Nothing Special   »   [go: up one dir, main page]
Next Article in Journal
TableExtractNet: A Model of Automatic Detection and Recognition of Table Structures from Unstructured Documents
Previous Article in Journal
Why Do People Gather? A Study on Factors Affecting Emotion and Participation in Group Chats
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Informatics
Volume 11
Issue 4
10.3390/informatics11040076
informatics-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Systematic Review

In-Bed Monitoring: A Systematic Review of the Evaluation of In-Bed Movements Through Bed Sensors

by
Honoria Ocagli
,
Corrado Lanera
,
Carlotta Borghini
,
Noor Muhammad Khan
,
Alessandra Casamento
and
Dario Gregori
*
Unit of Biostatistics, Epidemiology and Public Health, Department of Cardiac, Thoracic and Vascular Sciences, University of Padova, Via Loredan, 18, 35121 Padova, Italy
*
Author to whom correspondence should be addressed.
Informatics 2024, 11(4), 76; https://doi.org/10.3390/informatics11040076
Submission received: 1 July 2024 / Revised: 26 September 2024 / Accepted: 1 October 2024 / Published: 22 October 2024
Browse Figures
Figure 1
<p>PRISMA flow chart.</p> "> Figure 2
<p>Type of bed sensors (columns) and their positions (colors). In the y-axis are reported the number of studies, while the x-axis categorizes the types of sensors.</p> "> Figure 3
<p>Bubble chart of the mean minimal accuracy (color) of the best model divided by outcome (vital signs, sleep, multiple output data, in-bed pose estimation, bedsores) and type of input data (acceleration data, multiple input data, other, pressure data, pressure image/map, vital sign data).</p> "> Figure 4
<p>Preprocessing techniques in in-bed patient monitoring studies with machine learning approach.</p> "> Figure 5
<p>Roadmap for Future Smart-Bed Technologies.</p> ">
Review Reports Versions Notes

Abstract

:
The growing popularity of smart beds and devices for remote healthcare monitoring is based on advances in artificial intelligence (AI) applications. This systematic review aims to evaluate and synthesize the growing literature on the use of machine learning (ML) techniques to characterize patient in-bed movements and bedsore development. This review is conducted according to the principles of PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) and is registered in the International Prospective Register of Systematic Reviews (PROSPERO CRD42022314329). The search was performed through nine scientific databases. The review included 78 articles, including 142 ML models. The applied ML models revealed significant heterogeneity in the various methodologies used to identify and classify patient behaviors and postures. The assortment of ML models encompassed artificial neural networks, deep learning architectures, and multimodal sensor integration approaches. This review shows that the models for analyzing and interpreting in-bed movements perform well in experimental settings. Large-scale real-life studies are lacking in diverse patient populations.

1. Introduction

Recently, there has been an increase in attention to patient-centric approaches in the healthcare environment, primarily through remote health monitoring [1]. Remote health monitoring allows healthcare personnel to monitor patient health conditions and physiological signs in real time [2]. In addition, electric medical beds have been developed in recent years due to technological progress [3]. These beds are characterized by various features, including alarms, exit sensors in the bed, and integrated accessory controls [3]. They are the so-called “smart beds”, where this label describes the interconnection between materials, design and functionality, and user interfaces [3]. Smart beds provide continuous data that can be integrated into the health system to identify risk patterns [4]. Smart beds have been used to prevent episodes such as agitation [5], falls among elderly patients [6], pressure ulcers [7], and sleep disorders [8,9]. Smart beds and their interface offer an opportunity to assess three key objectives, as Karvounis et al. [4] reported: patient monitoring, prevention of falls, and prevention of pressure ulcers.
The growing popularity of smart beds and devices for remote healthcare monitoring is based on the advancement of artificial intelligence (AI) applications [10]. Just lying in a smart bed could provide a significant amount of information. Smart beds are widely used in specific hospital settings, including intensive care units. They have recently found applications in long-term care units and nursing homes to prevent bedsore-related complications [11]. Furthermore, sensors placed in a hospital bed could inform about the position of the patient in the bed [12]. Knowing and consequently anticipating the position of a patient in the bed has been proven helpful in detecting falls in at-risk populations, particularly older people [12]. The large amount of data from these modern devices is dependent on AI support, as traditional database technology cannot store and analyze them [10]. Machine learning techniques are known to be especially suitable for performing data processing tasks, given training examples [13]. Various reviews in the literature on machine learning techniques (ML) and the prevention of pressure injury have shown that ML techniques are widely implemented with promising results [14,15,16].
This systematic review aims to evaluate and synthesize the growing literature on the use of ML techniques to characterize patient movements in bed.

