搜索结果：MacBook

This dataset comprises 5452 images of durian plant parts-including leaves, flowers, branches, stems, and roots-affected by ten common disease classes. The images were captured from one family-owned durian orchard and four nearby orchards in Vinh Long Province, Vietnam. Each class contains approximately 405-427 raw images, photographed using an iPhone 14 under natural field conditions. These conditions simulate typical farmer photography practices, featuring varied angles, inconsistent lighting, and complex environmental backgrounds, resulting in significant visual noise. All raw JPEG images were manually reviewed and cropped on macOS systems using MacBook devices equipped with Apple M4 chips to focus on disease-affected regions, reduce file size, and minimize background noise. The processed, cropped images are provided in PNG format with variable dimensions. Images were resized to 224×224 pixels only during model training for machine learning experiments. Disease symptoms were verified in collaboration with plant pathologists to ensure accurate classification. This dataset is publicly available on Mendeley Data and is suitable for developing and evaluating machine learning models in plant disease classification. It is particularly valuable for testing model performance under real-world, noisy conditions and for supporting the creation of mobile or edge-based diagnostic tools in agriculture.

To address the triple bottlenecks of data scarcity, oversized models, and slow inference that hinder Cantonese automatic speech recognition (ASR) in low-resource and edge-deployment settings, this study proposes a cost-effective Cantonese ASR system based on LoRA fine-tuning and INT8 quantization. First, Whisper-tiny is parameter-efficiently fine-tuned on the Common Voice zh-HK training set using LoRA with rank = 8. Only 1.6% of the original weights are updated, reducing the character error rate (CER) from 49.5% to 11.1%, a performance close to full fine-tuning (10.3%), while cutting the training memory footprint and computational cost by approximately one order of magnitude. Next, the fine-tuned model is compressed into a 60 MB INT8 checkpoint via dynamic quantization in ONNX Runtime. On a MacBook Pro M1 Max CPU, the quantized model achieves an RTF = 0.20 (offline inference 5 &#xd7; real-time) and 43% lower latency than the FP16 baseline; on an NVIDIA A10 GPU, it reaches RTF = 0.06, meeting the requirements of high-concurrency cloud services. Ablation studies confirm that the LoRA-INT8 configuration offers the best trade-off among accuracy, speed, and model size. Limitations include the absence of spontaneous-speech noise data, extreme-hardware validation, and adaptive LoRA structure optimization. Future work will incorporate large-scale self-supervised pre-training, tone-aware loss functions, AdaLoRA architecture search, and INT4/NPU quantization, and will establish an mJ/char energy-accuracy curve. The ultimate goal is to achieve CER &#x2264; 8%, RTF < 0.1, and mJ/char < 1 for low-power real-time Cantonese ASR in practical IoT scenarios.

Visual analysis of paranasal sinuses (PNS) on computed tomography (CT) images requires interpretation and reporting of sinus involvement and other anatomical factors. This is time consuming, labour intensive and subjective. Artificial intelligence (AI)-based machine learning (ML) tools are under development for analysis of radiological images. The scope of this study was to develop and evaluate a coding-free ML model for automated identification of PNS on CT images. A total of 19,119 anonymous coronal images retrieved from 90 CT studies were included. All images were annotated with locations, names and opacification status of the sinuses. The images were divided into training, validation and testing datasets. The ML model was trained for 2000 iterations using YOLOv2 algorithm, and its accuracy was evaluated using F1 score and Intersection over Union (IoU) metrics. An ML model was developed using "Create ML" application on an Apple MacBook computer. A mean F1 score of 0.89 and a mean IoU50 of 79% was achieved during evaluation of the model on the testing dataset. The highest accuracy was seen in the detection of normal sphenoid sinus, and the lowest in the detection of opacified frontal sinus. The study demonstrates the utility of AI and ML in automating the interpretation of PNS CT images. From the results of our study, it can be concluded that a coding-free ML model can be developed and deployed for automated identification of PNS on CT images with accuracy similar to custom-coded ML models.

