Small Bias Research Articles

Uncovering gender gap in academia: A comprehensive analysis within the software engineering community

Gender gap in education has gained considerable attention in recent years, as it carries profound implications for the academic community. However, while the problem has been tackled from a student perspective, research is still lacking from an academic point of view. In this work, our main objective is to address this unexplored area by shedding light on the intricate dynamics of gender gap within the Software Engineering (SE) community. To this aim, we first review how the problem of gender gap in the SE community and in academia has been addressed by the literature so far. Results show that men in SE build more tightly-knit clusters but less global co-authorship relations than women, but the networks do not exhibit homophily. Concerning academic promotions, the Software Engineering community presents a higher bias in promotions to Associate Professors and a smaller bias in promotions to Full Professors than the overall Informatics community.

The Impact of Mind Mapping on Young Children's Writing Skills: An Experimental Study

This study investigates the application of the mind-mapping learning model to enhance children's writing skills. Conducted over two weeks during the 2022/2023 academic year at a school in Medan, North Sumatra, the research employed a quantitative experimental design. The study population included 106 children aged 3-8 years, with samples systematically selected from two classes. Data collection involved observation sheets to assess student engagement and activity levels. Statistical tests ensured data validity and reliability, with normality tests (Kolmogorov-Smirnov and Shapiro-Wilk) revealing normal and non-normal distributions and homogeneity tests confirming data uniformity. The t-test results indicated a significant positive impact of mind mapping on writing skills (Sig. = 0.000; t = 24.810). These findings suggest that mind mapping in early education can significantly enhance writing abilities, improving academic performance and cognitive development. The implications are substantial for educational practice, advocating for integrating mind mapping into teaching methods and supporting teacher training programs. Despite limitations such as a small sample size and potential sampling bias, this study provides a foundation for further research. Future studies should involve larger samples to validate these results and explore additional applications of mind mapping in educational settings, aiming to refine and enhance teaching strategies for early childhood education. Additionally, educational policymakers should consider integrating innovative teaching strategies like mind mapping to improve academic outcomes.

VEXP: A virtual enhanced cross screen panel for off-target pharmacology alerts

We describe the development of the GSK vEXP (virtual enhanced cross screen panel) for off-target pharmacology alerts. The derivation of a panel of machine learning classification models or QSAR models (Quantitative Structure-Activity Relationship) for off-target safety assessment allows early alerting to risk factors in candidate drugs. The models are matched to an internal in-vitro biochemical screening panel described previously with some updates reported here. The extreme imbalance of some internal GSK datasets and most of the related external ChEMBL datasets is shown when considering potency thresholds relevant to in-vitro screening. The small size and bias to the active class make many ChEMBL datasets un-modellable using such thresholds. Although larger, many GSK datasets remain too imbalanced to give a performant model. The value of merging internal and external data to help rebalance datasets and improve the domain of applicability is demonstrated with improvements in model performance frequently seen on merged data. Efforts to collate public datasets with a far better balance of the missing in-actives would likely do more to improve opensource models than simply increasing dataset size. We investigate the use of moving the probability threshold and applying imbalanced learners to help overcome the imbalance problem. Both methods can produce models with improved performance when applied to imbalanced datasets. Datasets with class imbalance 95:5 % or with <100 compounds were un-modellable. Where datasets had a class imbalance of 90:10 % the imbalanced learn methods were often more performant than standard tree-based classifiers. No one classification algorithm consistently out-performed all others and our approach emphasises a standardised, automated build and evaluate approach across all classifiers to identify the best model. The application of vEXP includes ranking of hit compounds for fast prioritisation, flagging of hit series that contain systematic scaffold or functional group related risks and the confirmation that late-stage optimisation is not introducing new off-target activities in established chemical series.

On the Magnification of Small Biases in Hiring

ABSTRACTWe analyze a setting in which a board must hire a chief executive officer (CEO) after exerting effort to learn about the quality of each candidate. Optimal effort is asymmetric, implying asymmetric likelihoods of each candidate being chosen. If the board has an infinitesimal bias in favor of one candidate, it allocates effort to maximize the likelihood of that candidate being chosen. Even when the board's prior is that its preferred candidate is inferior, she may still be chosen most often. A glass ceiling can also arise whereby the tendency to hire favored candidates increases as the importance of the position increases.