2. Materials and Methods

This review is reported according to the statement of PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [17]. It is registered in the International Prospective Register of Systematic Reviews (PROSPERO registration no. CRD42022314329).

2.1. Information Sources and Search Strategy

The search was carried out through the following databases: MEDLINE through PubMed, EMBASE (through Ovid), CINAHL, Scopus, Cochrane, IEEX, arXiv, ACM digital library, and Web of Science. The string search was updated on the seventh of September 2024. The words related to the concept of positions in bed and the other outcomes, bedsore development, and sleep quality, related to the terms related to machine learning techniques and the term “bed”. The three groups of concepts were linked with the Boolean operator “AND”. Table S1 shows the applied search strategy.

2.2. Eligibility Criteria

The eligibility criteria were based on the PICO (Population, Intervention, Comparison, Outcome) acronym: Participants, Healthy/non-healthy, volunteer/patients; Intervention/exposure; use of smart beds, sensors in the bed; Comparator, wearable devices; Outcome, movement patterns.
Articles with the following characteristics were included in this systematic review: (1) studies in which bed sensors measured parameters such as movements, heart rate, sleep patterns, and pressure ulcers; (2) studies that used any ML technique; (3) studies that used wearable devices based on accelerometry to evaluate movements. Articles (1) in languages other than English, publications (2) with populations other than adults (e.g., animals, children), or (3) letters, commentaries, or reviews without relevant data were excluded.

2.3. Selection Process

The selection of the title/abstract and the full text was performed by two independent reviewers (CB and AC). Any conflicts were resolved through discussion or the judgment of an experienced third-party researcher (HO and CL). The title/abstract and full text were reviewed using the Rayyan free Web tool [18]. Duplicates were manually removed using Rayyan.

2.4. Data Extraction

Data were extracted from all selected publications using a predefined format created in a Microsoft Excel worksheet. The two investigators (CB and AC) independently performed data extraction; they compared the results and resolved any conflict through discussion or involvement of a third investigator (CL or HO).
The following study information was collected: characteristics of the study, authors, year, journal of publication, study design (observational, longitudinal, and simulated designs), setting (home, controlled environment, hospital, and nursing home), experiment duration, and outcome type (in-bed pose estimation, multiple output data, and other specific health indicators); characteristics of the population (volunteers, healthy volunteers, patients), number of participants, number of male and female participants, and age (mean, range); characteristics of the bed, sensor type (bed sensors, wearable devices, arrays of pressure sensors), number of load cells, bed sensor placement (embedded, outside the bed structure), list of in-bed positions (supine, prone, left lateral, right lateral, and other), input data (acceleration data, pressure data, pressure images/maps, and vital sign data), and input data pre-processing; characteristics of the ML techniques applied, ML category (deep learning, shallow learning, and combinations of both), models used (clustering, supervised learning algorithms, and other specific machine learning techniques), and performance metrics (accuracy, F1 score, sensitivity, specificity, and precision).

2.5. Assessment of Risk of Bias

In assessing the quality of the included studies, we utilized the Prediction Model Risk of Bias Assessment Tool (PROBAST) [19]. The tool is structured into four key domains: participants, predictors, outcome, and analysis. Each domain contains multiple signaling questions to guide the assessment. Reviewers use these questions to classify each study as having a ”low”, “high”, or “unclear” risk of bias. Despite some limitations in its applicability to our study design, using PROBAST provided a consistent framework for evaluating study quality across the included studies.

3. Results

This review followed the PRISMA process [20] to select the included articles, as presented in the flow diagram in Figure 1. The search in the various databases retrieved 6009 publications. After checking for duplicates, 4145 items were screened based on title and abstract. This screening led to 3060 studies being removed due to their ineligibility (for example, another outcome), and 78 articles were selected for this review.

