A Neurosurgical Readmissions Reduction Program in an Academic Hospital Leveraging Machine Learning, Workflow Analysis, and Simulation.

BACKGROUND: Predicting 30-day hospital readmissions is crucial for improving patient outcomes, optimizing resource allocation, and achieving financial savings. Existing studies reporting the development of machine learning (ML) models predictive of neurosurgical readmissions do not report factors related to clinical implementation. OBJECTIVES: Train individual predictive models with good performance (area under the receiver operating characteristic curve or AUROC > 0.8), identify potential interventions through semi-structured interviews, and demonstrate estimated clinical and financial impact of these models. METHODS: Electronic health records were utilized with five ML methodologies: gradient boosting, decision tree, random forest, ridge logistic regression, and linear support vector machine. Variables of interest were determined by domain experts and literature. The dataset was split divided 80% for training and validation and 20% for testing randomly. Clinical workflow analysis was conducted using semi-structured interviews to identify possible intervention points. Calibrated agent-based models (ABMs), based on a previous study with interventions, were applied to simulate reductions of the 30-day readmission rate and financial costs. RESULTS: The dataset covered 12,334 neurosurgical intensive care unit (NSICU) admissions (11,029 patients); 1,903 spine surgery admissions (1,641 patients), and 2,208 traumatic brain injury (TBI) admissions (2,185 patients), with readmission rate of 13.13, 13.93, and 23.73%, respectively. The random forest model for NSICU achieved best performance with an AUROC score of 0.89, capturing potential patients effectively. Six interventions were identified through 12 semi-structured interviews targeting preoperative, inpatient stay, discharge phases, and follow-up phases. Calibrated ABMs simulated median readmission reduction rates and resulted in 13.13 to 10.12% (NSICU), 13.90 to 10.98% (spine surgery), and 23.64 to 21.20% (TBI). Approximately $1,300,614.28 in saving resulted from potential interventions. CONCLUSION: This study reports the successful development and simulation of an ML-based approach for predicting and reducing 30-day hospital readmissions in neurosurgery. The intervention shows feasibility in improving patient outcomes and reducing financial losses.

User-Centered Evaluation of a Visual Annotation Tool for Rapid Assessment of Pediatric Weight Entry Errors.

Weight entry errors can cause significant patient harm in pediatrics due to pervasive weight-based dosing practices. While computerized algorithms can assist in error detection, they have not achieved high sensitivity and specificity to be further developed as a clinical decision support tool. To train an advanced algorithm, expert-annotated weight errors are essential but difficult to collect. In this study, we developed a visual annotation tool to gather large amounts of expertly annotated pediatric weight charts and conducted a formal user-centered evaluation. Key features of the tool included configurable grid sizes and annotation styles. The user feedback was collected through a structured survey and user clicks on the interface. The results show that the visual annotation tool has high usability (average SUS=86.4). Different combinations of the key features, however, did not significantly improve the annotation efficiency and duration. We have used this tool to collect expert annotations for algorithm development and benchmarking.

A Scoping Review of Health Information Technology in Clinician Burnout.

BACKGROUND: Clinician burnout is a prevalent issue in healthcare, with detrimental implications in healthcare quality and medical costs due to errors. The inefficient use of health information technologies (HIT) is attributed to having a role in burnout. OBJECTIVE: This paper seeks to review the literature with the following two goals: (1) characterize and extract HIT trends in burnout studies over time, and (2) examine the evidence and synthesize themes of HIT's roles in burnout studies. METHODS: A scoping literature review was performed by following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines with two rounds of searches in PubMed, IEEE Xplore, ACM, and Google Scholar. The retrieved papers and their references were screened for eligibility by using developed inclusion and exclusion criteria. Data were extracted from included papers and summarized either statistically or qualitatively to demonstrate patterns. RESULTS: After narrowing down the initial 945 papers, 36 papers were included. All papers were published between 2013 and 2020; nearly half of them focused on primary care (n = 16; 44.4%). The most commonly studied variable was electronic health record (EHR) practices (e.g., number of clicks). The most common study population was physicians. HIT played multiple roles in burnout studies: it can contribute to burnout; it can be used to measure burnout; or it can intervene and mitigate burnout levels. CONCLUSION: This scoping review presents trends in HIT-centered burnout studies and synthesizes three roles for HIT in contributing to, measuring, and mitigating burnout. Four recommendations were generated accordingly for future burnout studies: (1) validate and standardize HIT burnout measures; (2) focus on EHR-based solutions to mitigate clinician burnout; (3) expand burnout studies to other specialties and types of healthcare providers, and (4) utilize mobile and tracking technology to study time efficiency.