Exploring the Linkages between farm efficiency, farm environmental performance, and agri-environmental scheme adoption: Lessons from France

The agricultural detrimental effects on the environment are a source of concern. Public mea-sures, such as agri-environmental schemes (AES), have been designed to incentivize farmers to adopt more sound environmental practices on the farm. In this study, we examine the effects of past initial economic and environmental performances on AES adoption by focusing on crop farms. Using Firth's logistic regression to address small sample bias with French FADN data from 1997 to 2007, we mainly find that technical efficiency has heterogeneous effects on AES adoption, depending on environmental indexes. This result suggests the presence of windfall effects. We also show complex interactions (antagonism or synergy) between economic and environmental performances in adoption decisions, and heterogeneous effects depending on the type of farming.The agricultural detrimental effects on the environment are a source of concern. Public mea-sures, such as agri-environmental schemes (AES), have been designed to incentivize farmers to adopt more sound environmental practices on the farm. In this study, we examine the effects of past initial economic and environmental performances on AES adoption by focusing on crop farms. Using Firth's logistic regression to address small sample bias with French FADN data from 1997 to 2007, we mainly find that technical efficiency has heterogeneous effects on AES adoption, depending on environmental indexes. This result suggests the presence of windfall effects. We also show complex interactions (antagonism or synergy) between economic and environmental performances in adoption decisions, and heterogeneous effects depending on the type of farming.

Validation of formulae predicting stroke volume from arterial pressure: with particular emphasis on upright individuals in hot ambient conditions.

During heatwaves, it is important to monitor workers' cardiovascular health since 35% of those working in hot environments experience symptoms of heat strain. Wearable technology has been popularized for monitoring heart rate (HR) during recreational activities, but it can also be used to monitor occupational heat strain based on core and skin temperatures and HR. To our knowledge, no devices estimate the cardiovascular strain directly based on stroke volume (SV) or cardiac output (CO). In addition to the hardware, there are limitations regarding the lack of suitable algorithms that would provide such an index based on relevant physiological responses. The validation of the formulae already existing in literature was the principle aim of the present study. We monitored the cardiovascular responses of our participants to a supine and 60° head-up tilt at the same time each day. During the test, we measured blood pressure derived by finger photoplethysmography, which also provided beat-by-beat measures of SV and CO. Afterwards, we compared the SV derived from the photoplethysmography with the one calculated with the different equations that already exist in literature. The evaluation of the formulae was based on comparing the error of prediction. This residual analysis compared the sum of the squared residuals generated by each formula using the same data set. Our findings suggest that estimating SV with existing formulae is feasible, showing a good correlation and a relatively small bias. Thus, simply measuring workers' blood pressure during breaks could estimate their cardiac strain.

Optical Biometry versus Ultrasound A-scan in Measuring Anterior Chamber Depth, Axial Length, and Lens Thickness in Patients Undergoing Cataract Surgery

Ocular biometry measurements use ultrasonography and optical biometry, offering high precision for noninvasive intraocular distance measurement in cataract surgery. This study aims to Compare agreement values between an optical biometry machine and an ocular ultrasound-based biometry device for axial length, anterior chamber depth, and lens thickness measurements. A study enrolled 64 eyes of 42 patients scheduled for cataract surgery at Benghazi Teaching Eye Hospital, Libya, from January 1st to February 1st, 2024. Patients' medical and ocular histories were taken, and slit lamp examinations were conducted. All eyes with visual impairment and good fixation were included. Axial length, anterior chamber depth, and lens thickness were measured using Aladdin optical biometer and an ellex eye-cubed ultrasonic contact biometer. The study involved 64 eyes with a mean age of 64.36 years, The most common type of cataract was posterior subcapsular cataract [46.9%], nuclear cataract [14.1%], and cortical cataract [14.1%]. Combined cataracts were also present in 31.3% of the participants. The ICC analysis showed a strong agreement [0.976] between the two measurement devices in assessing axial length. difference in measurements of the mean anterior chamber depth. ICC showed a high level of agreement between the two methods [0.545]. There was no significant difference in lens thickness as measured by optical biometry and A-scan ultrasound, with a medium positive correlation between lens thickness measurement by the two devices [P&lt;0.001]. The interclass correlation further supported these findings, with a strong correlation between lens thickness and cataract development [r = 0.753]. The study reveals consistency in the measurements between optical and ultrasound biometry, highlighting the need for future research, to address limitations like small sample size and potential biases.