3.1. Descriptive Characteristics of the Study

Table 1 reports the descriptive characteristics of the articles included in this review. The 78 articles included in our study were published in journals (41, 52%) or conference collections (37, 48%) after 2018 (53, 68%)—or before it (25, 32%)—and comprised a total of 5647 participants. In many of the studies (65, 75.6%), the data were collected in a controlled environment (such as a laboratory under supervision). For the remaining research, instead, data sets were generated from a variety of settings, including hospitals (8, 9.3%), nursing homes (5, 5.8%), and homes (7, 8.1%).
Regarding the participants enrolled in the included studies, in 48 (60.8%) articles, healthy adult volunteers were recruited; similarly, 15 (19%) articles involved volunteers. Furthermore, only five articles involved older people (6.3%), while eleven studies (14%) focused on patients. In the latter case, the diseases considered were Parkinson’s [32], sleep disorders [45], obstructive syndrome apnea [29], deteriorated cognitive function [73], atrial fibrillation [43], and heart disease [72]. The age of the participants ranged between 19 [41,51,53,54,57,58] and 99 years [72]. Eleven studies [22,34,35,36,37,38,39,40,41,42] considered publicly available datasets or simulated data.

3.2. Characteristics of the Collected Data

Experiments in this review lasted less than 0.5 h in 14 studies [21,22,23,24,28,35,36,38,40,44,50,52,55,60] or up to months when investigating sleeping patterns [71,72]. The input data were classified into kinetic data, pressure data, pressure image or map, and vital sign data (Table 2, Table S2). As expected, before analysis, these input data underwent a preprocessing phase (70, 90%), which, in most models, involved signal manipulation (46, 58%), feature extraction (24, 31%), and analysis of the principal component (15, 19%) (Table S3).

3.3. Smart Bed Characteristics

The participant data were acquired from wearable devices (11, 13.4%) and bed sensors (67, 81.7%), and four of them from both (4.8%). The study utilized various sensor technologies to monitor pressure distribution and related parameters. The majority of the devices used were load cells, accounting for 32.1% of the cases, followed closely by arrays of pressure sensors, representing 27.2% (Figure 2). The load cells are specialized transducers that convert mechanical force or weight into an electrical signal. In the context of bed monitoring, load cells are often integrated into the bed frame or mattress to measure changes in weight distribution and provide quantitative data on patient movements, postural changes, and restlessness during sleep.
These pressure sensors consist of an array of pressure-sensitive elements distributed across the bed’s surface. They function by detecting changes in pressure distribution when a patient lies on the bed, providing valuable information about the patient’s movements and positioning during sleep or rest. A small percentage of studies used multiple sensors (6.2%) and EMFi sensors (3.7%), highlighting some diversity in sensor choice. Additionally, a few studies employed specific devices like hydraulic bed sensors (3.7%) and piezoelectric sensors (2.5%) (Table 1).
The number of sensors that made up the arrays was between 15 [25] and 3560 [27], with 2048 being the most common size [50,51,53,54,57,58,62,75]. Hydraulic sensors use fluid-filled chambers or tubes to measure pressure variations, providing information on patient movements and body positioning. Pad sensors typically consist of pressure-sensitive pads or mats placed on the bed surface to detect pressure distribution and movement patterns. Oscillosensors use oscillatory motion or vibration detection mechanisms to monitor patient movements and sleep disturbances.
Most sensors were embedded into a mattress (25, 30.9%), followed by those integrated into the bed structure (18, 22.2%). Some were placed outside the mattress and bed structure (13, 16.0%), while others were positioned outside the mattress (6, 7.4%) or under the mattress (6, 7.4%). A few studies involved multiple placements (2, 2.5%). Notably, eight entries (9.9%) had missing information on sensor placement (Figure 2).

3.4. In-Bed Movements

In 53 articles (68%), participants were instructed to perform a set of movements or acquire specific positions while lying in bed. The most popular postures detected were supine (47, 60%), prone (21, 27%), and lateral (47, 60%) (Table S4). Regarding the positions, we categorize the primary static ones (prone, supine, and lateral). When using the term “movements”, we refer to all distinct types of movement considered by the authors in the articles, whether static or dynamic. The in-bed movements, as considered by authors, ranged from three [25,31] to twenty-eight [41]. Postures such as supine (lying on the back), lateral (lying on the side), prone (lying face down), and various transitional movements between these positions contribute to the diversity of in-bed movements captured by the sensors. The differences in in-bed movements between different postures can be attributed to several factors, including changes in weight distribution, pressure points, and muscle activation patterns associated with each posture. For example, in the supine position, movements can primarily involve changes in body position and adjustments for comfort, whereas, in the lateral or prone positions, movements can include rolling, turning, or repositioning to alleviate pressure or discomfort.