The AS7341 is a compact, multi-spectral digital sensor that functions as a spectrometer, measuring spectral response across eight visible channels (350-750&#xa0;nm), plus total intensity and near-infrared (NIR). Integrated with an open-source microcontroller, it was used to estimate the correlated color temperature (CCT)-a measure in Kelvin of the perceived warmth or coolness of a light source-of various display monitors. Calibration was performed using an industry-standard X-Rite Color Checker chart by capturing chromaticity coordinates. After calibration, five monitors were evaluated under consistent settings. The MacBook Air M2 (2022) showed CCT values closest to predicted standards, while all devices fell within expected manufacturer ranges. Results highlight the AS7341's accuracy and cost-effectiveness for CCT estimation. Its small size and portability enhance its practicality as a low-cost spectrometer for digital display analysis.

The usefulness and effectiveness of telepractice have been reported in recent years. Treatment of cleft palate patients with compensatory articulation is based on perceptual identification. Telepractice using videoconferencing platforms causes voice signal distortion and impacts auditory-perceptual perception. This study aimed acoustically examine voice signal distortion and determine the optimal videoconferencing platforms, in addition to the phonemes that can be discriminated with the same quality as in face-to-face interactions. ATR503 with 50 phoneme-balanced Japanese speech sentences was used as a reference corpus. Four videoconferencing platforms, -Zoom, Cisco Webex, Skype, and Google Meet, -and five devices, -iPhone, Android, iPad Air, Windows, and MacBook Pro were used as transmission conditions to examine voice signal distortions with the objective measure log-spectral distortion (LSD). Tukey's test was conducted to evaluate the degree of consonant distortion related to voicings (voiceless and voiced), places of articulation (bilabial, alveolar, alveolo-palatal, palatal, velar, labial-velar, and glottal), and manners of articulation (plosive, fricative, affricate, tap or flap, nasal, and approximant). With statistically significant differences, voiced, bilabial, labial-velar, nasal, and plosive consonants exhibited smaller distortions. In contrast, voiceless, alveolo-palatal, fricative, and affricate consonants exhibited larger distortions. Google Meet exhibited the lowest distortion among videoconferencing platforms and MacBook exhibited the lowest distortion among devices. This study provides significant insights into the telepractice strategies with the appropriate videoconferencing platform and device, and useful settings for cleft palate patients with compensatory articulations with respect to acoustics.

This study aimed to determine the molecular epidemiology of colistin-resistant A. baumannii in the last ten years and the frequency of gene regions related to pathogenesis, to compare the methods used to detect genes, and to confirm colistin resistance. This meta-analysis study was conducted under Preferred Reporting Items for Systematic Reviews and Meta-Analysis Guidelines. In the meta-analysis, research articles published in English and Turkish in electronic databases between January 2012 and November 2023 were examined. International Business Machines (IBM) Statistical Package for the Social Sciences (SPSS) Statistics for Macbook (Version 25.0. Armonk, NY, USA) was used for statistical analysis. The Comprehensive Meta-Analysis (CMA) (Version 3.0. Biostat, NJ, USA) program was used for heterogeneity assessment in the articles included in the meta-analysis. After evaluating the studies according to the elimination criteria, 18 original articles were included. Among colistin-resistant strains, blaOXA-51 positivity was 243 (19.61%), blaOXA-23 was 113 (9.12%), blaOXA-58 was 7 (0.56%), blaOXA-143 was 15 (1.21%), and blaOXA-72 was seen in two (0.16%) strains. The positivity rates of pmrA, pmrB, and pmrC were found to be 22 (1.77%), 26 (2.09%), and 6 (0.48%). The mcr-1 rate was found to be 91 (7.34%), the mcr-2 rate was 78 (6.29%), and the mcr-3 rate was 82 (6.61%). The colistin resistance rate in our study was found to be high. However, only some research articles report and/or investigate more than one resistance gene together. Additionally, it may be challenging to explain colistin resistance solely by expressing resistance genes without discussing accompanying components such as efflux pumps, virulence factors, etc.