Particle composition-specific approach to estimate the particulate organic carbon concentration off the coastal areas of Guangdong and eastern Hainan Island.

High optical complexity caused by the variability of marine particles poses a major challenge to the development of bio-optical algorithms for particulate organic carbon (POC) concentration retrievals from optical measurements in coastal waters. Here, we developed a particle composition-specific approach to estimate POC off the coastal areas of Guangdong and eastern Hainan Island, China. The ratio of phytoplankton absorption to detritus absorption coefficient aph(443)/ad(443) was used to optically discriminate water types. The samples with aph(443)/ad(443) ≤ 4.9 showed a significant correlation between POC and absorption line height at 676 nm aLH(676) (R2 = 0.75, n = 70, p < 0.01). In contrast, aph-dominant samples with aph(443)/ad(443) > 4.9 had a high covariance between POC and particle scattering coefficient at 675 nm bp(675) (R2 = 0.85, n = 37, p < 0.01). Validation with an independent dataset yielded a small positive bias (R2 = 0.81, APD = 23.10%, RMSE = 29.01 mg m-3, RPD = 16.31%). The approach provided a better estimation of POC concentration in coastal waters compared with univariate algorithms. A depth-resolved index aLH(676)/bbp(442) was defined as the ratio of absorption line height to particle backscattering coefficient. Using the depth-resolved index instead of aph(443)/ad(443) for optical water type classification can be utilized to represent the vertical variations of POC in 1 m bins, and can complement remote sensing observations to accurately characterize the three-dimensional structure of POC distribution in the oceans.

Latent classification model for censored longitudinal binary outcome.

Latent classification model is a class of statistical methods for identifying unobserved class membership among the study samples using some observed data. In this study, we proposed a latent classification model that takes a censored longitudinal binary outcome variable and uses its changing pattern over time to predict individuals' latent class membership. Assuming the time-dependent outcome variables follow a continuous-time Markov chain, the proposed method has two primary goals: (1) estimate the distribution of the latent classes and predict individuals' class membership, and (2) estimate the class-specific transition rates and rate ratios. To assess the model's performance, we conducted a simulation study and verified that our algorithm produces accurate model estimates (ie, small bias) with reasonable confidence intervals (ie, achieving approximately 95% coverage probability). Furthermore, we compared our model to four other existing latent class models and demonstrated that our approach yields higher prediction accuracies for latent classes. We applied our proposed method to analyze the COVID-19 data in Houston, Texas, US collected between January first 2021 and December 31st 2021. Early reports on the COVID-19 pandemic showed that the severity of a SARS-CoV-2 infection tends to vary greatly by cases. We found that while demographic characteristics explain some of the differences in individuals' experience with COVID-19, some unaccounted-for latent variables were associated with the disease.

The Social Amplification of Distortions in Representations of Ratios

Abstract: When faced with a choice between two ratios, some people prefer the option with the larger numerator instead of the larger ratio. In this study, participants read texts containing two options involving ratios, indicated their preference between those options, and recalled the texts. Using the serial reproduction method, the recall protocols were then given to other participants to read, judge, and recall. Results show that some people’s representations of the problem are distorted, favoring the option with the largest numerator; these distortions are associated with increased ratio bias and are amplified as information is socially transmitted. Finally, this social amplification progressively increases the ratio bias. These results show how small biases can grow larger when information is socially transmitted.

