Search

OBJECTIVEArtificial intelligence (AI) and machine learning are central components of today’s medical environment. The fairness of AI, i.e. the ability of AI to be free from bias, has repeatedly come into question. This study investigates the diversity of the members of academia whose scholarship poses questions about the fairness of AI.METHODSThe articles that combine the topics of fairness, artificial intelligence, and medicine were selected from Pubmed, Google Scholar, and Embase using keywords. Eligibility and data extraction from the articles were done manually and cross-checked by another author for accuracy. 375 articles were selected for further analysis, cleaned, and organized in Microsoft Excel; spatial diagrams were generated using Public Tableau. Additional graphs were generated using Matplotlib and Seaborn. The linear and logistic regressions were analyzed using Python.RESULTSWe identified 375 eligible publications, including research and review articles concerning AI and fairness in healthcare. When looking at the demographics of all authors, out of 1984, 794 were female, and 1190 were male. Out of 375 first authors, 155 (41.33%) were female, and 220 (58.67%) were male. For last authors 110 (31.16%) were female, and 243 (68.84%) were male. In regards to ethnicity, 234 (62.40%) of the first authors were white, 103 (27.47%) were Asian, 24 (6.40%) were black, and 14 (3.73%) were Hispanic. For the last authors, 234 (66.29%) were white, 96 (27.20%) were Asian, 12 (3.40%) were black, and 11 (3.11%) were Hispanic. Most authors were from the USA, Canada, and the United Kingdom. The trend continued for the first and last authors of the articles. When looking at the general distribution, 1631 (82.2%) were based in high-income countries, 209 (10.5 %) were based in upper-middle-income countries, 135 (6.8%) were based in lower-middle-income countries, and 9 (0.5 %) were based in low-income countries.CONCLUSIONSAnalysis of the bibliographic data revealed that there is an overrepresentation of white authors and male authors, especially in the roles of first and last author. The more male authors a paper had the more likely they were to be cited. Additionally, analysis showed that papers whose authors are based in higher-income countries were more likely to be cited more often and published in higher impact journals. These findings highlight the lack of diversity among the authors in the AI fairness community whose work gains the largest readership, potentially compromising the very impartiality that the AI fairness community is working towards.

Background The difference in clinical outcomes after endoscopic submucosal dissection (ESD) and endoscopic mucosal resection (EMR) for early Barrett's esophagus (BE) neoplasia remains unclear. We compared the recurrence/residual tissue rates, resection outcomes, and adverse events after ESD and EMR for early BE neoplasia. Methods We included patients who underwent EMR or ESD for BE-associated high grade dysplasia (HGD) or T1a esophageal adenocarcinoma (EAC) at eight academic hospitals. We compared demographic, procedural, and histologic characteristics, and follow-up data. A time-to-event analysis was performed to evaluate recurrence/residual disease and a Kaplan–Meier curve was used to compare the groups. Results 243 patients (150 EMR; 93 ESD) were included. EMR had lower en bloc (43 % vs. 89 %; P 0.99), early bleeding (0.7 % vs. 1 %; P > 0.99), delayed bleeding (3.3 % vs. 2.1 %; P = 0.71), and stricture (10 % vs. 16 %; P = 0.16) between EMR and ESD. Patients with non-curative resections who underwent further therapy were excluded from the recurrence analysis. Recurrent/residual disease was 31.4 % [44/140] for EMR and 3.5 % [3/85] for ESD during a median (interquartile range) follow-up of 15.5 (6.75–30) and 8 (2–18) months, respectively. Recurrence-/residual disease-free survival was significantly higher in the ESD group. More patients required additional endoscopic resection procedures to treat recurrent/residual disease after EMR (EMR 24.2 % vs. ESD 3.5 %; P Conclusions ESD is safe and results in more definitive treatment of early BE neoplasia, with significantly lower recurrence/residual disease rates and less need for repeat endoscopic treatments than with EMR.