In certain healthcare settings, such as emergency or critical care units, where quick and accurate real-time analysis and decision-making are required, the healthcare system can leverage the power of artificial intelligence (AI) models to support decision-making and prevent complications. This paper investigates the optimization of healthcare AI models based on time complexity, hyper-parameter tuning, and XAI for a classification task. The paper highlights the significance of a lightweight convolutional neural network (CNN) for analysing and classifying Magnetic Resonance Imaging (MRI) in real-time and is compared with CNN-RandomForest (CNN-RF). The role of hyper-parameter is also examined in finding optimal configurations that enhance the model's performance while efficiently utilizing the limited computational resources. Finally, the benefits of incorporating the XAI technique (e.g. GradCAM and Layer-wise Relevance Propagation) in providing transparency and interpretable explanations of AI model predictions, fostering trust, and error/bias detection are explored. Our inference time on a MacBook laptop for 323 test images of size 100x100 is only 2.6 sec, which is merely 8 milliseconds per image while providing comparable classification accuracy with the ensemble model of CNN-RF classifiers. Using the proposed model, clinicians/cardiologists can achieve accurate and reliable results while ensuring patients' safety and answering questions imposed by the General Data Protection Regulation (GDPR). The proposed investigative study will advance the understanding and acceptance of AI systems in connected healthcare settings.

There has been incremental progress in moving BCI out of the laboratory environment and into the homes of those who would benefit most, especially children living with severe physical disabilities. Practical issues, such as available computational resources and long calibration times, have slowed down the adoption of such systems. To develop an efficient and scalable machine learning framework consistent with early approaches that facilitate at-home BCI use, this study provides valuable insights into measuring the behavioral characteristics of a Raspberry Pi 4 (RPi4) during the operation and execution of standard BCI processes, including the training and evaluation of classifier models. The results, which evaluated ten standard classifiers, including the Riemannian Geometry (RG) framework and more advanced deep learning approaches like Artificial Neural Network (ANN), were profiled on RPi4. These were compared to Desktop and MacBook computations for metrics such as training time, inference time, peak memory, and incremental memory usage, with computational bottlenecks identified. Our assessment revealed comparable performance metrics (84.3% of accuracy, recall, and f1_score, and 84.7% precision) for the neural network models despite the lower computational resources. Profiling results, including 1.74 sec training time, 0.405 sec inference time, 1154.9 MiB peak memory, and 405.2 MiB incremental memory usage, also demonstrated that the RPi4 is a potentially viable device for low-cost BCI systems. However, high-resource demanding classifiers such as ANN may need to be carefully considered in their implementation, which, in turn, will scale down the potential cost and complexity of adopting practical, impactful at-home BCI systems.

Thispaper compares the usability of various Apple MacBook Pro laptops were tested for basic machine learning research applications, including text-based, vision-based, and tabular data. Four tests/benchmarks were conducted using four different MacBook Pro models-M1, M1 Pro, M2, and M2 Pro. A script written in Swift was used to train and evaluate four machine learning models using the Create ML framework, and the process was repeated three times. The script also measured performance metrics, including time results. The results were presented in tables, allowing for a comparison of the performance of each device and the impact of their hardware architectures.

Cloud computing is widely used in various sectors such as finance, health care, and education. Factors such as cost optimization, interoperability, data analysis, and data ownership functionalities are attracting healthcare industry to use cloud services. Security and forensic concerns are associated in cloud environments as sensitive healthcare data can attract the outside attacker and inside malicious events. Storage is the most used service in cloud computing environments. Data stored in iCloud (Apple Inc. Cloud Service Provider) is accessible via a Web browser, cloud client application, or mobile application. Apple Inc. provides iCloud service to synchronize data from MacBook, iPhone, iPad, etc. Core applications such as Mail, Contacts, Calendar, Photos, Notes, Reminders, and Keynote are synced with iCloud. Various operations can be performed on cloud data, including editing, deleting, uploading, and downloading data, as well as synchronizing data between devices. These operations generate log files and directories that are essential from an investigative perspective. This paper presents a taxonomy of iCloud forensic tools that provides a searchable catalog for forensic practitioners to identify the tools that meet their technical requirements. A case study involving healthcare data storage on iCloud service demonstrates that artifacts related to environmental information, browser activities (history, cookies, cache), synchronization activities, log files, directories, data content, and iCloud user activities are stored on a MacBook system. A GUI-based dashboard is developed to support iCloud forensics, specifically the collection of artifacts from a MacBook system.

