Protecting endangered megafauna through AI analysis of drone images in a low-connectivity setting: a case study from Namibia.

Assessing the numbers and distribution of at-risk megafauna such as the black rhino (Diceros bicornis) is key to effective conservation, yet such data are difficult to obtain. Many current monitoring technologies are invasive to the target animals and expensive. Satellite monitoring is emerging as a potential tool for very large animals (e.g., elephant) but detecting smaller species requires higher resolution imaging. Drones can deliver the required resolution and speed of monitoring, but challenges remain in delivering automated monitoring systems where internet connectivity is unreliable or absent. This study describes a model built to run on a drone to identify in situ images of megafauna. Compared with previously reported studies, this automated detection framework has a lower hardware cost and can function with a reduced internet bandwidth requirement for local network communication. It proposes the use of a Jetson Xavier NX, onboard a Parrot Anafi drone, connected to the internet throughout the flight to deliver a lightweight web-based notification system upon detection of the target species. The GPS location with the detected target species images is sent using MQ Telemetry Transport (MQTT), a lightweight messaging protocol using a publisher/subscriber architecture for IoT devices. It provides reliable message delivery when internet connection is sporadic. We used a YOLOv5l6 object detection architecture trained to identify a bounding box for one of five objects of interest in a frame of video. At an intersection over union (IoU) threshold of 0.5, our model achieved an average precision (AP) of 0.81 for black rhino (our primary target) and 0.83 for giraffe (Giraffa giraffa). The model was less successful at identifying the other smaller objects which were not our primary targets: 0.34, 0.25, and 0.42 for ostrich (Struthio camelus australis), springbok (Antidorcas marsupialis) and human respectively. We used several techniques to optimize performance and overcome the inherent challenge of small objects (animals) in the data. Although our primary focus for the development of the model was rhino, we included other species classes to emulate field conditions where many animal species are encountered, and thus reduce the false positive occurrence rate for rhino detections. To constrain model overfitting, we trained the model on a dataset with varied terrain, angle and lighting conditions and used data augmentation techniques (i.e., GANs). We used image tiling and a relatively larger (i.e., higher resolution) image input size to compensate for the difficulty faced in detecting small objects when using YOLO. In this study, we demonstrated the potential of a drone-based AI pipeline model to automate the detection of free-ranging megafauna detection in a remote setting and create alerts to a wildlife manager in a relatively poorly connected field environment.

Modeling Emergency Department crowding: Restoring the balance between demand for and supply of emergency medicine.

Emergency Departments (EDs) worldwide are confronted with rising patient volumes causing significant strains on both Emergency Medicine and entire healthcare systems. Consequently, many EDs are in a situation where the number of patients in the ED is temporarily beyond the capacity for which the ED is designed and resourced to manage-a phenomenon called Emergency Department (ED) crowding. ED crowding can impair the quality of care delivered to patients and lead to longer patient waiting times for ED doctor's consult (time to provider) and admission to the hospital ward. In Singapore, total ED attendance at public hospitals has grown significantly, that is, roughly 5.57% per year between 2005 and 2016 and, therefore, emergency physicians have to cope with patient volumes above the safe workload. The purpose of this study is to create a virtual ED that closely maps the processes of a hospital-based ED in Singapore using system dynamics, that is, a computer simulation method, in order to visualize, simulate, and improve patient flows within the ED. Based on the simulation model (virtual ED), we analyze four policies: (i) co-location of primary care services within the ED, (ii) increase in the capacity of doctors, (iii) a more efficient patient transfer to inpatient hospital wards, and (iv) a combination of policies (i) to (iii). Among the tested policies, the co-location of primary care services has the largest impact on patients' average length of stay (ALOS) in the ED. This implies that decanting non-emergency lower acuity patients from the ED to an adjacent primary care clinic significantly relieves the burden on ED operations. Generally, in Singapore, there is a tendency to strengthen primary care and to educate patients to see their general practitioners first in case of non-life threatening, acute illness.

Emergency department 'outbreak rostering' to meet challenges of COVID-19.

The COVID-19 outbreak has posed unique challenges to the emergency department rostering. Additional infection control, the possibility of quarantine of staff and minimising contact among staff have significant impact on the work of doctors in the emergency department. Infection of a single healthcare worker may require quarantine of close contacts at work. This may thus affect a potentially large number of staff. As such, we developed an Outbreak Response Roster. This Outbreak Response Roster had fixed teams of doctors working in rotation, each team that staff the emergency department in turn. Members within teams remained constant and were near equally balanced in terms of manpower and seniority of doctors. Each team worked fixed 12 hours shifts with as no overlapping of staff or staggering of shifts. Handovers between shifts were kept as brief as possible. All these were measures to limit interactions among healthcare workers. With the implementation of the roster, measures were also taken to bolster the psychological wellness of healthcare workers. With face-to-face contact limited, we also had to maintain clear, open channels for communication through technology and continue educating residents through innovative means.