Can the Visibility of Both Prosthetic Posterior Femoral Condyles on a Postoperative Radiograph Assure that Limb Rotation Is Appropriate to Allow Accurate Measurement of the Anatomic Knee Axis?

BACKGROUND: To better define radiographic parameters for a true anterior-posterior (AP) knee radiograph after total knee arthroplasty, we cataloged the radiographic appearance of 7 different designs of commercially available femoral components at various points of rotation to correlate the visibility of the prosthetic posterior femoral condyles (PPFCs) with the amount of rotation of the femoral component, and hence, the limb. METHODS: AP radiographs of 7 left-sided, cruciate-retaining femoral trial components were obtained at 5° increments of rotation from 20° internal rotation (IR) to 20° external rotation (ER). Rotational profiles were cataloged based on the visibility of either or both of the PPFCs. RESULTS: Three categories of femoral component rotation profiles were noted, based on the visibility of the PPFC: overt ER with only the medial PFC visible at greater than 10° ER, overt IR with only the lateral PFC visible at greater than 20° IR, and near-neutral rotation with both medial and lateral PPFCs visible between 5° ER and 15° IR. CONCLUSION: An acceptable AP radiograph to measure the anatomic knee axis after total knee arthroplasty is one where both the medial and lateral PPFCs are visible on either side of the trochlear flange.

The Effect of Patellar Thickness on Intraoperative Knee Flexion and Patellar Tracking in Patients With Arthrofibrosis Undergoing Total Knee Arthroplasty.

BACKGROUND: We evaluated the intraoperative effect of patellar thickness on intraoperative passive knee flexion and patellar tracking during total knee arthroplasty (TKA) in patients with preoperative arthrofibrosis and compared them to patients with normal preoperative range of motion (ROM) documented in a prior study. METHODS: Routine posterior cruciate ligament-retaining TKA was performed in a total of 34 knees, 23 with normal ROM and 11 with arthrofibrosis, defined as ≤100° of passive knee flexion against gravity under anesthesia. Once clinical balance and congruent patellar tracking were established, custom trial patellar components thicker than the standard trial by 2-mm increments (2-8 mm) were sequentially placed and trialed. Passive flexion against gravity was recorded using digital photograph goniometry. Gross mechanics of patellofemoral tracking were visually assessed. RESULTS: On average, passive knee flexion decreased 2° for every 2-mm increment of patellar thickness (P < .0001), which was similar to patients with normal preoperative ROM. In addition, increased patellar thickness had no gross effect on patellar subluxation and tilt in patients with arthrofibrosis as well as those with normal ROM. CONCLUSIONS: Patellar thickness had a modest effect on intraoperative passive flexion and no effect on patellar tracking in patients with arthrofibrosis undergoing TKA. There was no marked difference in intraoperative flexion and patellar tracking between patients with arthrofibrosis and patients with normal preoperative ROM.

Prevalence of thoracic scoliosis in adults 25 to 64 years of age detected during routine chest radiographs.

