Machine learning for psychiatric patient triaging: an investigation of cascading classifiers.

Objective: Develop an approach, One-class-at-a-time, for triaging psychiatric patients using machine learning on textual patient records. Our approach aims to automate the triaging process and reduce expert effort while providing high classification reliability. Materials and Methods: The One-class-at-a-time approach is a multistage cascading classification technique that achieves higher triage classification accuracy compared to traditional multiclass classifiers through 1) classifying one class at a time (or stage), and 2) identification and application of the highest accuracy classifier at each stage. The approach was evaluated using a unique dataset of 433 psychiatric patient records with a triage class label provided by "I2B2 challenge," a recent competition in the medical informatics community. Results: The One-class-at-a-time cascading classifier outperformed state-of-the-art classification techniques with overall classification accuracy of 77% among 4 classes, exceeding accuracies of existing multiclass classifiers. The approach also enabled highly accurate classification of individual classes-the severe and mild with 85% accuracy, moderate with 64% accuracy, and absent with 60% accuracy. Discussion: The triaging of psychiatric cases is a challenging problem due to the lack of clear guidelines and protocols. Our work presents a machine learning approach using psychiatric records for triaging patients based on their severity condition. Conclusion: The One-class-at-a-time cascading classifier can be used as a decision aid to reduce triaging effort of physicians and nurses, while providing a unique opportunity to involve experts at each stage to reduce false positive and further improve the system's accuracy.

CE: Original Research: Patient Handling and Mobility Course Content: A National Survey of Nursing Programs.

: Purpose: Despite the evidence supporting safe patient handling and mobility (SPHM) practices, anecdotal evidence suggests that such practices are not universally taught in academic nursing programs. The primary goal of this cross-sectional descriptive study was to understand what nursing programs teach students about lifting, turning, transferring, repositioning, and mobilizing patients. METHODS: Faculty from baccalaureate and associate's degree nursing programs in the United States were invited via e-mail to complete a 64-item survey questionnaire, which was accessible through an online link. Participants were also invited to send documents related to SPHM course content to the research team. RESULTS: Faculty from 228 baccalaureate and associate's degree nursing programs completed the questionnaire. Most curricula included outdated manual techniques, taught reliance on body mechanics to reduce the risk of musculoskeletal injuries, and made use of nonergonomic aids such as draw sheets. Elements of SPHM in the curricula were less common, and nearly half of the respondents didn't know whether their affiliated clinical facilities had an SPHM program. CONCLUSIONS: The survey results suggest many possibilities for improvement-such as partnering with faculty in physical and occupational therapy departments, clinical partnering, and working with equipment vendors-to better incorporate evidence-based SPHM principles and practices into nursing curricula.

Patient Health Record Systems Scope and Functionalities: Literature Review and Future Directions.

BACKGROUND: A new generation of user-centric information systems is emerging in health care as patient health record (PHR) systems. These systems create a platform supporting the new vision of health services that empowers patients and enables patient-provider communication, with the goal of improving health outcomes and reducing costs. This evolution has generated new sets of data and capabilities, providing opportunities and challenges at the user, system, and industry levels. OBJECTIVE: The objective of our study was to assess PHR data types and functionalities through a review of the literature to inform the health care informatics community, and to provide recommendations for PHR design, research, and practice. METHODS: We conducted a review of the literature to assess PHR data types and functionalities. We searched PubMed, Embase, and MEDLINE databases from 1966 to 2015 for studies of PHRs, resulting in 1822 articles, from which we selected a total of 106 articles for a detailed review of PHR data content. RESULTS: We present several key findings related to the scope and functionalities in PHR systems. We also present a functional taxonomy and chronological analysis of PHR data types and functionalities, to improve understanding and provide insights for future directions. Functional taxonomy analysis of the extracted data revealed the presence of new PHR data sources such as tracking devices and data types such as time-series data. Chronological data analysis showed an evolution of PHR system functionalities over time, from simple data access to data modification and, more recently, automated assessment, prediction, and recommendation. CONCLUSIONS: Efforts are needed to improve (1) PHR data quality through patient-centered user interface design and standardized patient-generated data guidelines, (2) data integrity through consolidation of various types and sources, (3) PHR functionality through application of new data analytics methods, and (4) metrics to evaluate clinical outcomes associated with automated PHR system use, and costs associated with PHR data storage and analytics.