Improving the wait time to admission by reducing bed rejections.

Congestion at the emergency department (ED) is associated with increased wait times, morbidity and mortality. We have identified prolonged wait time to admission as a significant contributor to ED congestion. One of the main contributors to prolonged wait time to admission was due to rejections by ward staff for the beds allocated to newly admitted patients by the Bed Management Unit (BMU). We have identified this as a systemic issue and through this quality improvement effort, seek to reduce the incidence of bed rejections for all admitted patients by 50% from 9% to 4.5% within 6 months. We used PDSA (Plan, Do, Study, Act) cycles to implement a series of interventions, such as updating legacy categorisation of wards, instituting a 'no rejects' policy and performing ward level audits. Compared with baseline, there was reduction in rejected BMU allocation requests from 9% to 5% (p<0.01). The monthly percentage of patients with at least one rejection dropped from an average of 7% to 4% (p<0.01). With reduction in the number of rejections, the average wait time to the final request acknowledged by the ward for all admission sources decreased from 2 hours 19 min to 1 hour (p<0.01), thereby allowing the overall wait time to admission to decrease by 68 min, from 5 hours 13 min to 4 hours 5 min. Improvements in the absolute duration and variance of wait times were sustained. Although the team's initial impetus was to improve ED wait times, this hospital-wide effort improved wait times across all admission sources. There has been a resultant increase in ownership of the admissions process by both nursing and BMU staff. With the conclusion of this effort, we are looking to further reduce the wait time to admission by optimising the current bed allocation logic through another quality improvement effort.

Can we understand population healthcare needs using electronic medical records?

INTRODUCTION: The identification of population-level healthcare needs using hospital electronic medical records (EMRs) is a promising approach for the evaluation and development of tailored healthcare services. Population segmentation based on healthcare needs may be possible using information on health and social service needs from EMRs. However, it is currently unknown if EMRs from restructured hospitals in Singapore provide information of sufficient quality for this purpose. We compared the inter-rater reliability between a population segment that was assigned prospectively and one that was assigned retrospectively based on EMR review. METHODS: 200 non-critical patients aged ≥ 55 years were prospectively evaluated by clinicians for their healthcare needs in the emergency department at Singapore General Hospital, Singapore. Trained clinician raters with no prior knowledge of these patients subsequently accessed the EMR up to the prospective rating date. A similar healthcare needs evaluation was conducted using the EMR. The inter-rater reliability between the two rating sets was evaluated using Cohen's Kappa and the incidence of missing information was tabulated. RESULTS: The inter-rater reliability for the medical 'global impression' rating was 0.37 for doctors and 0.35 for nurses. The inter-rater reliability for the same variable, retrospectively rated by two doctors, was 0.75. Variables with a higher incidence of missing EMR information such as 'social support in case of need' and 'patient activation' had poorer inter-rater reliability. CONCLUSION: Pre-existing EMR systems may not capture sufficient information for reliable determination of healthcare needs. Thus, we should consider integrating policy-relevant healthcare need variables into EMRs.

Predicting hospital admission at the emergency department triage: A novel prediction model.

BACKGROUND: Emergency department (ED) overcrowding is a growing international patient safety issue. A major contributor to overcrowding is long wait times for inpatient hospital admission. The objective of this study is to create a model that can predict a patient's need for hospital admission at the time of triage. METHODS: Retrospective observational study of electronic clinical records of all ED visits over ten years to a large urban hospital in Singapore. The data was randomly divided into a derivation set and a validation set. We used the derivation set to develop a logistic regression model that predicts probability of hospital admission for patients presenting to the ED. We tested the model on the validation set and evaluated the performance with receiver operating characteristic (ROC) curve analysis. RESULTS: A total of 1,232,016 visits were included for final analysis, of which 38.7% were admitted. Eight variables were included in the final model: age group, race, postal code, day of week, time of day, triage category, mode of arrival, and fever status. The model performed well on the validation set with an area under the curve of 0.825 (95% CI 0.824-0.827). Increasing age, increasing triage acuity, and mode of arrival via private patient transport were most predictive of the need for admission. CONCLUSIONS: We developed a model that accurately predicts admission for patients presenting to the ED using demographic, administrative, and clinical data routinely collected at triage. Implementation of the model into the electronic health record could help reduce the burden of overcrowding.

Improving the wait time to consultation at the emergency department.