PURPOSE: To investigate the prevalence of thoracic scoliosis and determine the effect of both age and gender on coronal curve magnitude among asymptomatic adults aged 25-64 years old, using standing posterior-anterior chest radiographs. METHODS: This was a retrospective, cross-sectional study evaluating 500 randomly selected digital posterior-anterior chest radiographs taken at a single institution on an outpatient basis between January 2010 and December 2011. Males (n = 184) and females (n = 316) ranged in age from 25 to 64 years. Patients with symptoms of back pain; including a history of back pain, spinal instrumentation, or known pre-existing spinal disease were excluded. Radiographs were evaluated using Centricity PACS Web Diagnostic 2.1 system (General Electric Co. Fairfield, CT). Coronal Cobb angle measurements of the thoracic spine were quantified by the authors, with scoliosis defined as coronal curves greater than 10°. Curvatures were subdivided into groups: a control group with coronal curves less than 10°, curves measuring 10° to 19°, 20° to 29°, and greater than 30°. The effect of age and gender on curve magnitude was examined using Pearson correlation analysis and linear regression analysis. RESULTS: There was a 13.4 % (67 patients) prevalence of thoracic scoliosis. The prevalence among asymptomatic males was 10.9 %, while the prevalence among asymptomatic females was 14.9 %. 11.6 % demonstrated a coronal curvature between 10° and 19° (58 patients), 1.6 % between 20° and 29° (8 patients), and 0.2 % greater than 30° (1 patient). Age and gender were not found to be significant independent predictors of curve severity. CONCLUSIONS: We found a 13.4 % prevalence of thoracic scoliosis among asymptomatic adults aged 25-64 years on routine outpatient chest radiographs. 11.6 % of patients demonstrated a coronal curvature between 10° and 19°. Unlike prior studies evaluating asymptomatic thoracic curves in elderly patients, age and gender did not significantly affect curve magnitude in our younger cohort. These data may provide a reference point to help clinicians counsel asymptomatic patients diagnosed with thoracic scoliosis on routine chest radiographs.

Factors affecting response rates to mailed preoperative surveys among arthroplasty patients.

AIM: To identify factors that affect patient response rates to preoperative functional surveys in hip and knee arthroplasty patients. METHODS: From May 2008 to March 2009, 247 patients were scheduled more than 4 wk in advance for hip or knee arthroplasty by one of two participating surgeons at our center. A personalized questionnaire comprised of the Short Form 12 (SF-12) and Western Ontario and McMaster Universities (WOMAC) Index was mailed to patients at random time points ranging from 7 to 101 d prior to surgery. Nine independent factors were documented prospectively, including age, gender, ethnicity, marital status, type of surgery, surgeon, days prior to surgery (DPS) of survey mailing, WOMAC score and SF-12 score. The date of the completed survey receipt was also documented. For non-responders, the surveys were completed with the research team at the hospital upon admission. Multivariate regression and χ(2) analysis were performed with Statistical Analysis Software software. RESULTS: DPS was the only factor that affected patient response. Mailing surveys 26 d to 31 d prior to surgery dates led to a peak response rate of 80% that was significantly higher (P < 0.023) than response rates for patients who were mailed their surveys ≤ 16 d (62.5%), 17 d to 25 d (70%) or ≥ 32 d prior to surgery (55%). No other factors, including preoperative WOMAC and SF-12 scores, significantly influenced response behavior. CONCLUSION: The DPS was independently the most significant predictor of response rates for pre-operative functional data among patients scheduled for hip and knee arthroplasty.

Formation of pancreatic islets involves coordinated expansion of small islets and fission of large interconnected islet-like structures.

The islets of Langerhans, micro-organs for maintaining glucose homeostasis, range in size from small clusters of <10 cells to large islets consisting of several thousand endocrine cells. Islet size distributions among various species are similar and independent of body size, suggesting an intrinsic limit to islet size. Little is known about the mechanisms regulating islet size. We have carried out a comprehensive analysis of changes of islet size distribution in the intact mouse pancreas from birth to eight months, including mathematical modeling to quantify this dynamic biological process. Islet growth was size-dependent during development, with preferential expansion of smaller islets and fission of large interconnected islet-like structures occurring most actively at approximately three weeks of age at the time of weaning. The process of islet formation was complete by four weeks with little or no new islet formation thereafter, and all the ß-cells had low proliferation potential in the adult, regardless of islet size. Similarly, in insulinoma-bearing mice, the early postnatal developmental process including fission followed the same time course with no new islet formation in adults. However, tumor progression led to uncontrolled islet growth with accelerated expansion of larger islets. Thus, islet formation and growth is a tightly regulated process involving preferential expansion of small islets and fission of large interconnected islet-like structures.

Computer-assisted large-scale visualization and quantification of pancreatic islet mass, size distribution and architecture.

