Optimization models for patient and technician scheduling in hemodialysis centers.

Patient and technician scheduling problem in hemodialysis centers presents a unique setting in healthcare operations as (1) unlike other healthcare problems, dialysis appointments have a steady state and the treatment times are determined in advance of the appointments, and (2) once the appointments are set, technicians will have to be assigned to two types of jobs per appointment: putting on and taking off patients (connecting to and disconnecting from dialysis machines). In this study, we design a mixed-integer programming model to minimize technicians' operating costs (regular and overtime costs) at large-scale hemodialysis centers. As this formulation proves to be computationally challenging to solve, we propose a novel reformulation of the problem as a discrete-time assignment model and prove that the two formulations are equivalent under a specific condition. We then simulate instances based on the data from our collaborating hemodialysis center to evaluate the performance of our proposed formulations. We compare our results to the current scheduling policy at the center. In our numerical analysis, we reduced the technician operating costs by 17% on average (up to 49%) compared to the current practice. We further conduct a post-optimality analysis and develop a predictive model that can estimate the number of required technicians based on the center's attributes and patients' input variables. Our predictive model reveals that the optimal number of technicians is strongly related to the time flexibility of patients and their dialysis times. Our findings can help clinic managers at hemodialysis centers to accurately estimate the technician requirements.

Machine learning for stone artifact identification: Distinguishing worked stone artifacts from natural clasts using deep neural networks.

Stone artifacts are often the most abundant class of objects found in archaeological sites but their consistent identification is limited by the number of experienced analysts available. We report a machine learning based technology for stone artifact identification as part of a solution to the lack of such experts directed at distinguishing worked stone objects from naturally occurring lithic clasts. Three case study locations from Egypt, Australia, and New Zealand provide a data set of 6769 2D images, 3868 flaked artifact and 2901 rock images used to train and test a machine learning model based on an openly available PyTorch implementation of Faster R-CNN ResNet 50. Results indicate 100% agreement between the model and original human derived classifications, a better performance than the results achieved independently by two human analysts who reassessed the 2D images available to the machine learning model. Machine learning neural networks provide the potential to consistently assess the composition of large archaeological assemblages composed of objects modified in a variety of ways.

Coastal Image Classification and Pattern Recognition: Tairua Beach, New Zealand.

The study of coastal processes is critical for the protection and development of beach amenities, infrastructure, and properties. Many studies of beach evolution rely on data collected using remote sensing and show that beach evolution can be characterized by a finite number of "beach states". However, due to practical constraints, long-term data displaying all beach states are rare. Additionally, when the dataset is available, the accuracy of the classification is not entirely objective since it depends on the operator. To address this problem, we collected hourly coastal images and corresponding tidal data for more than 20 years (November 1998-August 2019). We classified the images into eight categories according to the classic beach state classification, defined as (1) reflective, (2) incident scaled bar, (3) non-rhythmic, attached bar, (4) attached rhythmic bar, (5) offshore rhythmic bar, (6) non-rhythmic, 3-D bar, (7) infragravity scaled 2-D bar, (8) dissipative. We developed a classification model based on convolutional neural networks (CNN). After image pre-processing with data enhancement, we compared different CNN models. The improved ResNext obtained the best and most stable classification with F1-score of 90.41% and good generalization ability. The classification results of the whole dataset were transformed into time series data. MDLats algorithms were used to find frequent temporal patterns in morphology changes. Combining the pattern of coastal morphology change and the corresponding tidal data, we also analyzed the characteristics of beach morphology and the changes in morphodynamic states.

Blind testing of shoreline evolution models.

Beaches around the world continuously adjust to daily and seasonal changes in wave and tide conditions, which are themselves changing over longer time-scales. Different approaches to predict multi-year shoreline evolution have been implemented; however, robust and reliable predictions of shoreline evolution are still problematic even in short-term scenarios (shorter than decadal). Here we show results of a modelling competition, where 19 numerical models (a mix of established shoreline models and machine learning techniques) were tested using data collected for Tairua beach, New Zealand with 18 years of daily averaged alongshore shoreline position and beach rotation (orientation) data obtained from a camera system. In general, traditional shoreline models and machine learning techniques were able to reproduce shoreline changes during the calibration period (1999-2014) for normal conditions but some of the model struggled to predict extreme and fast oscillations. During the forecast period (unseen data, 2014-2017), both approaches showed a decrease in models' capability to predict the shoreline position. This was more evident for some of the machine learning algorithms. A model ensemble performed better than individual models and enables assessment of uncertainties in model architecture. Research-coordinated approaches (e.g., modelling competitions) can fuel advances in predictive capabilities and provide a forum for the discussion about the advantages/disadvantages of available models.

A Quality Improvement Initiative Featuring Peer-Comparison Prescribing Feedback Reduces Emergency Department Opioid Prescribing.

BACKGROUND: Opioid prescribing in the United States nearly tripled from 1999 to 2015, and opioid overdose deaths doubled in the same time frame. Emergency departments (EDs) may play a pivotal role in the opioid epidemic as a source of first-time opioid exposure; however, many prescribers are generally unaware of their prescribing behaviors relative to their peers. METHODS: All 117 ED prescribers at an urban academic medical center were provided with regular feedback on individual rates of opioid prescribing relative to their de-identified peers. To evaluate the effect of this intervention on the departmental rate of opioid prescribing, a statistical process control (SPC) chart was created to identify special cause variation, and an interrupted time series analysis was conducted to evaluate the immediate effect of the intervention and any change in the postintervention trend due to the intervention. RESULTS: The aggregate opioid prescribing rate in the preintervention period was 8.6% (95% confidence interval [CI]: 8.3%-8.9%), while the aggregate postintervention prescribing rate was 5.8% (95% CI: 5.5%-6.1%). The SPC chart revealed special cause variation in both the pre- and postintervention periods, with an overall downtrend of opioid prescribing rates across the evaluation period and flattening of rates in the final four blocks. Interrupted time series analysis demonstrated a significant immediate downward effect of the intervention and a nonsignificant additional decrease in postintervention trend. CONCLUSION: Implementation of peer-comparison opioid prescribing feedback was associated with a significant immediate reduction in the rate of ED discharge opioid prescribing.

Towards validating stent induced micro flow patterns in left main coronary artery bifurcations.

We investigated if blood flow changes induced through the presence of a stent could be detected using in vitro dynamically scaled 4D Phase-Contrast Magnetic Resonance Imaging (PC-MRI). Using idealized and patient-specific left main coronary artery bifurcations, we 3D-printed the dynamically large scaled geometries and incorporated them into a flow circuit for non-invasive acquisition with a higher effective spatial resolution. We tested the effects of using non-Newtonian and Newtonian fluids for the experiment. We also numerically simulated the same geometries in true scale for comparison using computational fluid dynamics (CFD). We found that the experimental setup increased the effective spatial resolution enough to reveal stent induced blood flow changes close to the vessel wall. Non-Newtonian fluid replicated all of the flow field well with a strong agreement with the computed flow field (R2 > 0.9). Fine flow structures were not as prominent for the Newtonian compared to non-Newtonian fluid consideration. In the patient-specific geometry, arterial non-planarity increased the difficulty to capture the near wall slow velocity changes. Findings demonstrate the potential to dynamically scale in vitro 4D MRI flow acquisition for micro blood flow considerations.