Contextualizing Polarimetric Retrievals of Boundary Layer Height Using State-of-the-Art Boundary Layer Profiling

Abstract Knowledge about the depth of the planetary boundary layer (PBL) is crucial for a variety of applications, but direct observations of PBL depth are spatiotemporally sparse. Recent studies have proposed using operational dual-polarization weather radars to observe the evolution of PBL depth by capitalizing on unique differential reflectivity (ZDR) signatures of Bragg scatter at the top of the PBL. While this approach appears promising and cost-effective, uncertainties remain about the representativeness of these estimates and how its efficacy may vary by geography and climatology. To address these outstanding uncertainties, this study compares collocated observations collected from two WSR-88D radars and two state-of-the-art mobile boundary layer profiling systems and evaluates the proposed methodology over the full diurnal cycle. Results indicate good overall correspondence between the profiling- and radar-based PBL depth estimates, with an abrupt divergence during the early evening transition and large discrepancies overnight. Relatively large root-mean-square-deviations (RMSDs) coupled with small biases match expectations when comparing spatially averaged data with point observations during PBL growth, which capture frequent fluctuations. A qualitative examination of the radar data reveals signatures of elevated residual layers, clouds, and ground clutter, all of which can obfuscate the desired surface-based PBL signal but which may have their own utility. The prominence of the Bragg scatter signal is found to be correlated with the observed moisture gradient at the top of the PBL, reflecting climatological variability that should be considered. These findings motivate further work to improve the automated detection of Bragg scatter layers from polarimetric radar data. Significance Statement Knowledge of the height of the planetary boundary layer matters for weather forecasting, air quality, and renewable energy production. Currently, boundary layer height measurements are taken at select locations twice a day. However, a method to use the existing national network of polarimetric weather radars for this purpose has been proposed. This work evaluates this method against specialized boundary layer measurements. The results show that the method is generally reliable during the daytime and could be used for a variety of applications including climatologies and model evaluation. There remain a number of situational caveats, including residual turbulence, clouds/precipitation, ground clutter, and certain meteorological environments, that may require modification of the approach and need to be considered in future work.

A new, adaptive, self-administered, and generalizable method used to measure visual acuity.

Angular Indication Measurement (AIM) is an adaptive, self-administered, and generalizable orientation-judgment method designed to interrogate visual functions. We introduce AIM Visual Acuity (VA) and show its features and outcome measures. Angular Indication Measurement VA's ability to detect defocus was comparable with that of an Early Treatment of Diabetic Retinopathy Study (ETDRS) letter chart and showed greater sensitivity to astigmatic blur. This proof-of-concept study introduces Angular Indication Measurement and applies it to VA. First, we compared the ability of AIM-VA and ETDRS to detect defocus and astigmatic blur in 22 normally sighted adults. Spherical and cylindrical lenses in the dominant eye induced blur. Second, we compared repeatability over two tests of AIM-VA and ETDRS. A repeated-measure analysis of variance showed a main effect for defocus blur and test. For the astigmatism experiment, an interaction between blur and orientation was found. Pairwise comparisons showed that AIM was more sensitive to astigmatic-induced VA loss than ETDRS. Bland-Altman plots showed small bias and no systematic learning effect for either test type and improved repeatability with more than two adaptive steps for AIM-VA. Angular Indication Measurement VA's ability to detect defocus was comparable with that of an ETDRS letter chart and showed greater sensitivity to induced astigmatic blur, and AIM-VA's repeatability is comparable with ETDRS when using two or more adaptive steps. Angular Indication Measurement's self-administered orientation judgment approach is generalizable to interrogate other visual functions, e.g., contrast, color, motion, and stereovision.

Is race adjustment necessary to estimate glomerular filtration rate in South Brazilians?