Over 16 million cases worldwide, severe acute respiratory syndrome coronavirus 2 has profoundly affected healthcare as we know it. Given reports of gastrointestinal involvement and viral shedding in the stool, it is unsurprising there are concerns that endoscopic equipment may be a potential vector of viral transmission. Here, we provide an overview of existing practices for endoscope reprocessing, recent developments in the field, and challenges in the COVID-19 environment. Current multi-society guidelines do not advise any change to endoscope disinfection protocols but emphasize strict adherence to recommended practices. However, endoscopy reprocessing staff may benefit from supplemental personal protective equipment measures, especially in high risk situations. Because thorough endoscope reprocessing is highly operator dependent, adequate training of personnel is critical for proper manual cleaning and disinfection of endoscopes that have potential to harbor virus. Bacterial contamination of duodenoscopes has caused outbreaks of infection from multidrug-resistant organisms, highlighting vulnerable areas. The emphasis of current studies is on optimization of disinfection and drying, minimization of simethicone use, and on quality control of endoscope reprocessing with sampling and microbiological culturing. Recent advances include novel approaches to endoscope sterilization, infection barrier methods, and design of partially or fully disposable duodenoscopes. Overall, the available data indicate that, when correctly executed, current reprocessing practices are sufficient in preventing SARS-COV-2 transmission.

The outcomes of endoscopic submucosal dissection (ESD) for T1b esophageal cancer (EC) and its recurrence rates remain unclear in the West. Using a multicenter cohort, we evaluated technical outcomes and recurrence rates of ESD in the treatment of pathologically staged T1b EC.We included patients who underwent ESD of T1b EC at 7 academic tertiary referral centers in the United States (n = 6) and Brazil (n = 1). We analyzed demographic, procedural, and histopathologic characteristics and follow-up data. Time-to-event analysis was performed to evaluate recurrence rates.Sixty-six patients with pathologically staged T1b EC after ESD were included in the study. A preprocedure staging EUS was available in 54 patients and was Tis/T1a in 27 patients (50%) and T1b in 27 patients (50%). En-bloc resection rate was 92.4% (61/66) and R0 resection rate was 54.5% (36/66). Forty-nine of 66 patients (74.2%) did not undergo surgery immediately after resection and went on to surveillance. Ten patients had ESD resection within the curative criteria, and no recurrences were seen in a 13-month (range, 3-18.5) follow-up period in these patients. Ten of 39 patients (25.6%) with noncurative resections had residual/recurrent disease. Of the 10 patients with noncurative resection, local recurrence alone was seen in 5 patients (12.8%) and metastatic recurrence in 5 patients (12.8%). On univariate analysis, R1 resection had a higher risk of recurrent disease (hazard ratio, 6.25; 95% confidence interval, 1.29-30.36; P = .023).EUS staging of T1b EC has poor accuracy, and a staging ESD should be considered in these patients. ESD R0 resection rates were low in T1b EC, and R1 resection was associated with recurrent disease. Patients with noncurative ESD resection of T1b EC who cannot undergo surgery should be surveyed closely, because recurrent disease was seen in 25% of these patients.

Background: A requirement of current good manufacturing practices for dietary supplements is that manufacturers must identify their dietary ingredients. DNA-based methods can provide species-level authentication that may sometimes be difficult to achieve using conventional morphological and chemical analysis methods. However, because of varying levels of DNA degradation in botanical materials, many commercial tests fail to generate consistent test results across all types of botanical materials. AOAC published guidelines for validation of botanical identification methods and proposed probability of identification (POI) as a method performance parameter. However, few DNA-based botanical authentication methods in the literature follow these guidelines and evaluate POI. Objective: To provide a targeted PCR method validation example that follows AOAC guidelines for validation of botanical identification methods. Methods: Using Matricaria chamomilla (chamomile) as an example, we performed a single-laboratory validation for a targeted PCR method that aimed to identify both raw and processed chamomile materials. The performance parameters of the test were evaluated by carrying out an inclusivity/exclusivity study and a Specified Superior Test Material/Specified Inferior Test Material study to demonstrate that the method’s POI meets industry requirements. Results: The chamomile samples were identified by the method and achieved a POI greater than 0.9 with respect to all types of chamomile botanical materials. Conclusions: The method was validated for DNA-based identification of raw and processed chamomile materials, such as sterilized powders and extracts. Highlights: This work will provide insight for laboratories and manufacturers that aim to develop and validate DNA-based botanical identification methods.