3.5. Machine Learning Techniques Applied

Among the 78 articles, we considered a model for each outcome and MLT for 142 models (Table S4). Deep learning models were used primarily featuring CNNs for in-bed position estimation. Shallow models appeared in seven cases (32%), mainly using SVM, KNN, and decision trees, often paired with feature extraction and clustering methods. For sleep outcome, deep learning models (e.g., CNNs) were used in five cases (56%) with preprocessing steps such as SMOTE and signal manipulation. Vital sign monitoring featured in five cases (13%), where shallow models dominated with four instances (80%), including Random Forests, Naive Bayes, and clustering techniques like KMCA and FCM. Table S3 describes the characteristics of the models included and their reported performance.
Accuracy was used as an evaluation indicator to assess the performance of the ML model; we considered the accuracy of the number of correctly identified positions. Pressure images and pressure data are the most frequently used types of input data for in-bed monitoring applications, particularly in pose estimation and bedsore prevention (Figure 3). Pressure data and pressure image/map data are associated with higher accuracy rates in detecting and predicting various outcomes. The varying sizes and colors of the bubbles further illustrate the different levels of accuracy achieved across various applications, with warmer colors indicating higher accuracy.
The best-performing models in each article exhibited an accuracy range from 31% (as seen in the AdaLSTM in Alinia et al. [22]) to 100% (Feed Forward Neural Network in Davoodnia et al. [36] and RF in Youngkong et al. [101]). Globally, 36 models (34%) reached an accuracy of more than 95%.
Figure 4 shows if the models have used pre-processing techniques or not. The bar chart highlights that the majority of studies, regardless of the machine learning approach used, apply preprocessing techniques. Specifically, thirty-six of the shallow learning studies (51.4%), twenty-five deep learning studies (35.7%), and nine of the combined shallow and deep learning studies (12.8%) utilized preprocessing. This underscores the importance of preprocessing in improving the performance of in-bed patient monitoring systems.
In Table S3, the information on pre-processing is reported. The most commonly used preprocessing methods were feature extraction methods, accounting for 20 occurrences (28.57%). This was followed closely by signal manipulation methods with 23 occurrences (32.86%). Principal Component Analysis (PCA) was used in 12 instances (17.14%). Additionally, image processing techniques appeared three times (4.29%), while data cleaning and augmentation were employed in three cases (4.29%). Other methods comprised six instances (8.57%).

3.6. Risk of Bias Assessment

Overall, about 40% of the studies are rated as “Unclear”, indicating uncertainty in their overall quality, while around 35% are rated as “High”, suggesting notable concerns. Only about 25% have a “Low” rating, reflecting a lower risk of bias in their overall assessment. This distribution suggests that many studies may have methodological limitations affecting their overall reliability.