To evaluate the accuracy of different viewing monitors for image reading and grading of diabetic retinopathy (DR). Single-centre, experimental case series-evaluation of reading devices for DR screening. A total of 100 sets of three-field (optic disc, macula, and temporal views) colour retinal still images (50 normal and 50 with DR) captured by FF 450 plus (Carl Zeiss) were interpreted on 27-inch iMac, 15-inch MacBook Pro, and 9.7-inch iPad. All images were interpreted by a retinal specialist and a medical officer. We calculated the sensitivity and specificity of 15-inch MacBook Pro and 9.7-inch iPad in detection of DR signs and grades with reference to the reading outcomes obtained using a 27-inch iMac reading monitor. In detection of any grade of DR, the 15-inch MacBook Pro had sensitivity and specificity of 96% (95% confidence interval (CI): 85.1-99.3) and 96% (95% CI: 85.1-99.3), respectively, for retinal specialist and 91.5% (95% CI: 78.7-97.2) and 94.3% (95% CI: 83.3-98.5), respectively, for medical officer, whereas for 9.7-inch iPad, they were 91.8% (95% CI: 79.5-97.4) and 94.1% (95% CI: 82.8-98.5), respectively, for retinal specialist and 91.3% (95% CI: 78.3-97.1) and 92.6% (95% CI: 81.3-97.6), respectively, for medical officer. The 15-inch MacBook Pro and 9.7-inch iPad had excellent sensitivity and specificity in detecting DR and hence, both screen sizes can be utilized to effectively interpret colour retinal still images for DR remotely in a routine, mobile or tele-ophthalmology setting. Future studies could explore the use of more economical devices with smaller viewing resolutions to reduce cost implementation of DR screening services.

In this paper, the authors have compared all of the currently available Apple MacBook Pro laptops, in terms of their usability for basic machine learning research applications (text-based, vision-based, tabular). The paper presents four tests/benchmarks, comparing four Apple Macbook Pro laptop versions: Intel based (i5) and three Apple based (M1, M1 Pro and M1 Max). A script in the Swift programming language was prepared, whose goal was to conduct the training and evaluation process for four machine learning (ML) models. It used the Create ML framework-Apple's solution dedicated to ML model creation on macOS devices. The training and evaluation processes were performed three times. While running, the script performed measurements of their performance, including the time results. The results were compared with each other in tables, which allowed to compare and discuss the performance of individual devices and the benefits of the specificity of their hardware architectures.

Modern cell phones allow for easy communication and transfer of data between devices. Unfortunately, some of the data transferred can be of unwelcomed, illicit, or threatening imagery and other files; digital forensic examiners are often asked to identify the source of these files. In this project, we developed a method to gain insights into the device used to send a file via Apple AirDrop. Our method brute forces the partial SHA256 hash entries found in the receiving Apple device's sysdiagnose logs to reveal the sender's phone number, even if that phone number was not known by the receiving device. This research publishes a method to generate permutations of the partial hash values using potential US area codes to identify the complete phone number of the sending device. In this research project, exemplar photographs were transmitted via AirDrop between Apple devices running iOS 15. A sysdiagnose was then generated on the receiving phone and exported by AirDrop to a MacBook Air for analysis. The analysis of the generated sysdiagnose archive found a partial SHA-256 hash of the sending device's phone number. This research identified a method to generate permutations of the partial SHA-256 hashes using a possible country and area code for the sending device in order to successfully identify the sending device's phone number. As a result, it was found that the sender of an unknown AirDrop file's phone number can be identified from the receiving device's sysdiagnose log files.