The pancreatic islet is a unique micro-organ composed of several hormone secreting endocrine cells such as beta-cells (insulin), alpha-cells (glucagon), and delta-cells (somatostatin) that are embedded in the exocrine tissues and comprise 1-2% of the entire pancreas. There is a close correlation between body and pancreas weight. Total beta-cell mass also increases proportionately to compensate for the demand for insulin in the body. What escapes this proportionate expansion is the size distribution of islets. Large animals such as humans share similar islet size distributions with mice, suggesting that this micro-organ has a certain size limit to be functional. The inability of large animal pancreata to generate proportionately larger islets is compensated for by an increase in the number of islets and by an increase in the proportion of larger islets in their overall islet size distribution. Furthermore, islets exhibit a striking plasticity in cellular composition and architecture among different species and also within the same species under various pathophysiological conditions. In the present study, we describe novel approaches for the analysis of biological image data in order to facilitate the automation of analytic processes, which allow for the analysis of large and heterogeneous data collections in the study of such dynamic biological processes and complex structures. Such studies have been hampered due to technical difficulties of unbiased sampling and generating large-scale data sets to precisely capture the complexity of biological processes of islet biology. Here we show methods to collect unbiased "representative" data within the limited availability of samples (or to minimize the sample collection) and the standard experimental settings, and to precisely analyze the complex three-dimensional structure of the islet. Computer-assisted automation allows for the collection and analysis of large-scale data sets and also assures unbiased interpretation of the data. Furthermore, the precise quantification of islet size distribution and spatial coordinates (i.e. X, Y, Z-positions) not only leads to an accurate visualization of pancreatic islet structure and composition, but also allows us to identify patterns during development and adaptation to altering conditions through mathematical modeling. The methods developed in this study are applicable to studies of many other systems and organisms as well.

Legal ramifications for physicians of patients who drive with hepatic encephalopathy.

BACKGROUND & AIMS: Hepatic encephalopathy, a spectrum of neuropsychiatric abnormalities that can occur in patients with liver dysfunction, negatively affects driving performance, but no study has examined legal ramifications. We studied state requirements for reporting hepatic encephalopathy and investigated whether lawsuits have been completed against physicians or patients for motor vehicle accidents that were related to hepatic encephalopathy. METHODS: We contacted motor vehicle departments from all 50 states and examined motor vehicle codes and legal databases to search for hepatic encephalopathy-related lawsuits. RESULTS: Definitions of a medically impaired driver varied considerably. No state specifically mentioned hepatic encephalopathy or patients with advanced liver disease. Only 6 (12%) of the states had mandatory reporting laws for drivers who have medical impairment, and 25 of the remaining 44 states (57%) provided legal immunity to physicians for reporting such patients. The legal databases did not contain any cases against physicians for failure to warn against driving or diagnose hepatic encephalopathy that resulted in an accident. There were no lawsuits identified against an encephalopathic patient for causing a motor vehicle accident. CONCLUSIONS: Only 6 states require physicians to report drivers with medical impairments. Hepatic encephalopathy is not specifically addressed in any state vehicle code. There are no completed lawsuits against physicians or patients for motor vehicle accidents associated with driving impairment from hepatic encephalopathy. In the absence of definitive laws, the onus of responsibility for identifying potentially hazardous drivers might still lie with the physician; physicians should carefully evaluate patients for driving abilities.

Hypovitaminosis D in patients scheduled to undergo orthopaedic surgery: a single-center analysis.

BACKGROUND: Vitamin D is essential for optimal bone health and muscle function. An alarmingly high rate of vitamin-D deficiency in the general population has been reported recently. The purpose of the present study was to characterize the extent of low serum levels of vitamin D among orthopaedic surgery patients. METHODS: We performed a retrospective chart review of 723 patients who were scheduled for orthopaedic surgery between January 2007 and March 2008. Preoperative serum 25-hydroxyvitamin D (25[OH]D) levels were measured. The prevalence of normal (≥32 ng/mL), insufficient (<32 ng/mL), and deficient (<20 ng/mL) vitamin-D levels was determined. Logistic regression was used to assess risk factors for insufficient (<32 ng/mL) 25(OH)D levels. RESULTS: Overall, 43% of all patients had insufficient serum vitamin-D levels, and, of these, 40% had deficient levels. Among the orthopaedic services, the highest rates of low serum vitamin-D levels were seen in the trauma and sports services, in which the rates of abnormal (insufficient and deficient) vitamin-D levels were 66% and 52%, respectively. The lowest rate of abnormal vitamin-D levels was seen in the metabolic bone disease service. Patients between the ages of fifty-one and seventy years were 35% less likely to have low vitamin-D levels than patients between the ages of eighteen and fifty years (p = 0.018). The prevalence of low vitamin-D levels was significantly higher in men (p = 0.006). Individuals with darker skin tones (blacks and Hispanics) were 5.5 times more likely to have low vitamin-D levels when compared with those with lighter skin tones (whites and Asians) (p < 0.001). CONCLUSIONS: The prevalence of low serum levels of vitamin D among patients undergoing orthopaedic surgery is very common. Given the importance of vitamin D in musculoskeletal health, such low levels may negatively impact patient outcomes.

