Search

As machine learning (ML) has emerged as the predominant technological paradigm for artificial intelligence (AI), complex black box models such as GPT-4 have gained widespread adoption. Concurrently, explainable AI (XAI) has risen in significance as a counterbalancing force. But the rapid expansion of this research domain has led to a proliferation of terminology and an array of diverse definitions, making it increasingly challenging to maintain coherence. This confusion of languages also stems from the plethora of different perspectives on XAI, e.g. ethics, law, standardization and computer science. This situation threatens to create a “tower of Babel” effect, whereby a multitude of languages impedes the establishment of a common (scientific) ground. In response, this paper first maps different vocabularies, used in ethics, law and standardization. It shows that despite a quest for standardized, uniform XAI definitions, there is still a confusion of languages. Drawing lessons from these viewpoints, it subsequently proposes a methodology for identifying a unified lexicon from a scientific standpoint. This could aid the scientific community in presenting a more unified front to better influence ongoing definition efforts in law and standardization, often without enough scientific representation, which will shape the nature of AI and XAI in the future.

Importance: Identifying new prognostic features in colon cancer has the potential to refine histopathologic review and inform patient care. Although prognostic artificial intelligence systems have recently demonstrated significant risk stratification for several cancer types, studies have not yet shown that the machine learning-derived features associated with these prognostic artificial intelligence systems are both interpretable and usable by pathologists. Objective: To evaluate whether pathologist scoring of a histopathologic feature previously identified by machine learning is associated with survival among patients with colon cancer. Design, Setting, and Participants: This prognostic study used deidentified, archived colorectal cancer cases from January 2013 to December 2015 from the University of Milano-Bicocca. All available histologic slides from 258 consecutive colon adenocarcinoma cases were reviewed from December 2021 to February 2022 by 2 pathologists, who conducted semiquantitative scoring for tumor adipose feature (TAF), which was previously identified via a prognostic deep learning model developed with an independent colorectal cancer cohort. Main Outcomes and Measures: Prognostic value of TAF for overall survival and disease-specific survival as measured by univariable and multivariable regression analyses. Interpathologist agreement in TAF scoring was also evaluated. Results: A total of 258 colon adenocarcinoma histopathologic cases from 258 patients (138 men [53%]; median age, 67 years [IQR, 65-81 years]) with stage II (n = 119) or stage III (n = 139) cancer were included. Tumor adipose feature was identified in 120 cases (widespread in 63 cases, multifocal in 31, and unifocal in 26). For overall survival analysis after adjustment for tumor stage, TAF was independently prognostic in 2 ways: TAF as a binary feature (presence vs absence: hazard ratio [HR] for presence of TAF, 1.55 [95% CI, 1.07-2.25]; P = .02) and TAF as a semiquantitative categorica

Artificial intelligence (AI) solutions that automatically extract information from digital histology images have shown great promise for improving pathological diagnosis. Prior to routine use, it is important to evaluate their predictive performance and obtain regulatory approval. This assessment requires appropriate test datasets. However, compiling such datasets is challenging and specific recommendations are missing. A committee of various stakeholders, including commercial AI developers, pathologists, and researchers, discussed key aspects and conducted extensive literature reviews on test datasets in pathology. Here, we summarize the results and derive general recommendations on compiling test datasets. We address several questions: Which and how many images are needed? How to deal with low-prevalence subsets? How can potential bias be detected? How should datasets be reported? What are the regulatory requirements in different countries? The recommendations are intended to help AI developers demonstrate the utility of their products and to help pathologists and regulatory agencies verify reported performance measures. Further research is needed to formulate criteria for sufficiently representative test datasets so that AI solutions can operate with less user intervention and better support diagnostic workflows in the future.