Factors associated with lumbar fusion surgery: a case-control study.

PURPOSE: The objective of this study is to identify the demographic and payer factors that are associated with lumbar fusion surgery. METHODS: A case-control study was conducted utilizing a population of 38,092 patients from the 2010 Florida Agency for Health Care Administration (AHCA), USA hospital discharge data. The case population included 16,236 records with any of five ICD-9-CM principal procedure codes for initial lumbar fusion. The control group was comprised of 21,856 patients who were admitted for the same principal diagnoses as the cases, but who did not have initial fusion surgery. Logistic regression was used to analyze the association of age, gender, race and principal payer type with initial lumbar fusions. The interaction between age and payer was also examined, as payer type may moderate the association between age and lumbar fusion surgery. RESULTS: Gender, race, principal payer and age were all found to be significantly associated with lumbar fusion surgery. The interaction of payer and age was also found to be significant. Being female was significantly associated with having a fusion (OR = 1.11, 95 % CI 1.07-1.16). The association between age and receiving surgery was greatest for the less than 20 age group (OR = 10.43, 95 % CI 8.74-12.45). Employees and dependents of Federal government agencies (Tricare, etc.) and patients with commercial insurance were significantly associated with surgery (OR = 1.48, 95 % CI 1.29-1.70 and OR = 1.12, 95 % CI 1.04-1.20, respectively). Patients insured through Medicaid (a social health care program for those with low incomes and limited resources), and the uninsured were negatively associated with surgery (OR = 0.53, 95 % CI 0.47-0.60 and OR = 0.52, 95 % CI 0.46-0.58, respectively). CONCLUSIONS: Lumbar fusion surgery is not recommended in clinical practice guidelines for the top four principal diagnoses in this study. Yet, patients covered by certain types of insurance were found to be significantly associated with fusion surgery.

Predictors of the charges for lumbar fusion surgery in Florida hospitals.

STUDY DESIGN: A mixed-effects model was used to evaluate the effects specific surgical procedure by International Classification of Diseases, Ninth Revision, Clinical Modification, procedure code, patient age, sex, ethnic group, payers for the inpatient hospital stay, and number of additional diagnoses beyond the principal diagnosis that led to the procedure (as a proxy for severity of illness) on the charges for lumbar fusion surgery. OBJECTIVE: The present research examined the charges and the predictors of the charges for lumbar fusion surgery in Florida hospitals in 2010. SUMMARY OF BACKGROUND DATA: The number of spinal fusion surgical procedures in the United States has grown exponentially in recent years despite the procedure's high costs and questionable efficacy for many of the principal diagnoses associated with it. METHODS: All records with any of the 5 International Classification of Diseases, Ninth Revision, Clinical Modification, principal procedure codes for lumbar fusion were extracted (cases) from the Florida Agency for Health Care Administration (AHCA) hospital discharge data for the year 2010. A control group was obtained by taking all patients who had the same principal diagnoses as the cases, but who did not have fusion surgery. This produced 16,236 cases and 21,856 controls. RESULTS: The total hospital charges for lumbar fusion surgery in Florida in 2010 were $2,095,413,584. Despite having the same principal diagnoses and a similar number of additional diagnoses, patients who underwent a fusion surgery had 3 times the charges as those incurred by the controls. The number of additional diagnoses, sex, age, payer, and principal procedure, were all found to be statistically significant predictors of charges. Ethnicity was not significant. Of all the predictors, the number of additional diagnoses was the most significant in the model (F=2577, P<0.0001). CONCLUSION: The high incidence and charges for fusion surgical procedures shown in this study emphasize the need for a better understanding of when these surgical procedures are justified and for which patients. LEVEL OF EVIDENCE: N/A.