Pancreatic islet plasticity: interspecies comparison of islet architecture and composition.

The pancreatic islet displays diverse patterns of endocrine cell arrangement. The prototypic islet, with insulin-secreting beta-cells forming the core surrounded by other endocrine cells in the periphery, is largely based on studies of normal rodent islets. Recent reports on large animals, including humans, show a difference in islet architecture, in which the endocrine cells are randomly distributed throughout the islet. This particular species difference has raised concerns regarding the interpretation of data based on rodent studies to humans. On the other hand, further variations have been reported in marsupials and some nonhuman primates, which possess an inverted ratio of beta-cells to other endocrine cells. This review discusses the striking plasticity of islet architecture and cellular composition among various species including changes in response to metabolic states within a single species. We propose that this plasticity reflects evolutionary acquired adaptation induced by altered physiological conditions, rather than inherent disparities between species.

Intraislet production of GLP-1 by activation of prohormone convertase 1/3 in pancreatic α-cells in mouse models of ß-cell regeneration.

The islet of Langerhans is a highly vascularized micro-organ consisting of not only ß-cells but multiple cell types such as α-, delta-, pancreatic polypeptide- and epsilon-cells that work together to regulate glucose homeostatis. We have recently proposed a new model of the neonatal islet formation in mice by a process of fission following contiguous endocrine cell proliferation in the form of branched cord-like structures in embryos and newborns. There exist large stretches of interconnected islet structures along large blood vessels in the neonatal pancreas, which, upon further development, segregate into smaller fragments (i.e., islets) that eventually become more spherical by internal proliferation as seen in the adult pancreas. α-cells span these elongated islet-like structures in the developing pancreas, which we hypothesize represent sites of fission and facilitate the eventual formation of discrete islets. The α-cells express both prohormone convertase 2 and 1/3 (PC 2 and PC 1/3, respectively), which resulted in the processing of the proglucagon precursor into glucagon-like peptide 1, thereby leading to local production of this important ß-cell growth factor. Furthermore, while α-cells in the adult basically only express PC 2, significant activation of PC 1/3 is also observed in mouse models of insulin resistance such as pregnant, ob/ ob, db/db and prediabetic NOD mice, which may be a common mechanism in proliferating ß-cells. Our study suggests an important role of α-cells for ß-cell proliferation and further for the endocrine cell network within an islet.

In situ quantification of pancreatic beta-cell mass in mice.

Tracing changes of specific cell populations in health and disease is an important goal of biomedical research. The process of monitoring pancreatic beta-cell proliferation and islet growth is particularly challenging. We have developed a method to capture the distribution of beta-cells in the intact pancreas of transgenic mice with fluorescence-tagged beta-cells with a macro written for ImageJ (rsb.info.nih.gov/ij/). Following pancreatic dissection and tissue clearing, the entire pancreas is captured as a virtual slice, after which the GFP-tagged beta-cells are examined. The analysis includes the quantification of total beta-cell area, islet number and size distribution with reference to specific parameters and locations for each islet and for small clusters of beta-cells. The entire distribution of islets can be plotted in three dimensions, and the information from the distribution on the size and shape of each islet allows a quantitative and qualitative comparison of changes in overall beta-cell area at a glance.

Targeting extracellular signal-regulated kinase (ERK) signaling has therapeutic implications for inflammatory osteolysis.

