This systematic review evaluates recent literature from January 2021 to March 2024 on large language model (LLM) applications across diverse medical specialties. Searching PubMed, Web of Science, and Scopus, we included 84 studies. LLMs were applied to tasks such as clinical natural language processing, medical decision support, education, and aiding diagnostic processes. While studies reported benefits such as improved efficiency and, in some specific NLP tasks, high accuracy above 90%, significant challenges persist concerning reliability, ethical implications, and performance consistency, with accuracy in broader diagnostic support applications showing substantial variability, with some as low as 3%. The overall risk of bias in the reviewed literature was considerably low in 72 studies. Key findings highlight a substantial heterogeneity in LLM performance across different medical tasks and contexts, preventing meta-analysis due to a lack of standardized methodologies. Future efforts should prioritize developing domain-specific LLMs using robust medical data and establishing rigorous validation standards to ensure their safe and effective clinical integration. Trial registration: PROSPERO (CRD42024561381).

The lack of a common framework often complicates the process of provider selection and marginal resource allocation decision. The nonlinear relationships among selection criteria greatly impact the decision-making process. The paper address the critical issue by proposing a centralised Quality of Experience (QoE) and Quality of Service (QoS)- CQoES framework. The framework considers customised priority criteria, determine the relative importance of each criterion and intelligently assign relative weights to each criterion. The framework assists the service provider to in decision making for marginal resources. To achieve the objective, we employ the Analytical Hierarchical Process (AHP), Induced OWA (IOWA) operator, Probabilistic OWA (POWA) operator, user-based collaborative filtering method with enhanced top KNN algorithm. The method handles complex nonlinear relationship of the selection criteria. It signifies consumer's customised criteria in relation to other criteria, then reorders inputs based on the ordered-inducing variable. The proposed method smartly unifies the provider's probabilistic information and the attitudinal characteristics for marginal resource allocation. To demonstrate the effectiveness of the approach, we present two scenarios and use a real cloud and other web service dataset. The experimental results show that the proposed system handles the issue of service selection and marginal resource allocation decision.

A cloud service marketplace is the first point for a consumer to discovery, select and possible composition of different services. Although there are some private cloud service marketplaces, such as Microsoft Azure, that allow consumers to search service advertainment belonging to a given vendor. However, due to an increase in the number of cloud service advertisement, a consumer needs to find related services across the worldwide web (WWW). A consumer mostly uses a search engine such as Google, Bing, for the service advertisement discovery. However, these search engines are insufficient in retrieving related cloud services advertainments on time. There is a need for a framework that effectively and efficiently discovery of the related service advertisement for ordinary users. This paper addresses the issue by proposing a user-friendly harvester and a centralised cloud service repository framework. The proposed Centralised Cloud Service Repository (CCSR) framework has two modules -Harvesting as-a-Service (HaaS) and the service repository module. The HaaS module allows users to extract real-time data from the web and make it available to different file format without the need to write any code. The service repository module provides a centralised cloud service repository that enables a consumer for efficient and effective cloud service discovery. We validate and demonstrate the suitability of our framework by comparing its efficiency and feasibility with three widely used open-source harvesters. From the evaluative result, we observe that when we harvest a large number of services advertisements, the HaaS is more efficient compared with the traditional harvesting tools. Our cloud services advertisements dataset is publicly available for future research at: http://cloudmarketregistry.com/cloud-market-registry/home.html.

There is increasing interest in sharing the experience of products and services on the web platform, and social media has opened a way for product and service providers to understand their consumers needs and expectations. This paper explores reviews by cloud consumers that reflect consumers experiences with cloud services. The reviews of around 6,000 cloud service users were analysed using sentiment analysis to identify the attitude of each review, and to determine whether the opinion expressed was positive, negative, or neutral. The analysis used two data mining tools, KNIME and RapidMiner, and the results were compared. We developed four prediction models in this study to predict the sentiment of users reviews. The proposed model is based on four supervised machine learning algorithms: K-Nearest Neighbour (k-NN), Nave Bayes, Random Tree, and Random Forest. The results show that the Random Forest predictions achieve 97.06% accuracy, which makes this model a better prediction model than the other three.

Road damage detection and assessment are crucial components of infrastructure maintenance. However, detecting multiple crack types in a single image remains a challenging limitation, particularly at varying scales. This is due to the lack of road datasets with various types of damage on varying scales. To address this challenge, we introduce: 1) A novel dataset called the Diverse Road Damage Dataset (DRDD) that captures multiple damage types per individual image. 2) A new detection model called RDD4D that exploits Attention4D blocks to enable better feature refinement across multiple scales. The Attention4D module processes feature maps through an attention mechanism combining positional encoding and talking head components to extract both local and global contextual features. In our comprehensive experimental analysis comparing various state-of-the-art models, our enhanced model demonstrates superior performance in detecting large-sized road cracks with an Average Precision (AP) of 0.458 and maintained competitive performance with an overall mAP of 0.446 on our proposed dataset. Moreover, we also provide results on the CrackTinyNet dataset; our model achieved around a 0.21 increase in performance. The code, model weights, dataset, and our results are publicly available at https://github.com/msaqib17/Road_Damage_Detection.