4. Discussion

This review focuses on the use of sensor devices and bed sensors to primarily classify movements in the bed, the onset of bedsores, or sleep quality. In detail, we focus on the types of ML models implemented so far in the field.
ML models were classified into deep learning models and shallow models. As suggested by Chollet et Allaire, shallow learning models use one or two hidden layers [13] or are non-neural-network ML models (e.g., SVM or RF). In contrast, deep learning considers numerous hidden layers [78]. In this present review, we show that deep learning and shallow learning were used to detect outcomes in the bed, which is the main outcome considered in this work. Among shallow learning models, the best models achieve the best accuracy, ranging from 71.95% to 99.93% for different outcomes. DL models also demonstrate competitive performance, with the best accuracy ranging from 31.05% to 100%. Similarly to other studies [103], we observed higher accuracy rates in controlled simulated settings (up to 95.43%) compared to real-world clinical environments (around 86.80%). This discrepancy emphasizes the challenge of achieving high detection accuracy in practical applications, a point that is not always fully addressed in the existing literature. Among deep learning models, a common approach used in these studies is the application of artificial neural networks (ANNs). For example, Chica et al. [21] propose a real-time recognition system that uses ANNs to classify patient intentions based on sequences of pressure maps. Similarly, Hu et al. [26] present an ANN-based model to classify posture in the bed using pressure sensors on the bed sheets. ANNs allow these models to learn and recognize patterns in the pressure data, enabling accurate classification of different postures or behaviors.
Another approach mentioned in the references is the use of deep learning models. Deep residual networks [22] and deep learning [23] are used for sleep posture recognition and sleep stage classification, respectively. Deep learning models, particularly convolutional neural networks (CNNs) and recurrent neural networks (RNNs), have successfully processed sequential data, making them suitable for analyzing time-series data collected from sensors.
In addition, some studies combine different sensor modalities and machine learning techniques to enhance the accuracy of posture recognition and behavior detection. For example, in the study by Pupic et al. [31], they considered the bed reaction and the skin–bed interface forces to detect the position of the patient and monitor the skin pressure to prevent pressure injury.
Both deep and shallow learning models offer distinct advantages: deep models excel in complex pattern recognition with large datasets, while shallow models are more feasible for real-time, resource-limited clinical settings. Combining both approaches can optimize patient monitoring and outcomes.
The studies considered in this review use different types of sensors, such as pressure sensors, bed sensors, load cells, and wearable devices. Several studies used pressure-sensitive sensors attached to the bed or patients’ clothing to detect sleep posture, bed-exit intention, and in-bed postural changes. Chica et al. [21] developed an artificial neural network (ANN) that can recognize patient intentions from sequences of pressure maps in real time. Kuwahara and Wada [27] developed a deep learning algorithm that predicts bed leave using a pressure-sensitive sheet-type sensor base. Similarly, Gargees et al. [44] used a hydraulic bed sensor to classify the stages of sleep using deep learning. Hsiao et al. [25] developed a body posture recognition and turning recording system for bedbound patients. Matar et al. [28] used an ANN to classify posture in the bed using bedsheet pressure sensors. Hu et al. [26] developed a patient-specific real-time sleep posture recognition system using a pressure-sensitive conductive sheet and transfer learning. Furthermore, Metsis et al. [38] and Walsh et al. [33] used bed pressure mats to recognize sleep patterns using machine learning. Finally, some studies used sensors to monitor pressure changes to prevent pressure ulcers and sleep apnea. Monroy et al. [30] used inertial sensors attached to clothing to monitor postural changes in the bed to avoid pressure ulcers. Pupic et al. [31] used bed reaction forces and monitored skin–bed interface forces for the prevention and management of pressure injuries. Waltisberg et al. [8] used non-contact pressure-based sensors to discriminate between sleep and wakefulness.
The market for in-bed patient monitoring has seen some commercialized products, such as smart beds and wearable devices integrated with sensor technologies. These products are designed to monitor vital signs, detect exits from the bed, and assess sleep patterns, improving patient care and safety. However, widespread adoption is hindered by several factors. A significant challenge is that most studies have been conducted on prototypes rather than on real hospital beds, which limits the generalizability of the findings to actual clinical settings. Additionally, high costs, technological limitations, regulatory hurdles, data privacy concerns, and the need for user acceptance and training further impede commercialization. Overcoming these challenges requires increased funding, collaborative efforts, standardization of protocols, advancements in technology, regulatory support, and public engagement to build trust and facilitate broader implementation. Large-scale real-life studies across diverse patient populations are still lacking due to logistical challenges, high costs, and ethical and privacy concerns. To improve this scenario, several steps can be taken: increased funding and resource allocation, fostering collaborations between institutions and countries, developing standardized protocols for data collection and analysis, leveraging advanced technologies like electronic health records (EHRs) and AI, and ensuring robust data protection measures to build public trust.
The range of movements considered in the included studies varied significantly, from three to seventeen distinct dynamics. This level of detail in movement categorization is more comprehensive compared to previous reviews, which often did not differentiate as clearly between static and dynamic movements [103]. This variability suggests a lack of standardization in the categorization and measurement of movements, which could impact the comparability of results across studies. Additionally, the ability of sensors to accurately detect and classify different postures and movements may differ, indicating the need to discuss the precision and limitations of the various sensor types used. Movements associated with different postures, such as supine versus lateral, have different characteristics due to differences in weight distribution, pressure points, and muscle activation. This aspect warrants further investigation to understand how well sensors capture these nuances. Furthermore, movements made for comfort adjustments versus those made to alleviate discomfort or pressure could provide insight into patient well-being and the effectiveness of sensors. Understanding these distinctions could enhance the interpretation of data collected from monitoring in the bed. Finally, discussing the implications of these findings for clinical practice is crucial. For instance, understanding typical movements in different positions could help design better intervention strategies for bedridden patients, improving patient care.
This review, which includes a variety of study designs, highlights the significant efforts made to predict events affecting patient outcomes. However, only nine studies were conducted on actual patients, and these involved relatively small sample sizes. Studies in controlled environments often do not account for the variability in movements and conditions present in real hospital settings, such as the use of pillows or the presence of bed rails. Additionally, we identified significant heterogeneity in the machine learning (ML) models used across studies, with notable differences in input data types, preprocessing methods, and model architectures. This variability makes it challenging to compare results between studies, as differences in sensor types, data granularity, and movement categorizations lead to inconsistent findings. For example, models trained in controlled environments with healthy volunteers may not generalize well to clinical settings, where patient conditions and behaviors are more diverse. Furthermore, the lack of standardized performance metrics complicates the evaluation of model accuracy and limits the ability to determine which models are truly robust across different populations. To address these issues, future research should prioritize standardized data collection, preprocessing, and reporting practices, along with the development of benchmark datasets that reflect real-world variability. Standardization would facilitate the comparison of results and improve the generalizability of findings across diverse clinical environments.
This review, considering that it includes all kinds of studies, shows that there is a great effort to try to predict events that can affect patient outcomes. However, there are only nine studies conducted in actual patients, involving only a few patients. Studies in a controlled environment do not consider the same variability in movements available in a real-hospital setting. Most studies must provide information on the use of a pillow and the presence/absence of bed rails.