The extracellular signal-regulated kinase 1/2 (ERK) pathway, part of the mitogen-activated protein kinase (MAPK) family, is well-known for its role in cell differentiation and proliferation. In the context of osteoclastogenesis, macrophage colony stimulating factor (M-CSF) is an upstream activator of ERK signals for the survival of osteoclast precursors prior to their differentiation into multinucleated osteoclasts. In this study, we demonstrate by using both in vivo and in vitro models that the ERK signaling pathway involves an inflammatory response of various cells mediating osteolysis. Osteoblasts exhibit innate immune response by expressing M-CSF in response to lipopolysaccharide (LPS). LPS induced M-CSF expression is mediated by ERK. The inhibition of ERK signaling attenuated the inflammatory response to LPS both in vivo and in vitro. Thus, the ERK pathway may be a potentially important therapeutic target in the treatment of inflammatory osteolysis.

Patient education before hip or knee arthroplasty lowers length of stay.

From April 2006 to May 2007, 261 patients undergoing primary unilateral total hip arthroplasty or total knee arthroplasty were offered voluntary participation in a one-on-one preoperative educational program. Length of stay (LOS) and inpatient data were monitored and recorded, prospectively. Education participants enjoyed a significantly shorter LOS than nonparticipants for both total hip arthroplasty (3.1 +/- 0.8 days vs 3.9 +/- 1.4 days; P = .0001) and total knee arthroplasty (3.1 +/- 0.9 days vs 4.1 +/- 1.9 days; P = .001).

Islet formation during the neonatal development in mice.

The islet of Langerhans is a unique micro-organ within the exocrine pancreas, which is composed of insulin-secreting beta-cells, glucagon-secreting alpha-cells, somatostatin-secreting delta-cells, pancreatic polypeptide-secreting PP cells and ghrelin-secreting epsilon-cells. Islets also contain non-endocrine cell types such as endothelial cells. However, the mechanism(s) of islet formation is poorly understood due to technical difficulties in capturing this dynamic event in situ. We have developed a method to monitor beta-cell proliferation and islet formation in the intact pancreas using transgenic mice in which the beta-cells are specifically tagged with a fluorescent protein. Endocrine cells proliferate contiguously, forming branched cord-like structures in both embryos and neonates. Our study has revealed long stretches of interconnected islets located along large blood vessels in the neonatal pancreas. Alpha-cells span the elongated islet-like structures, which we hypothesize represent sites of fission and facilitate the eventual formation of discrete islets. We propose that islet formation occurs by a process of fission following contiguous endocrine cell proliferation, rather than by local aggregation or fusion of isolated beta-cells and islets. Mathematical modeling of the fission process in the neonatal islet formation is also presented.

Quantification of pancreatic islet distribution in situ in mice.

Tracing changes of specific cell populations in health and disease is an important goal of biomedical research. Precisely monitoring pancreatic beta-cell proliferation and islet growth is a challenging area of research. We have developed a method to capture the distribution of beta-cells in the intact pancreas of transgenic mice with fluorescence-tagged beta-cells with a macro written for ImageJ (rsb.info.nih.gov/ij/). Total beta-cell area and islet number and size distribution are quantified with reference to specific parameters and location for each islet and for small clusters of beta-cells. The entire distribution of islets can now be plotted in three dimensions, and the information from the distribution on the size and shape of each islet allows a quantitative and a qualitative comparison of changes in overall beta-cell area at a glance.

No mantle formation in rodent islets -- the prototype of islet revisited.

Emerging reports on human islets emphasize distinct differences from the widely accepted prototype of rodent islets, raising questions over their suitability for human studies. Here we aim at elucidating architectural differences and similarities of human versus rodent islets. The cellular composition and architecture of human and rodent islets were compared through three-dimensional (3D) reconstructions. Physiological and pathological changes were examined using islets from various mouse models such as non-obese diabetic (NOD), ob/ob, db/db mice and during pregnancy. A subpopulation of human islets is composed of clusters of alpha-cells within the central beta-cell cores, while the overall proportion of alpha-cells varies among islets. In mouse islets under normal conditions, alpha-cells are localized in the islet periphery, but they do not envelop the entire beta-cell core, so that beta-cells are exposed on the outer layer of the islet, as in most human islets. Also, an increased proportion of alpha-cells within the central core is observed in the pancreas of mouse models exhibiting increased demand for insulin. In summary, human and mouse islets share common architectural features as endocrine micro-organs. Since these may hold a key to better understanding islet plasticity, our concept of the prototypic islet should be revised.