Background: The applications of deoxyribonucleic acid (DNA) barcoding methods have been extended from authenticating taxonomic provenance of animal products to identifying botanicals used as herbal medicine and in botanical dietary supplements. DNA barcoding methods for botanical identification must be adequately validated to meet regulatory compliance. Objective: The goal of this study is to provide a validation protocol for a two-tiered DNA barcoding method that aims to identify raw botanicals. Methods: A barcode database was computationally validated to define the barcode combinations that can unambiguously identify botanicals in the database. A maximum variation sampling technique was used to capture a wide range of perspectives relating to DNA barcode-based botanical identification, including plant parts and species distance, for the experimental validation. Twenty-two authenticated botanicals were purposively sampled from different plant parts—covering both closely related and distantly related species—to validate the two-tiered DNA barcoding method. The performance of the method was assessed on accuracy, precision, ruggedness, and uncertainty. Results: High accuracy (100%) and precision (1.0) were obtained from the validation samples. The method was also found to be rugged and have acceptable uncertainty. Conclusions: The method was validated and suitable for DNA-based identification of botanical raw materials listed in the current database. Highlights: This work will provide support guidance for manufacturers and regulatory policy makers to implement equivalent validated and compliant DNA-based testing in quality control processes to improve botanical raw material identification and authentication.

Quality control in botanical products begins with the raw material supply. Traditionally, botanical identification is performed through morphological assessment and chemical analytical methods. However, the lack of availability of botanists, especially in recent years, coupled with the need to enhance quality control to combat the stresses on the supply chain brought by increasing consumer demand and climate change, necessitates alternative approaches. The goal of this protocol is to facilitate botanical species identification using a portable qPCR system on the field or in any setting, where access to laboratory equipment and expertise is limited. Target DNA is amplified using dye-based qPCR, with DNA extracted from botanical reference materials serving as a positive control. The target DNA is identified by its specific amplification and matching its melting peak against the positive control. A detailed description of the steps and parameters, from hands-on field sample collection, to DNA extraction, PCR amplification, followed by data interpretation, has been included to ensure that readers can replicate this protocol. The results produced align with traditional laboratory botanical identification methods. The protocol is easy to perform and cost-effective, enabling quality testing on raw materials as close to the point of origin of the supply chain as possible.

Quality control in botanical products begins with the raw material supply. Traditionally, botanical identification is performed through morphological assessment and chemical analytical methods. However, the lack of availability of botanists, especially in recent years, coupled with the need to enhance quality control to combat the stresses on the supply chain brought by increasing consumer demand and climate change, necessitates alternative approaches. The goal of this protocol is to facilitate botanical species identification using a portable qPCR system on the field or in any setting, where access to laboratory equipment and expertise is limited. Target DNA is amplified using dye-based qPCR, with DNA extracted from botanical reference materials serving as a positive control. The target DNA is identified by its specific amplification and matching its melting peak against the positive control. A detailed description of the steps and parameters, from hands-on field sample collection, to DNA extraction, PCR amplification, followed by data interpretation, has been included to ensure that readers can replicate this protocol. The results produced align with traditional laboratory botanical identification methods. The protocol is easy to perform and cost-effective, enabling quality testing on raw materials as close to the point of origin of the supply chain as possible.

Parsley (Petroselinum crispum) leaf is an herb widely consumed for its health benefits. Due to similar morphological and chemical profiles, celery leaf may be a source of substitution in commercial parsley materials. In order to detect this substitution, the present work characterized parsley and celery leaf at the cultivar level by physical, chemical and DNA approaches. In contrast to the variations observed in physical appearances and chemical profiles that make verification of authenticity difficult, consistent differences observed between their DNA sequences are suitable for verifying parsley material authenticity. To identify parsley and detect celery simultaneously, a multiplex qPCR assay was developed and validated with respect to efficiency and specificity. Further testing indicated the assay can be used to detect 1% (w/w) celery in parsley materials with a probability of detection greater than 0.9. The developed method is well-suited for routine quality control to prevent parsley material misidentification in commercial trade.

