg and Enhanced Reg are 18.70 and 19.69% respectively. That is, Enhanced Reg can have competitive prediction performance compared to LASSO. In addition, results show that Enhanced Reg usually relies on fewer features under the setting of simulations of this paper. CONCLUSIONS: We conclude that autoencoder can create useful features representing the entire space of EHR data and which are applicable to a wide array of predictive tasks. Together with important response-specific predictors, we can derive efficient and robust predictive models with less labor in data extraction and model training.

Scaling and contextualizing personalized healthcare: A case study of disease prediction algorithm integration.

Today, advances in medical informatics brought on by the increasing availability of electronic medical records (EMR) have allowed for the proliferation of data-centric tools, especially in the context of personalized healthcare. While these tools have the potential to greatly improve the quality of patient care, the effective utilization of their techniques within clinical practice may encounter two significant challenges. First, the increasing amount of electronic data generated by clinical processes can impose scalability challenges for current computational tools, requiring parallel or distributed implementations of such tools to scale. Secondly, as technology becomes increasingly intertwined in clinical workflows these tools must not only operate efficiently, but also in an interpretable manner. Failure to identity areas of uncertainty or provide appropriate context creates a potentially complex situation for both physicians and patients. This paper will present a case study investigating the issues associated with first scaling a disease prediction algorithm to accommodate dataset sizes expected in large medical practices. It will then provide an analysis on the diagnoses predictions, attempting to provide contextual information to convey the certainty of the results to a physician. Finally it will investigate latent demographic features of the patient's themselves, which may have an impact on the accuracy of the diagnosis predictions.

A Rare Case of Complete Inguinoscrotal Bladder Herniation With Ureteric Involvement: Assessing Diagnostic Challenges and Complex Surgical Management.

Inguinoscrotal hernias involving the urinary bladder are exceedingly rare, constituting a small subset of inguinal hernias. We present a case of a 47-year-old male with long-standing scrotal enlargement and obstructive uropathy due to complete herniation of the bladder with ureteric involvement. Diagnostic imaging confirmed the condition. Following an open laparotomy, the bladder was reduced, and a modified Bassini technique with orchiopexy was used for repair. Recurrence of the inguinoscrotal hernia with evidence of the bladder in the scrotal sac required additional surgery. This case underscores the rarity, diagnostic complexity, and potential complications of inguinoscrotal bladder hernias. Specialized surgical techniques and a multidisciplinary approach are crucial for successful management, especially in cases of complete bladder herniation. Future considerations should include innovative approaches to enhance primary repair outcomes for extensive hernias involving the bladder.

Bringing big data to personalized healthcare: a patient-centered framework.

Faced with unsustainable costs and enormous amounts of under-utilized data, health care needs more efficient practices, research, and tools to harness the full benefits of personal health and healthcare-related data. Imagine visiting your physician's office with a list of concerns and questions. What if you could walk out the office with a personalized assessment of your health? What if you could have personalized disease management and wellness plan? These are the goals and vision of the work discussed in this paper. The timing is right for such a research direction--given the changes in health care, reimbursement, reform, meaningful use of electronic health care data, and patient-centered outcome mandate. We present the foundations of work that takes a Big Data driven approach towards personalized healthcare, and demonstrate its applicability to patient-centered outcomes, meaningful use, and reducing re-admission rates.

Comparative protein interactomics of neuroglobin and myoglobin.

Neuroglobin is a hypoxia-inducible O(2)-binding protein with neuroprotective effects in cell and animal models of stroke and Alzheimer's disease. The mechanism underlying neuroglobin's cytoprotective action is unknown, although several possibilities have been proposed, including anti-oxidative and anti-apoptotic effects. We used affinity purification-mass spectrometry methods to identify neuroglobin-interacting proteins in normoxic and hypoxic murine neuronal (HN33) cell lysates, and to compare these interactions with those of a structurally and functionally related protein, myoglobin. We report that the protein interactomes of neuroglobin and myoglobin overlap substantially and are modified by hypoxia. In addition, neuroglobin-interacting proteins include partners consistent with both anti-oxidative and anti-apoptotic functions, as well as with a relationship to several neurodegenerative diseases.

Genome and proteome annotation using automatically recognized concepts and functional networks.

