Human activity recognition has become an active research field over the past few years due to its wide application in various fields such as health-care, smart home monitoring, and surveillance. Existing approaches for homes have achieved promising results. Most of these evaluate real-time activities using only sensor activations that precede evaluation time (where decision is made). However, several critical situations, diagnosing people with dementia, "preceding activations" are not always...

This work presents a novel architecture for context-aware interactions within smart environments, leveraging Large Language Models (LLMs) to enhance user experiences. Our system integrates location data obtained through UWB tags and sensor-equipped homes with real-time human activity recognition (HAR) provide comprehensive understanding of context. contextual information is then fed an LLM-powered chatbot, enabling it generate personalised recommendations based on the user's current...

This work presents a comparative study of low cost and invasiveness sensors (plantar pressure inertial measurement units) for classifying cross-country skiing techniques. A dataset was created symmetrical analysis, with data collected from skiers using instrumented insoles that measured plantar pressure, foot angles, acceleration. deep learning model based on CNN LSTM trained various sensor combinations, ranging two specific to full multisensory array per incorporating 4 an unit...

In the context of ambient-assisted living and monitoring activities behaviors, concept smart homes/environments has emerged as a research development field with many examples appearing in range different contexts (universities, centers, hospitals, residences). Recently, lab called UJAmI based on ambient intelligence was created at University Jaén order to provide an environment where solutions assistive technologies could be developed. This paper presents experience developing this within...

The IoT describes a development field where new approaches and trends are in constant change. In this scenario, devices sensors offering higher precision everyday life an increasingly less invasive way. work, we propose the use of spatial-temporal features by means fuzzy logic as general descriptor for heterogeneous sensors. This sensor representation is highly efficient enables with low computing power to develop learning evaluation tasks activity recognition using light classifiers. To...

Cardiac rehabilitation is a key program which significantly reduces the mortality in at-risk patients with ischemic heart disease; however, there lack of accessibility to these programs health centers. To resolve this issue, home-based for cardiac have arisen as potential solution. In work, we present an approach based on new generation wrist-worn devices improved quality rate sensors and applications. Real-time monitoring sessions high-quality clinical guidelines embedded wearable...

Falling is a common issue within the aging population. The immediate detection of fall key to guarantee early and attention avoid other potential immobility risks reduction in recovery time. Video-based approaches for monitoring detection, although being highly accurate, are largely perceived as intrusive if deployed living environments. As an alternative, thermal vision-based methods can be offer more acceptable level privacy. To date, have focused on single-occupancy scenarios, which not...

Cardiac rehabilitation is a key program which significantly decreases mortality rates in high‐risk patients with ischemic heart disease. Due to the huge lack of accessibility such programs at Health Centers, outdoor‐based for cardiac have been proposed as an excellent tool improve Centers. These make use wrist‐worn devices real‐time monitoring sessions based on clinical guidelines. In this way, greater number can fortunately gain access program. However, advantage also means that team has...

Preeclampsia affects from 5% to 14% of all pregnant women and is responsible for about maternal deaths per year in the world. This paper focused on use a decision analysis tool early detection preeclampsia at risk. applies fuzzy linguistic approach implemented wearable device. In order develop this tool, real dataset containing data with high risk health center has been analyzed, methodology two main phases used. Firstly, transformation applied increase interpretability flexibility...

In this work, we detail a methodology based on Convolutional Neural Networks (CNNs) to detect falls from non-invasive thermal vision sensors. First, include an agile data collection label images in order create dataset that describes several cases of single and multiple occupancy. These standing inhabitants target situations with fallen inhabitant. Second, provide augmentation techniques increase the learning capabilities classification reduce configuration time. Third, have defined 3 types...

Fog Computing is an approach involving smart devices. These devices carry out data processing to provide collaborative services reach a common goal, usually, in the cloud. In fog computing paradigm, uncertainty and vagueness are inherent due limitations of compu tational communication capabilities Fuzzy logic protoforms represent powerful tool model compute imprecise presented within fog-computing paradigm. this paper, we present fuzzy cloud-fog based on temporal windows aggregation. The...

The classic models used to predict the behavior of photovoltaic systems, which are based on physical process solar cell, limited defining analytical equation obtain its electrical parameter. In this paper, we evaluate several machine learning nowcast and energy production a (PV) system in conjunction with ambient data provided by IoT environmental devices. We have evaluated estimation output power generation human-crafted features multiple temporal windows deep approaches comparative results...

The recognition of activities daily living is an important research area interest in recent years. process activity aims to recognize the actions one or more people a smart environment, which set sensors has been deployed. Usually, all events produced during each are taken into account develop classification models. However, instant started unknown real environment. Therefore, only most usually used. In this paper, we use statistics determine appropriate length that interval for type...

Nowadays, smart environments (SEs) enable the monitoring of people with physical disabilities by incorporating activity recognition. Thermal cameras are being incorporated as they preserve privacy. Some deep learning (DL) solutions use pose users because it removes external noise. Although there robust DL in visible spectrum (VS), fail thermal domain. Thus, we propose human lite (THPoseLite), a convolutional neural network (CNN) based on MobileNetV2 that extracts from images (TIs). In novel...