Statins possess anti-inflammatory properties and have a protective effect in certain inflammatory conditions; however, their effect on the natural history of pancreatitis is unknown.The aim of this study is to assess the effect of statin exposure on the severity of pancreatitis and incidence of organ failure using a propensity-matched approach.A historical cohort study was conducted of adult patients with acute pancreatitis (AP) admitted in the Cleveland Clinic Health System between 2007 and 2014. All medication, clinical, and outcomes data were extracted from the electronic medical record. Factors that influence statin use were included in a propensity model to minimize selection bias. Patients on and off statins were matched (1:1) based on the propensity score to simulate a randomized controlled trial. Measured outcomes included pancreatitis severity (Revised Atlanta Classification), incidence of multisystem organ failure (MSOF), new MSOF, acute necrosis, and death. Additional surrogate markers of severity included hospital length of stay, Bedside Index of Severity of Acute Pancreatitis (BISAP), and presence of SIRS.A total of 110 subjects taking a statin at admission were matched with 210 subjects not on a statin. Known baseline factors that may influence statin use and severity of pancreatitis were evenly matched between the 2 groups. Patients on a statin were less likely to develop MSOF, severe AP and necrosis. Although less in-hospital death occurred in the statin group when compared to nonusers, the difference was not statistically significant (2% vs. 4%; P=0.38).Statin use is associated with decreased severity of AP observed as reduction in both overall MSOF incidence and new MSOF. Prospective randomized controlled trials are needed to determine the efficacy of statin drugs in the treatment of AP.

Necrosis significantly increases rates of morbidity and mortality in acute pancreatitis. Hospitalists and general internists are on the front lines in identifying severe cases and consulting the appropriate specialists for optimal multidisciplinary care.

"Endoscopic ultrasound (EUS)-guided ablative therapies have advanced significantly and have led to experimental applications in locations that have been difficult to image and/or reach with percutaneous approaches, such as the caudate and left lobe of the liver. EUS-guided treatments of the liver are under development. The literature has shown that many percutaneous ablative techniques are readily adaptable for EUS. In this review, the authors discuss the current developments on EUS-guided ablation of liver tumors, including injection of sclerosants, thermal therapy, and EUS-guided portal injection of chemotherapy."

Background We have established a multicenter international consortium to better understand the natural history of acute pancreatitis (AP) worldwide and to develop a platform for future randomized clinical trials. Methods The AP patient registry to examine novel therapies in clinical experience (APPRENTICE) was formed in July 2014. Detailed web-based questionnaires were then developed to prospectively capture information on demographics, etiology, pancreatitis history, comorbidities, risk factors, severity biomarkers, severity indices, health-care utilization, management strategies, and outcomes of AP patients. Results Between November 2015 and September 2016, a total of 20 sites (8 in the United States, 5 in Europe, 3 in South America, 2 in Mexico and 2 in India) prospectively enrolled 509 AP patients. All data were entered into the REDCap (Research Electronic Data Capture) database by participating centers and systematically reviewed by the coordinating site (University of Pittsburgh). The approaches and methodology are described in detail, along with an interim report on the demographic results. Conclusion APPRENTICE, an international collaboration of tertiary AP centers throughout the world, has demonstrated the feasibility of building a large, prospective, multicenter patient registry to study AP. Analysis of the collected data may provide a greater understanding of AP and APPRENTICE will serve as a future platform for randomized clinical trials. Keywords Acute pancreatitis, international multicenter consortium, methodology, APPRENTICE Ann Gastroenterol 2017; 30 (1): 106-113

Modern semiconductor chips are reaching higher power densities, requiring cooling capacities in excess of 100 W/cm2. Liquid cooling is widely seen as the appropriate technology to provide cooling for these devices, however, convenient, integrated cooling techniques are not available. We report the use of a chip cooling package designed to interface directly with an embedded fluidic system that routes cooling liquid on a printed circuit board. The package is designed as a “package-on-package” construction so that it can be packaged over an existing electronic chip to provide fluid-based cooling to a variety of components. This paper presents package design and construction, fluidic integration, and flow/thermal performance of this package.

Sensor and actuator networks are often the backbone of control applications. They are used in many different fields, such as health care, home automation, and industrial control. Despite the prevalence of sensor and actuator networks, there is a lack of tools to support the rapid prototyping of sensing and control applications. We have developed a system to support a generic platform for rapid prototyping of sensing and control applications in which different sensors and actuators are brought together to perform specific functions. At the heart of the system-of-systems is the Control INterface to Devices and Instruments (CINDI), a small Linux-based box with a number of standard interfaces. The system was built with simplicity in mind and to support component reuse.