Many tools have been developed for prediction of the function or disease association of genes and proteins, and this continues to be a highly active area of bioinformatics research. Typically, these methods predict which concepts should be annotated to genes or proteins, using terms from ontologies such as Gene Ontology (GO), largely overlooking other ontologies that are available. Here, we set out to broadly evaluate novel, automatically retrieved, gene-term annotations and identify those concepts of publicly available ontologies that can be predicted using a generalized tool for prediction of annotations. We identified terms that perform better than expected by chance using randomly generated gene sets and show that both manually curated terms in GO and automatically recognized terms can be used to develop reasonable predictive models. In all, we characterize terms in over 250 ontologies and identify more than 127,000 statistically significant terms that can be predicted on human genes.

A public-private partnership develops and externally validates a 30-day hospital readmission risk prediction model.

INTRODUCTION: Preventing the occurrence of hospital readmissions is needed to improve quality of care and foster population health across the care continuum. Hospitals are being held accountable for improving transitions of care to avert unnecessary readmissions. Advocate Health Care in Chicago and Cerner (ACC) collaborated to develop all-cause, 30-day hospital readmission risk prediction models to identify patients that need interventional resources. Ideally, prediction models should encompass several qualities: they should have high predictive ability; use reliable and clinically relevant data; use vigorous performance metrics to assess the models; be validated in populations where they are applied; and be scalable in heterogeneous populations. However, a systematic review of prediction models for hospital readmission risk determined that most performed poorly (average C-statistic of 0.66) and efforts to improve their performance are needed for widespread usage. METHODS: The ACC team incorporated electronic health record data, utilized a mixed-method approach to evaluate risk factors, and externally validated their prediction models for generalizability. Inclusion and exclusion criteria were applied on the patient cohort and then split for derivation and internal validation. Stepwise logistic regression was performed to develop two predictive models: one for admission and one for discharge. The prediction models were assessed for discrimination ability, calibration, overall performance, and then externally validated. RESULTS: The ACC Admission and Discharge Models demonstrated modest discrimination ability during derivation, internal and external validation post-recalibration (C-statistic of 0.76 and 0.78, respectively), and reasonable model fit during external validation for utility in heterogeneous populations. CONCLUSIONS: The ACC Admission and Discharge Models embody the design qualities of ideal prediction models. The ACC plans to continue its partnership to further improve and develop valuable clinical models.

Exploring and exploiting disease interactions from multi-relational gene and phenotype networks.

The availability of electronic health care records is unlocking the potential for novel studies on understanding and modeling disease co-morbidities based on both phenotypic and genetic data. Moreover, the insurgence of increasingly reliable phenotypic data can aid further studies on investigating the potential genetic links among diseases. The goal is to create a feedback loop where computational tools guide and facilitate research, leading to improved biological knowledge and clinical standards, which in turn should generate better data. We build and analyze disease interaction networks based on data collected from previous genetic association studies and patient medical histories, spanning over 12 years, acquired from a regional hospital. By exploring both individual and combined interactions among these two levels of disease data, we provide novel insight into the interplay between genetics and clinical realities. Our results show a marked difference between the well defined structure of genetic relationships and the chaotic co-morbidity network, but also highlight clear interdependencies. We demonstrate the power of these dependencies by proposing a novel multi-relational link prediction method, showing that disease co-morbidity can enhance our currently limited knowledge of genetic association. Furthermore, our methods for integrated networks of diverse data are widely applicable and can provide novel advances for many problems in systems biology and personalized medicine.

Network analysis of team structure in the neonatal intensive care unit.

OBJECTIVE: The goal was to examine nursing team structure and its relationship with family satisfaction. METHODS: We used electronic health records to create patient-based, 1-mode networks of nursing handoffs. In these networks, nurses were represented as nodes and handoffs as edges. For each patient, we calculated network statistics including team size and diameter, network centrality index, proportion of newcomers to care teams according to day of hospitalization, and a novel measure of the average number of shifts between repeat caregivers, which was meant to quantify nursing continuity. We assessed parental satisfaction by using a standardized survey. RESULTS: Team size increased with increasing length of stay. At 2 weeks of age, 50% of shifts were staffed by a newcomer nurse who had not previously cared for the index patient. The patterns of newcomers to teams did not differ according to birth weight. When the population was dichotomized according to median mean repeat caregiver interval value, increased reports of problems with nursing care were seen with less-consistent staffing by familiar nurses. This relationship persisted after controlling for factors including birth weight, length of stay, and team size. CONCLUSIONS: Family perceptions of nursing care quality are more strongly associated with team structure and the sequence of nursing participation than with team size. Objective measures of health care team structure and function can be examined by applying network analytic techniques to information contained in electronic health records.