Race coefficients (RC) in equations to estimate glomerular filtration rate (GFR) have been highly questioned. We aimed to evaluate the performance of three equations, namely 2009 Chronic Kidney Disease Epidemiology Collaboration (2009 CKD-EPI), 2021 CKD-EPI, and European Kidney Function Consortium (EKFC) in self-reported Black and White Brazilians. Our cross-sectional study compared estimated GFR (eGFR) with 51Cr-EDTA measured GFR (mGFR) in healthy adults, patients with type 2 diabetes mellitus with or without chronic kidney disease (CKD), and in non-diabetic individuals with CKD. The performance of these equations was assessed using Bland-Altman plots, Lin's concordance correlation coefficient (CCC), bias, P30, and P15 accuracy. Three hundred six White adults (aged 53 ± 17 years, 55% women, mean mGFR: 83 ± 32 mL/min/1.73m2) and 48 Black participants (aged 53 ± 17 years, 58% women, mGFR: 90 ± 34 mL/min/1.73m2) were included. No equation achieved the desirable P30 accuracy value of 90%, neither in White (2009 CKD-EPI:78%, 2021 CKD-EPI:76% and EKFC:77%, p = 0.368) nor in Black volunteers (respective values of 77%, 75%, and 77%; p = 0.882). The 2009 CKD-EPI showed the best performance in Black participants (bias: 4.04; CCC: 0.848), whereas the 2021 CKD-EPI performed better in Whites, with smaller bias (1.45), and better concordance correlation coefficient (0.790). The EKFC presented the worst performance. All equations underdiagnosed advanced CKD in White participants, but not in Black. The 2021 CKD-EPI does not outperform the 2009 CKD-EPI. Instead, it underestimated the occurrence of CKD in White participants. Thus, we do not recommend replacing the 2009 with the new 2021 CKD-EPI in the Brazilian population.

Differential centiloid scale normalization techniques: comparison between hybrid PET/MRI and independently acquired MRI.

Centiloid (CL) scales play an important role in semiquantitative analyses of amyloid-β (Aβ) PET. CLs are derived from the standardized uptake value ratio (SUVR), which needs Aβ positron emission tomography (PET) normalization processing. There are two methods to collect the T1-weighted imaging (T1WI) for normalization: (i) anatomical standardization using simultaneously acquired T1WI (PET/MRI), usually adapted to PET images from PET/MRI scanners, and (ii) T1WI from a separate examination (PET + MRI), usually adapted to PET images from PET/CT scanners. This study aimed to elucidate the correlations and differences in CLs between when using the above two T1WI collection methods. Among patients who underwent Aβ PET/MRI (using 11C-Pittuberg compound B (11C-PiB) or 18F-flutemetamol (18F-FMM)) at our institution from 2015 to 2023, we selected 49 patients who also underwent other additional MRI examinations, including T1WI for anatomic standardization within 3years. Thirty-one of them underwent 11C-PiB PET/MRI, and 18 participants underwent 18F-FMM PET/MRI. Twenty-five of them, additional MRI acquisition parameters were identical to simultaneous MRI during PET, and 24 participants were different.After normalization using PET/MRI or PET + MRI method each, SUVR was measured using the Global Alzheimer's Association Initiative Network cerebral cortical and striatum Volume of Interest templates (VOI) and whole cerebellum VOI. Subsequently, CLs were calculated using the previously established equations for each Aβ PET tracer. Between PET/MRI and PET + MRI methods, CLs correlated linearly in 11C-PiB PET (y = 1.00x - 0.11, R2 = 0.999), 18F-FMM PET (y = 0.97x - 0.12, 0.997), identical additional MRI acquisition (y = 1.00x + 0.33, 0.999), different acquisition (y = 0.98x - 0.43, 0.997), and entire study group (y = 1.00x - 0.24, 0.999). Wilcoxon signed-rank test revealed no significant differences: 11C-PiB (p = 0.49), 18F-FMM (0.08), and whole PET (0.46). However, significant differences were identified in identical acquisition (p = 0.04) and different acquisition (p = 0.02). Bland-Altman analysis documented only a small bias between PET/MRI and PET + MRI in 11C-PiB PET, 18F-FMM PET, identical additional MRI acquisition, different acquisition, and whole PET (-0.05, 0.67, -0.30, 0.78, and 0.21, respectively). Anatomical standardizations using PET/MRI and using PET + MRI can lead to almost equivalent CL. The CL values obtained using PET/MRI or PET + MRI normalization methods are consistent and comparable in clinical studies.

