Abstract
Health Remote Monitoring Systems (HRMS) offer the ability to address health-care human resource concerns. In developing nations, where pervasive mobile networks and device access are linking people like never before, HRMS are of special relevance. A fundamental aim of this research work is the realization of technological-based solution to triage and follow-up people living with dementias so as to reduce pressure on busy staff while doing this from home so as to avoid all unnecessary visits to hospital facilities, increasingly perceived as dangerous due to COVID-19 but also raising nosocomial infections, raising alerts for abnormal values. Sensing approaches are complemented by advanced predictive models based on Machine Learning (ML) and Artificial Intelligence (AI), thus being able to explore novel ways of demonstrating patient-centered predictive measures. Low-cost IoT devices composing a network of sensors and actuators aggregated to create a digital experience that will be used and exposure to people to simultaneously conduct several tests and obtain health data that can allow screening of early onset dementia and to aid in the follow-up of selected cases. The best ML for predicting AD was logistic regression with an accuracy of 86.9%. This application as demonstrated to be essential for caregivers once they can monitor multiple patients in real-time and actuate when abnormal values occur.

Abstract
The volume of data is growing exponentially and becoming more valuable to organizations that collect it, from e-commerce data, shipping, audio and video logs, text messages, internet search queries, stock market activity, financial transactions, the Internet of Things, and various other sources. The major challenges are related with the way to extract insights from such a rich data environment and whether Deep Learning can be successful with Big Data. To get some insight on these topics, social network data are employed as a case study on how sentiments can affect decisions in stock market environments. In this paper, we propose a generalized Deep Learning-based classification framework for Stock Market Sentiment Analysis. This work comprises the study, the development, and implementation of an automatic classification system based on Deep Learning and the validation of its adequacy and efficiency in any scenario, particularly Stock Market Sentiment Analysis. Distinct datasets and several Deep Learning approaches with different layers and embedded techniques are used, and their performances are evaluated. These developments show how Deep Learning reacts to distinct contexts. The results also give context on how different techniques with different parameter combinations react to certain types of data. Convolution obtained the best results when dealing with complex data inputs, and long short-term layers kept a memory of data, allowing inputs which are not as common to still be considered for decisions. The models that resulted from Stock Market Sentiment Analysis datasets were applied with some success to real-life problems. The best models reached accuracies of 73% in training and 69% in certain test datasets. In a simulation, a model was able to provide a Return on Investment of 4.4%. The results contribute to understanding how to process Big Data efficiently using Deep Learning and specialized hardware techniques.

Abstract
In the digital world, the demand for better interactions between subscribers and companies is growing, creating the need for personalized and individualized experiences. With the exponential growth of email usage over the years, broad flows of campaigns are sent and received by subscribers, which reveals itself to be a problem for both companies and subscribers. In this work, subscribers are segmented by their behaviors and profiles, such as (i) open rates, (ii) click-through rates, (iii) frequency, and (iv) period of interactions with the companies. Different regressions are used: (i) Random Forest Regressor, (ii) Multiple Linear Regression, (iii) K-Neighbors Regressor, and (iv) Support Vector Regressor. All these regressions’ results were aggregated into a final prediction achieved by an ensemble approach, which uses averaging and stacking methods. The use of Long Short-Term Memory is also considered in the presented case. The stacking model obtained the best performance, with an R (Formula presented.) score of 0.91 and a Mean Absolute Error of 0.204. This allows us to estimate the week’s days with a half-day error difference. This work presents promising results for subscriber segmentation based on profile information for predicting the best period for email marketing. In the future, subscribers can be segmented using the Recency, Frequency and Monetary value, the Lifetime Value, or Stream Clustering approaches that allow more personalized and tailored experiences for subscribers. The latter tracks segments over time without costly recalculations and handles continuous streams of new observations without the necessity to recompile the entire model. © 2022 by the authors.

Abstract

Abstract
In the digital world, the demand for better interactions between subscribers and companies is growing, creating the need for personalized and individualized experiences. With the exponential growth of email usage over the years, broad flows of campaigns are sent and received by subscribers, which reveals itself to be a problem for both companies and subscribers. In this work, subscribers are segmented by their behaviors and profiles, such as (i) open rates, (ii) click-through rates, (iii) frequency, and (iv) period of interactions with the companies. Different regressions are used: (i) Random Forest Regressor, (ii) Multiple Linear Regression, (iii) K-Neighbors Regressor, and (iv) Support Vector Regressor. All these regressions' results were aggregated into a final prediction achieved by an ensemble approach, which uses averaging and stacking methods. The use of Long Short-Term Memory is also considered in the presented case. The stacking model obtained the best performance, with an R-2 score of 0.91 and a Mean Absolute Error of 0.204. This allows us to estimate the week's days with a half-day error difference. This work presents promising results for subscriber segmentation based on profile information for predicting the best period for email marketing. In the future, subscribers can be segmented using the Recency, Frequency and Monetary value, the Lifetime Value, or Stream Clustering approaches that allow more personalized and tailored experiences for subscribers. The latter tracks segments over time without costly recalculations and handles continuous streams of new observations without the necessity to recompile the entire model.

Abstract
In the areas ofmachine learning / big data, when collecting data, sometimes too many features may be stored. Some of them may be redundant or irrelevant for the problem to be solved, adding noise to the dataset. Feature selection allows to create a subset from the original feature set, according to certain criteria. By creating a smaller subset of relevant features, it is possible to improve the learning accuracy while reducing the amount of data. This means means better results obtained in a shorter learning time. However, feature selection is normally regarded as a very important problem to be solved, as it directly impacts both data analysis and model creation. The problem of optimizing the selected features of a given dataset is not always trivial but, throughout the years, different ways to counter this optimization problem have been presented. This work presents how feature selection fits in the larger context of multi-objective problems as well as a review of how both multi-objective evolutionary algorithms and metaheuristics are being used in order to solve feature selection problems © MIR Labs, www.mirlabs.net/ijcisim/index.html