BACKGROUND: Preliminary evidence suggests a nephroprotective effect of urinary alkalinization in patients at risk of acute kidney injury. In this study, we tested whether prophylactic bicarbonate-based infusion reduces the incidence of acute kidney injury and tubular damage in patients undergoing open heart surgery. METHODS AND FINDINGS: In a multicenter, double-blinded (patients, clinical and research personnel), randomized controlled trial we enrolled 350 adult patients undergoing open heart surgery with the use of cardiopulmonary bypass. At induction of anesthesia, patients received either 24 hours of intravenous infusion of sodium bicarbonate (5.1 mmol/kg) or sodium chloride (5.1 mmol/kg). The primary endpoint was the proportion of patients developing acute kidney injury. Secondary endpoints included the magnitude of acute tubular damage as measured by urinary neutrophil gelatinase-associated lipocalin (NGAL), initiation of acute renal replacement therapy, and mortality. The study was stopped early under recommendation of the Data Safety and Monitoring Committee because interim analysis suggested likely lack of efficacy and possible harm. Groups were non-significantly different at baseline except that a greater proportion of patients in the sodium bicarbonate group (66/174 [38%]) presented with preoperative chronic kidney disease compared to control (44/176 [25%]; p = 0.009). Sodium bicarbonate increased urinary pH (from 6.0 to 7.5, p<0.001). More patients receiving bicarbonate (83/174 [47.7%]) developed acute kidney injury compared with control patients (64/176 [36.4%], odds ratio [OR] 1.60 [95% CI 1.04-2.45]; unadjusted p = 0.032). After multivariable adjustment, a non-significant unfavorable group difference affecting patients receiving sodium bicarbonate was found for the primary endpoint (OR 1.45 [0.90-2.33], p = 0.120]). A greater postoperative increase in urinary NGAL in patients receiving bicarbonate infusion was observed compared to control patients (p

INTRODUCTION: Conventional markers of acute kidney injury (AKI) lack diagnostic accuracy and are expressed only late after cardiac surgery with cardiopulmonary bypass (CPB). Recently, interest has focused on hepcidin, a regulator of iron homeostasis, as a unique renal biomarker. METHODS: We studied 100 adult patients in the control arm of a randomized, controlled trial http://www.clinicaltrials.gov/NCT00672334 who were identified as being at increased risk of AKI after cardiac surgery with CPB. AKI was defined according to the Risk, Injury, Failure, Loss, End-stage renal disease classification of AKI classification stage. Samples of plasma and urine were obtained simultaneously (1) before CPB (2) six hours after the start of CPB and (3) twenty-four hours after CPB. Plasma and urine hepcidin 25-isoforms were quantified by competitive enzyme-linked immunoassay. RESULTS: In AKI-free patients (N = 91), urine hepcidin concentrations had largely increased at six and twenty-four hours after CPB, and they were three to seven times higher compared to patients with subsequent AKI (N = 9) in whom postoperative urine hepcidin remained at preoperative levels (P = 0.004, P = 0.002). Furthermore, higher urine hepcidin and, even more so, urine hepcidin adjusted to urine creatinine at six hours after CPB discriminated patients who did not develop AKI (area under the curve (AUC) receiver operating characteristic curve 0.80 [95% confidence interval (95% CI) 0.71 to 0.87] and 0.88 [95% CI 0.78 to 0.97]) or did not need renal replacement therapy initiation (AUC 0.81 [95% CI 0.72 to 0.88] 0.88 [95% CI 0.70 to 0.99]) from those who did. At six hours, urine hepcidin adjusted to urine creatinine was an independent predictor of ruling out AKI (P = 0.011). Plasma hepcidin did not predict no development of AKI. The study findings remained essentially unchanged after excluding patients with preoperative chronic kidney disease. CONCLUSIONS: Our findings suggest that urine hepcidin is an early pr

Thin boron nitride films were deposited using ion beam assisted deposition. The fraction of the cubic phase relative to the non-cubic phase was varied by changing the assisting ion to boron atom arrival ratio during deposition. Different ex situ analytical techniques were used to resolve the layered growth, i.e. cubic BN on top of non-cubic BN. Variation of the argon incorporation measured with Rutherford backscattering spectroscopy yields information on the layered growth. Polarized IR reflection data were analysed with a multilayer model taking into account the anisotropy of hexagonal BN. The parameters of the optical model, such as the thickness of the non-cubic BN interlayer and the volume fraction of cubic BN in the toplayer, are more useful for the characterization of BN films than the established IR peak ratio. Reflection electron energy loss spectroscopy and X-ray reflectivity measurements were also performed to verify the results.

Six different types of boron nitride films were investigated by polarized infrared reflection spectroscopy. Films with a highly cubic, mixed cubic and non-cubic, and exclusively non-cubic phase composition were synthesized using ion beam assisted deposition. Additionally, post-deposition argon ion irradiated cubic and non-cubic boron nitride films as well as nitrogen implanted boron sample were analyzed. Using this technique, besides the cubic phase, two different non-cubic modifications, layered anisotropic and amorphous, could be distinguished. A preferentional orientation of the normal axis of the sp2-bonded basal planes parallel to the substrate surface was observed.