Cox proportional hazards regression in small studies of predictive biomarkers

Predictive biomarkers are essential for personalized medicine since they select the best treatment for a specific patient. However, of all biomarkers that are evaluated, only few are eventually used in clinical practice. Many promising biomarkers may be erroneously abandoned because they are investigated in small studies using standard statistical techniques which can cause small sample bias or lack of power. The standard technique for failure time endpoints is Cox proportional hazards regression with a multiplicative interaction term between binary variables of biomarker and treatment. Properties of this model in small studies have not been evaluated so far, therefore we performed a simulation study to understand its small sample behavior. As a remedy, we applied a Firth correction to the score function of the Cox model and obtained confidence intervals (CI) using a profile likelihood (PL) approach. These methods are generally recommended for small studies of different design. Our results show that a Cox model estimates the biomarker-treatment interaction term and the treatment effect in one of the biomarker subgroups with bias, and overestimates their standard errors. Bias is however reduced and power is increased with Firth correction and PL CIs. Hence, the modified Cox model and PL CI should be used instead of a standard Cox model with Wald based CI in small studies of predictive biomarkers.

Using chemical intramuscular fat percentage to predict visual marbling scores in Australian beef carcasses

Meat Standards Australia (MSA) marbling score and AUS-MEAT marbling represent key determinants of carcass value in the Australian beef industry and are well recognised traits in national and international markets. However, with the emergence of objective measurement technologies there are opportunities to grade beef carcasses using objective traits such as chemical intramuscular fat (IMF%) but abrupt changes to MSA and AUS-MEAT grading practices would cause significant disruption to the industry. Therefore, the objective of the study was to develop and validate models that transform chemical IMF% into equivalent MSA marbling scores and AUS-MEAT marbling. The second objective was to compare IMF%-derived and grader-derived MSA marbling scores when used as input values in the MSA model to generate predicted eating quality scores (MQ4). Carcasses (n = 5513) from industry experiments across 7 years (2017–2023) were graded for MSA marbling and AUS-MEAT marbling and sampled for chemical IMF%. Data were utilised to develop IMF%-derived models for MSA marbling score (IMF-MSAMB) and AUS-MEAT marbling (IMF-AUMB). Calibration performance was maintained when cross-validated and independently validated. The IMF-MSAMB model described 91% of the variation in MSA marbling score (Residual Standard Error (RSEV) = 57.9), with a slope of 0.90 and very small bias of −0.54. Similarly, IMF-AUMB described 88% of the variation in grader AUS-MEAT marbling (RSEV = 0.68) with a slope very close to 1 (0.94) and negligible bias (0.06). In addition, predicted MQ4 scores were almost equivalent irrespective of which marbling input value was used, across a suite of cut and cook combinations. Therefore, there is an opportunity for models to assist transition to the use of chemical IMF% in place of visual marbling scores. This would enable grading technologies to be calibrated and validated against chemical IMF%, whilst minimising industry disruption.

High-performance InGaAs/GaAsSb extended short-wave infrared Electron-Injection photodetector

In this paper, we report a high-sensitive extended short-wave infrared (e-SWIR) Electron-Injection (EI) photodetector based on InGaAs/GaAsSb type-II superlattice (T2SL). To achieve high gain and low dark current noise, the EI photodetector involved in this letter employs a multi-layer heterogeneous band structure, composed of InAlAs, GaAsSb and InGaAs/GaAsSb T2SL, which can facilitate electron injection and effectively reduce recombination and thermal generation within the device. Thanks to the unique band alignment, this EI photodetector operates in high-gain linear-mode and requires only a small bias voltage about 0.8 V. At 200 K, the detector exhibits a 100 % cut-off wavelength of ∼ 2.7 μm, a peak responsivity of 3693.2 A/W corresponding to a gain of 7979.0 at 1.4 V, and a gain-normalized dark current density (GNDCD) of 1.4 × 10-6 A/cm2, which results in a remarkable specific detectivity of 6.0 × 1013 cm·Hz1/2/W. When operating at room temperature, this EI photodetector presents a decent responsivity of over 2000.0 A/W. Our results pave a potential way for ultra-sensitive e-SWIR infrared detection at high temperature.

