International journal of medical informatics
-
Multicenter Study
Machine learning predicts mortality in septic patients using only routinely available ABG variables: a multi-centre evaluation.
To evaluate the application of machine learning methods, specifically Deep Neural Networks (DNN) models for intensive care (ICU) mortality prediction. The aim was to predict mortality within 96 hours after admission to mirror the clinical situation of patient evaluation after an ICU trial, which consists of 24-48 hours of ICU treatment and then "re-triage". The input variables were deliberately restricted to ABG values to maximise real-world practicability. ⋯ An LSTM-based model could help physicians with the "re-triage" and the decision to restrict treatment in patients with a poor prognosis.
-
The goal of this study was to examine the effects of medical notes (MD) in an electronic medical records (EMR) system on doctors' work practices at an Emergency Department (ED). ⋯ We suggest three guidelines for designing future EMR systems to be used in teaching hospitals. First, the design of documentation tools in EMR needs to take into account what we called "note-intensive tasks" to support the collaborative nature of medical work. Second, it should clearly define roles and responsibilities. Lastly, the system should provide a balance between flexibility and interruption to better manage the complex nature of medical work and to facilitate necessary interactions among ED staff and patients in the work environment.
-
The COVID-19 pandemic and its socio-economic impacts have disrupted our health systems and society. We sought to examine informatics and digital health strategies that supported the primary care response to COVID-19 in Australia. Specifically, the review aims to answer: how Australian primary health care responded and adapted to COVID-19, the facilitators and inhibitors of the Primary care informatics and digital health enabled COVID-19 response and virtual models of care observed in Australia. ⋯ COVID-19 has transformed Australian primary care with the rapid adaptation of digital technologies to complement "in-person" primary care with telehealth and virtual models of care. The pandemic has also highlighted several literacy, maturity/readiness, and micro, meso and macro-organisational challenges with adopting and adapting telehealth to support integrated person-centred health care. There is a need for more research into how telehealth and virtual models of care can improve the access, integration, safety, and quality of virtual primary care.
-
The impact of the ambulatory electronic health record (EHR) on physician productivity is poorly understood. Fear of productivity loss remains a major concern for practitioners and health care delivery organizations and inhibits system adoption. This study describes the changes in physician productivity after the implementation of a commercially available ambulatory EHR system in a large academic multi-specialty physician group. ⋯ Provider productivity, as measured by patient visit volume, charges, and wRVUs modestly increased for a cohort of multi-specialty providers that adopted a commercially available ambulatory EHR. The productivity gain appeared to become even more pronounced after several months of system experience. This objective data may help persuade apprehensive practitioners that EHR adoption need not harm productivity. The baseline differences in productivity metrics for the adopters and non-adopters in our study suggest that there are fundamental differences in these groups. Further characterizing these differences may help predict EHR adoption success and guide future implementation strategies.
-
Pneumonia is a common complication after stroke, causing an increased length of hospital stay and death. Therefore, the timely and accurate prediction of post-stroke pneumonia would be highly valuable in clinical practice. Previous pneumonia risk score models were often built on simple statistical methods such as logistic regression. This study aims to investigate post-stroke pneumonia prediction models using more advanced machine learning algorithms, specifically deep learning approaches. ⋯ The deep learning-based predictive model is feasible for stroke patient management and achieves the optimal performance compared to many classic machine learning methods.