Delivery of acute stroke endovascular intervention can be challenging because it requires complex coordination of patient and staff across many different locations. In this proof-of-concept paper we (a) examine whether WiFi fingerprinting is a feasible machine learning (ML)-based real-time location system (RTLS) technology that can provide accurate real-time location information within a hospital setting, and (b) hypothesize its potential application in streamlining acute stroke endovascular intervention. We conducted our study in a comprehensive stroke care unit in Melbourne, Australia that offers a 24-hour mechanical thrombectomy service. ML algorithms including K-nearest neighbors, decision tree, random forest, support vector machine and ensemble models were trained and tested on a public WiFi dataset and the study hospital WiFi dataset. The hospital dataset was collected using the WiFi explorer software (version 3.0.2) on a MacBook Pro (AirPort Extreme, Broadcom BCM43x&#xd7;1.0). Data analysis was implemented in the Python programming environment using the scikit-learn package. The primary statistical measure for algorithm performance was the accuracy of location prediction. ML-based WiFi fingerprinting can accurately predict the different hospital zones relevant in the acute endovascular intervention workflow such as emergency department, CT room and angiography suite. The most accurate algorithms were random forest and support vector machine, both of which were 98% accurate. The algorithms remain robust when new data points, which were distinct from the training dataset, were tested. ML-based RTLS technology using WiFi fingerprinting has the potential to streamline delivery of acute stroke endovascular intervention by efficiently tracking patient and staff movement during stroke calls.

Human bodily mechanisms and functions produce low-frequency vibrations. Our ability to perceive these vibrations is limited by our range of hearing. However, in-ear infrasonic hemodynography (IH) can measure low-frequency vibrations (<20&#x2009;Hz) created by vital organs as an acoustic waveform. This is captured using a technology that can be embedded into wearable devices such as in-ear headphones. IH can acquire sound signals that travel within arteries, fluids, bones, and muscles in proximity to the ear canal, allowing for measurements of an individual's unique audiome. We describe the heart rate and heart rhythm results obtained in time-series analysis of the in-ear IH data taken simultaneously with ECG recordings in two dedicated clinical studies. We demonstrate a high correlation (r&#x2009;=&#x2009;0.99) between IH and ECG acquired interbeat interval and heart rate measurements and show that IH can continuously monitor physiological changes in heart rate induced by various breathing exercises. We also show that IH can differentiate between atrial fibrillation and sinus rhythm with performance similar to ECG. The results represent a demonstration of IH capabilities to deliver accurate heart rate and heart rhythm measurements comparable to ECG, in a wearable form factor. The development of IH shows promise for monitoring acoustic imprints of the human body that will enable new real-time applications in cardiovascular health that are continuous and noninvasive.

A cross-sectional observational study involved the analysis of computed tomography (CT) scan data from 125 Indian subjects of 18 years or older with normal imaging findings. Scans were obtained from patients with head injuries as a part of the screening process along with brain CT scans. To establish the dimensions of lateral masses of the atlas vertebrae in normal disease-free Indian individuals. Lateral mass fixation has become the standard of care in fixation of the supra-axial cervical spine. Many studies have investigated the dimensions of lateral masses in cadaveric specimens; however, studies involving the radiological morphometric analysis of the lateral masses of the atlas vertebra in living patients are lacking. Subjects underwent craniovertebral junction CT scans during evaluations of head injury. All had normal radiology reports. The CT scans were obtained using a CT Philips Brilliance 64 machine (Philips, Amsterdam, Netherlands) with a slice thickness of 1 mm and then analyzed using Horos software ver. 2.0.2 (Horos Project, Annapolis, MD, USA) on a MacBook. Lateral masses of the atlas vertebrae were found to be larger in males than females and larger on the right than the left side. The angle of permissible medialization was found to be larger on the right side. The analysis of the average dimensions indicated the conventionally described screw positions to be safe. The present study provides information that may help to establish standard dimensions of lateral masses of the atlas vertebrae among the normal Indian population. We demonstrate that there is no significant difference when compared with the Western population. The results presented here will be of use to clinicians as they may inform preoperative planning for lateral mass fixation surgeries.