InP-based GaAsSb/AlGaAsSb/T2SL barrier-type low-bias tunable dual-band NIR/eSWIR photodetectors.

A bias-selectable near-infrared (NIR) and extended short wavelength infrared (eSWIR) dual-band bandgap engineered Ga0.51As0.49Sb/Al0.85Ga0.15AsSb/T2SL (In0.53Ga0.47 As/Ga0.51As0.49Sb) infrared photodetector, vertically stacked in a monolithic grown on InP substrate, is demonstrated. GaAsSb NIR sub-detector and T2SL eSWIR sub-detector are operated under small forward and reverse bias, respectively. The GaAsSb sub-detector functions within the NIR spectrum, with a 100% cutoff wavelength of 1.72 μm at 50 mV, achieving a peak responsivity of 0.560 A/W at 1.55 μm and a specific detectivity (D*) of 1.48 ×1011 c m⋅H z 1/2/W. At -250 mV, the T2SL eSWIR sub-detector functions in the eSWIR band, exhibiting a 100% cutoff wavelength of 2.6 μm. The peak responsivity is 0.273 A/W at 2.0 μm, with a specific detectivity of 6.11 ×109 c m⋅H z 1/2/W. The present work demonstrates the potential of the dual-band photodetector for multispectral SWIR applications.

Dermatology resident comfort level treating hair conditions related to patients with skin of color.

Although recent studies demonstrated resident satisfaction in the treatment of skin of color (SOC) related disease, comfort levels treating hair specific to populations within the SOC spectrum is unclear. The purpose of this study is to assess dermatology residents comfort level in recognizing and treating various common hair conditions with a focus on those specific to SOC. An Institutional Review Board-approved survey was distributed to United States residents of Accreditation Council for Graduate Medical Education-accredited dermatology programs. Data pertaining to hair care knowledge and treatment comfort levels were collected. Analysis was completed using equal variance 2-sample t tests and analysis of variance F tests, P < .05. Dermatology residents were relatively comfortable with common conditions such as androgenetic alopecia and alopecia areata, but uncomfortable with creating healthy hair regimens, discussing natural hair care products, and treating trichorrhexis nodosa. Resident self-identification as underrepresented in medicine significantly impacted resident knowledge of hair care and treatment in patients with SOC. This study was limited due to small sample size and potential recall bias. This study highlights knowledge gaps in understanding hair-related care for patients with SOC, affirming the continued importance of diversifying dermatology programs as well as hair-specific training for residents.

Zero-Shot Building Age Classification from Facade Image Using GPT-4

Abstract. A building’s age of construction is crucial for supporting many geospatial applications. Much current research focuses on estimating building age from facade images using deep learning. However, building an accurate deep learning model requires a considerable amount of labelled training data, and the trained models often have geographical constraints. Recently, large pre-trained vision language models (VLMs) such as GPT-4 Vision, which demonstrate significant generalisation capabilities, have emerged as potential training-free tools for dealing with specific vision tasks, but their applicability and reliability for building information remain unexplored. In this study, a zero-shot building age classifier for facade images is developed using prompts that include logical instructions. Taking London as a test case, we introduce a new dataset, FI-London, comprising facade images and building age epochs. Although the training-free classifier achieved a modest accuracy of 39.69%, the mean absolute error of 0.85 decades indicates that the model can predict building age epochs successfully albeit with a small bias. The ensuing discussion reveals that the classifier struggles to predict the age of very old buildings and is challenged by fine-grained predictions within 2 decades. Overall, the classifier utilising GPT-4 Vision is capable of predicting the rough age epoch of a building from a single facade image without any training. The code and dataset are available at https://zichaozeng.github.io/ba_classifier.