BackgroundHeart failure is one of the leading causes of hospitalization in the United States. Advances in big data solutions allow for storage, management, and mining of large volumes of structured and semi-structured data, such as complex healthcare data. Applying these advances to complex healthcare data has led to the development of risk prediction models to help identify patients who would benefit most from disease management programs in an effort to reduce readmissions and healthcare cost, but the results of these efforts have been varied. The primary aim of this study was to develop a 30-day readmission risk prediction model for heart failure patients discharged from a hospital admission.MethodsWe used longitudinal electronic medical record data of heart failure patients admitted within a large healthcare system. Feature vectors included structured demographic, utilization, and clinical data, as well as selected extracts of un-structured data from clinician-authored notes. The risk prediction model was developed using deep unified networks (DUNs), a new mesh-like network structure of deep learning designed to avoid over-fitting. The model was validated with 10-fold cross-validation and results compared to models based on logistic regression, gradient boosting, and maxout networks. Overall model performance was assessed using concordance statistic. We also selected a discrimination threshold based on maximum projected cost saving to the Partners Healthcare system.ResultsData from 11,510 patients with 27,334 admissions and 6369 30-day readmissions were used to train the model. After data processing, the final model included 3512 variables. The DUNs model had the best performance after 10-fold cross-validation. AUCs for prediction models were 0.664 ± 0.015, 0.650 ± 0.011, 0.695 ± 0.016 and 0.705 ± 0.015 for logistic regression, gradient boosting, maxout networks, and DUNs respectively. The DUNs model had an accuracy of 76.4% at the classification threshold that corresponded with maximum cost saving to the hospital.ConclusionsDeep learning techniques performed better than other traditional techniques in developing this EMR-based prediction model for 30-day readmissions in heart failure patients. Such models can be used to identify heart failure patients with impending hospitalization, enabling care teams to target interventions at their most high-risk patients and improving overall clinical outcomes.Electronic supplementary materialThe online version of this article (10.1186/s12911-018-0620-z) contains supplementary material, which is available to authorized users.

Many children with autism spectrum disorder (ASD) have notable difficulties in motor, speech and language domains. The connection between motor skills (oral-motor, manual-motor) and speech and language deficits reported in other developmental disorders raises important questions about a potential relationship between motor skills and speech-language deficits in ASD. To this end, we examined data from children with ASD (n = 1781), 2-17 years of age, enrolled in the Autism Speaks-Autism Treatment Network (AS-ATN) registry who completed a multidisciplinary evaluation that included diagnostic, physical, cognitive and behavioral assessments as part of a routine standard of care protocol. After adjusting for age, non-verbal IQ, Attention Deficit Hyperactivity Disorder (ADHD) medication use, and muscle tone, separate multiple linear regression analyses revealed significant positive associations of fine motor skills (FM) with both expressive language (EL) and receptive language (RL) skills in an impaired FM subgroup; in contrast, the impaired gross motor (GM) subgroup showed no association with EL but a significant negative association with RL. Similar analyses between motor skills and interpersonal relationships across the sample found both GM skills and FM skills to be associated with social interactions. These results suggest potential differences in the contributions of fine versus gross motor skills to autistic profiles and may provide another lens with which to view communication differences across the autism spectrum for use in treatment interventions.

BackgroundPersonal Emergency Response Systems (PERS) are traditionally used as fall alert systems for older adults, a population that contributes an overwhelming proportion of healthcare costs in the United States. Previous studies focused mainly on qualitative evaluations of PERS without a longitudinal quantitative evaluation of healthcare utilization in users. To address this gap and better understand the needs of older patients on PERS, we analyzed longitudinal healthcare utilization trends in patients using PERS through the home care management service of a large healthcare organization.MethodsRetrospective, longitudinal analyses of healthcare and PERS utilization records of older patients over a 5-years period from 2011–2015. The primary outcome was to characterize the healthcare utilization of PERS patients. This outcome was assessed by 30-, 90-, and 180-day readmission rates, frequency of principal admitting diagnoses, and prevalence of conditions leading to potentially avoidable admissions based on Centers for Medicare and Medicaid Services classification criteria.ResultsThe overall 30-day readmission rate was 14.2%, 90-days readmission rate was 34.4%, and 180-days readmission rate was 42.2%. While 30-day readmission rates did not increase significantly (p = 0.16) over the study period, 90-days (p = 0.03) and 180-days (p = 0.04) readmission rates did increase significantly. The top 5 most frequent principal diagnoses for inpatient admissions included congestive heart failure (5.7%), chronic obstructive pulmonary disease (4.6%), dysrhythmias (4.3%), septicemia (4.1%), and pneumonia (4.1%). Additionally, 21% of all admissions were due to conditions leading to potentially avoidable admissions in either institutional or non-institutional settings (16% in institutional settings only).ConclusionsChronic medical conditions account for the majority of healthcare utilization in older patients using PERS. Results suggest that PERS data combined with electronic medical records data can provide useful insights that can be used to improve health outcomes in older patients.