The aim of this study was to compare a medical-grade PACS (picture archiving and communication system) monitor, a consumer-grade monitor, a laptop computer, and a tablet computer for linear measurements of height and width for specific implant sites in the posterior maxilla and mandible, along with visualization of the associated anatomical structures. Cone beam computed tomography (CBCT) scans were evaluated. The images were reviewed using PACS-LCD monitor, consumer-grade LCD monitor using CB-Works software, a 13&#x2033; MacBook Pro, and an iPad 4 using OsiriX DICOM reader software. The operators had to identify anatomical structures in each display using a 2-point scale. User experience between PACS and iPad was also evaluated by means of a questionnaire. The measurements were very similar for each device. P-values were all greater than 0.05, indicating no significant difference between the monitors for each measurement. The intraoperator reliability was very high. The user experience was similar in each category with the most significant difference regarding the portability where the PACS display received the lowest score and the iPad received the highest score. The iPad with retina display was comparable with the medical-grade monitor, producing similar measurements and image visualization, and thus providing an inexpensive, portable, and reliable screen to analyze CBCT images in the operating room during the implant surgery.

To compare medical students' learning uptake and understanding of vitreoretinal surgeries by watching either 2D or 3D video recordings. Three vitreoretinal procedures (tractional retinal detachment, exposed scleral buckle removal, and four-point scleral fixation of an intraocular lens [TSS]) were recorded simultaneously with a conventional recorder for two-dimensional viewing and a VERION 3D HD system using Sony HVO-1000MD for three-dimensional viewing. Two videos of each surgery, one 2D and the other 3D, were edited to have the same content side by side. One hundred UMass medical students randomly assigned to a 2D group or 3D, then watched corresponding videos on a MacBook. All groups wore BiAL Red-blue 3D glasses and were appropriately randomized. Students filled out questionnaires about surgical steps or anatomical relationships of the pathologies or tissues, and their answers were compared. There was no significant difference in comprehension between the two groups for the extraocular scleral buckle procedure. However, for the intraocular TSS and tractional retinal detachment videos, the 3D group performed better than 2D (P < 0.05) on anatomy comprehension questions. Three-dimensional videos may have value in teaching intraocular ophthalmic surgeries. Surgical procedure steps and basic ocular anatomy may have to be reviewed to ensure maximal teaching efficacy.

The use of mobile devices such as tablets and laptops by students to support their learning is now ubiquitous. The clinical setting is an environment, which lends itself to the use of mobile devices as students are exposed to novel clinical scenarios that may require rapid location of information to address knowledge gaps. It is unknown what preferences students have for these devices and how they are used in the clinical environment. In this study we explored medical students' choices and their use of different devices in their first year of clinical attachments. We sought to evaluate learners' experiences with these devices using a mixed methods approach. All students newly entered into the clinical years were given the option of a MacBook Air or iPad. We surveyed these students using an online survey tool followed by individual semi-structured interviews to explore survey findings in more depth. Students owned a multitude of devices however their preferences were for the 11&#x2009;in. MacBook Air Laptop over the iPad mini. Students made constant use of online information to support their clinical learning, however three major themes emerged from the interview data: connection and devices (diverse personal ownership of technology by students and how this is applied to source educational materials), influence and interaction with patients (use of any device in a clinical setting) and influence and interaction with staff. In general students preferred to use their device in the absence of patients however context had a significant influence. These mobile devices were useful in the clinical setting by allowing access to online educational material. However, the presence of patients, and the behaviour of senior teaching staff significantly influenced their utilisation by students. Understanding the preferences of students for devices and how they use their preferred devices can help inform educational policy and maximise the learning from online educational content.