Predictive Modeling of the Total Joint Replacement Surgery Risk: a Deep Learning Based Approach with Claims Data.

Total joint replacement (TJR) is one of the most commonly performed, fast-growing elective surgical procedures in the United States. Given its huge volume and cost variation, it has been regarded as one of the top opportunities to reduce health care cost by the industry. Identifying patients with a high chance of undergoing TJR surgery and engaging them for shopping is the key to success for plan sponsors. In this paper, we experimented with different machine learning algorithms and developed a novel deep learning approach to predict TJR surgery based on a large commercial claims dataset. Our results demonstrated that the performance of the gated recurrent neural network is better than other methods regardless of data representation methods (multi-hot encoding or embedding). Additional pooling mechanism can further improve the performance of deep learning models for our case.

Estimating disease burden using Internet data.

Data on disease burden are often used for assessing population health, evaluating the effectiveness of interventions, formulating health policies, and planning future resource allocation. We investigated whether Internet usage and social media data, specifically the search volume on Google, page view count on Wikipedia, and disease mentioning frequency on Twitter, correlated with the disease burden, measured by prevalence and treatment cost, for 1633 diseases over an 11-year period. We also applied least absolute shrinkage and selection operator to predict the burden of diseases. We found that Google search volume is relatively strongly correlated with the burdens for 39 of 1633 diseases, including viral hepatitis, diabetes mellitus, multiple sclerosis, and hemorrhoids. Wikipedia and Twitter data strongly correlated with the burdens of 15 and 7 diseases, respectively. However, an accurate analysis must consider each condition's characteristics, including acute/chronic nature, severity, familiarity to the public, and the presence of stigma.

Development of a Behavioral Health Stigma Measure and Application of Machine Learning for Classification.

Objective: Given the growing public health importance of measuring the change in mental health stigma over time, the goal of this study was to demonstrate the potential for using machine learning as a tool to analyze patterns of social stigma as a complement to traditional research methods. Methods: A total of 1,904 participants were recruited through Sona Systems, Ltd (Tallinn, Estonia), an experiment management system for online research, to complete a self-reported survey. The collected data were used to develop a new measure of mental (behavioral) health stigma. To build a classification predictive model of stigma, a decision tree was used as the data mining tool, wherein a set of classification rules was generated and tested for its ability to examine the prevalence of stigma. Results: A three-factor stigma model was supported and confirmed. Results indicate that the measure is content-valid and internally consistent. Performance evaluation of the machine learning-based classification algorithm revealed a sufficient inter-rater reliability with a predictive accuracy of 92.4 percent. Conclusion: This study illustrates the potential for applying machine learning to derive a data-driven understanding of the extent to which stigma is prevalent in society. It establishes a framework for the development of an index to track stigma over time and to assist healthcare decision-makers with improving the health of populations and the experience of care for patients.

Systems modeling and simulation applications for critical care medicine.

Critical care delivery is a complex, expensive, error prone, medical specialty and remains the focal point of major improvement efforts in healthcare delivery. Various modeling and simulation techniques offer unique opportunities to better understand the interactions between clinical physiology and care delivery. The novel insights gained from the systems perspective can then be used to develop and test new treatment strategies and make critical care delivery more efficient and effective. However, modeling and simulation applications in critical care remain underutilized. This article provides an overview of major computer-based simulation techniques as applied to critical care medicine. We provide three application examples of different simulation techniques, including a) pathophysiological model of acute lung injury, b) process modeling of critical care delivery, and c) an agent-based model to study interaction between pathophysiology and healthcare delivery. Finally, we identify certain challenges to, and opportunities for, future research in the area.

Medical informatics: an essential tool for health sciences research in acute care.

Medical Informatics has become an important tool in modern health care practice and research. In the present article we outline the challenges and opportunities associated with the implementation of electronic medical records (EMR) in complex environments such as intensive care units (ICU). We share our initial experience in the design, maintenance and application of a customized critical care, Microsoft SQL based, research warehouse, ICU DataMart. ICU DataMart integrates clinical and administrative data from heterogeneous sources within the EMR to support research and practice improvement in the ICUs. Examples of intelligent alarms -- "sniffers", administrative reports, decision support and clinical research applications are presented.

Development of a functional, internet-accessible department of surgery outcomes database.

The need for surgical outcomes data is increasing due to pressure from insurance companies, patients, and the need for surgeons to keep their own "report card". Current data management systems are limited by inability to stratify outcomes based on patients, surgeons, and differences in surgical technique. Surgeons along with research and informatics personnel from an academic, hospital-based Department of Surgery and a state university's Department of Information Technology formed a partnership to develop a dynamic, internet-based, clinical data warehouse. A five-component model was used: data dictionary development, web application creation, participating center education and management, statistics applications, and data interpretation. A data dictionary was developed from a list of data elements to address needs of research, quality assurance, industry, and centers of excellence. A user-friendly web interface was developed with menu-driven check boxes, multiple electronic data entry points, direct downloads from hospital billing information, and web-based patient portals. Data were collected on a Health Insurance Portability and Accountability Act-compliant server with a secure firewall. Protected health information was de-identified. Data management strategies included automated auditing, on-site training, a trouble-shooting hotline, and Institutional Review Board oversight. Real-time, daily, monthly, and quarterly data reports were generated. Fifty-eight publications and 109 abstracts have been generated from the database during its development and implementation. Seven national academic departments now use the database to track patient outcomes. The development of a robust surgical outcomes database requires a combination of clinical, informatics, and research expertise. Benefits of surgeon involvement in outcomes research include: tracking individual performance, patient safety, surgical research, legal defense, and the ability to provide accurate information to patient and payers.