Prolonged wait times at the emergency department (ED) are associated with increased morbidity and mortality, and decreased patient satisfaction. Reducing wait times at the ED is challenging. The objective of this study is to determine if the implementation of a series of interventions would help decrease the wait time to consultation (WTC) for patients at the ED within 6 months. Interventions include creation of a common board detailing work output, matching manpower to patient arrivals and adopting a team-based model of care. A retrospective analysis of the period from January 2015 to May 2016 was undertaken to define baseline duration for WTC. Rapid PDSA (Plan, Do, Study, Act) cycles were used to implement a series of interventions, and changes in wait time were tracked, with concurrent patient load, rostered manpower and number of admissions from ED. Results of the interventions were tracked from 1 October 2016 to 30 April 2017. There was improvement in WTC within 6 months of initiation of interventions. The improvements demonstrated appeared consistent and sustained. The average 95th centile WTC decreased by 38 min to 124 min, from the baseline duration of 162 min. The median WTC improved to 21 min, compared with a baseline timing of 24 min. The improvements occurred despite greater patient load of 4317 patients per month, compared with baseline monthly average of 4053 patients. There was no increase in admissions from ED and no change in the amount of ED manpower over the same period. We demonstrate how implementation of low-cost interventions, enabling transparency, equitable workload and use of a team-based care model can help to bring down wait times for patients. Quality improvement efforts were sustained by employing a data-driven approach, support from senior clinicians and providing constant feedback on outcomes.

Association between the elderly frequent attender to the emergency department and 30-day mortality: A retrospective study over 10 years.

BACKGROUND: To determine if elderly frequent attenders are associated with increased 30-day mortality, assess resource utilization by the elderly frequent attenders and identify associated characteristics that contribute to mortality. METHODS: Retrospective observational study of electronic clinical records of all emergency department (ED) visits over a 10-year period to an urban tertiary general hospital in Singapore. Patients aged 65 years and older, with 3 or more visits within a calendar year were identified. Outcomes measured include 30-day mortality, admission rate, admission diagnosis and duration spent at ED. Chi-square-tests were used to assess categorical factors and Student t-test was used to assess continuous variables on their association with being a frequent attender. Univariate and multivariate logistic regressions were conducted on all significant independent factors on to the outcome variable (30-day mortality), to determine factor independent odds ratios of being a frequent attender. RESULTS: 1.381 million attendance records were analyzed. Elderly patients accounted for 25.5% of all attendances, of which 31.3% are frequent attenders. Their 30-day mortality rate increased from 4.0% in the first visit, to 8.8% in the third visit, peaking at 10.2% in the sixth visit. Factors associated with mortality include patients with neoplasms, ambulance utilization, male gender and having attended the ED the previous year. CONCLUSION: Elderly attenders have a higher 30-day mortality risk compared to the overall ED population, with mortality risk more marked for frequent attenders. This study illustrates the importance and need for interventions to address frequent ED visits by the elderly, especially in an aging society.

Data-Driven Approach to Defining the Emergency Department Frequent Attender Using a Cohort of 10 Years.

AIMS: To identify, based on the measure of resource utilization, the number of visits per calendar year that defines the emergency department (ED) frequent attender; and examine for significant trends in patient characteristics and outcomes which may support the use of our definition. MATERIALS AND METHODS: We conducted a retrospective observational study of electronic clinical records of all ED visits over a 10-year period from January 2005 to December 2014 to an urban tertiary general hospital. We defined the ED frequent attender based on the number of ED attendances per calendar year which would yield a patient group representing more than 20% of all patient visits. Chi-square tests were conducted on each categorical factor individually to assess if they were independent of time, and the Student's t-test was used to assess continuous variables on their association with being a frequent attender. RESULTS: 1.381 million attendance records were analyzed. Patients who attended three or more times per year accounted for about 22.1% of all attendances and were defined as frequent attenders. They were associated with higher triage acuity, complex chronic illnesses, greater 30-day mortality for patients with three to six visits, and increased markers of resource utilization, such as ambulance use (15.5% vs. 11.6%), time to disposition (180 vs. 155 minutes), admissions rate (47.4% vs. 30.7%) and inpatient length of stay (6 days vs. 4 days). All p values were statistically significant (p < 0.001). CONCLUSION: We have demonstrated a data-driven approach to defining an ED frequent attender. Frequent attenders are associated with increased resource utilization, more complex illness and may be associated with greater 30-day mortality rates.

Perforated Duodenal Diverticulum with Subtle Pneumoretroperitoneum on Abdominal X-Ray.

Abdominal pain is one of the most common presenting complaints at the Emergency Department (ED). Given the myriad of possible differential diagnoses for abdominal pain, it becomes more important to diagnose conditions requiring emergent surgical intervention early. We present a case of an elderly male patient with abdominal pain secondary to perforated hollow viscus, subtle evidence of pneumoretroperitoneum on the initial supine abdominal X-ray, and review the signs of pneumoperitoneum and pneumoretroperitoneum on plain abdominal X-rays.