5. Conclusions

This review highlights significant heterogeneity in the diverse range of ML models used to recognize and classify patient behaviors and postures. These models include artificial neural networks, deep learning architectures, and approaches that combine different sensor modalities. Each model has advantages and may be more suitable for specific applications based on the available data and the desired classification tasks.
Looking towards the future, in-bed patient monitoring through bed sensors holds promising prospects. Trends indicate a growing integration of advanced machine learning algorithms, both deep and shallow, to enhance accuracy and functionality. Continuous improvements in sensor technology and data processing capabilities will lead to more sophisticated monitoring systems that can provide real-time insight into patient movements and vital signs.
Overall, the studies demonstrate the potential of sensor technology and machine learning algorithms to monitor patient sleep posture, intention to leave the bed, postural changes in the bed, sleep patterns, and prevention of pressure ulcers and sleep apnea. The generalizability of the findings is limited by the fact that many studies were conducted in controlled environments or with small sample sizes. Furthermore, few studies involved real-world clinical settings, which restricts the applicability of the findings to broader patient populations. Future research should aim to validate these findings in more diverse, real-world settings.
However, several critical challenges must be addressed to realize these advancements. The high cost of these technologies remains a significant barrier to widespread adoption. Efforts to reduce production costs and develop cost-effective solutions will be crucial. Additionally, the integration of these systems with existing hospital infrastructure poses technical challenges that require robust, interoperable designs.
To address these issues, a structured roadmap can guide future developments (Figure 5). Establishing standardized protocols for data collection and utilizing high-resolution, unobtrusive sensors can enhance accuracy and patient comfort, ensuring more reliable monitoring systems. Developing and refining machine learning algorithms that balance accuracy and computational efficiency is essential. These algorithms should be optimized for real-time processing and low-power consumption, making them practical for various settings. Implementing robust data encryption and privacy-preserving techniques is necessary to protect patient data and comply with regulations. Furthermore, extensive validation studies in clinical environments will ensure the effectiveness and reliability of these technologies. Fostering interdisciplinary collaboration between engineers, healthcare professionals, and researchers will help develop holistic solutions. Providing comprehensive training for healthcare providers and patients is also crucial to maximizing the benefits of smart-bed technologies. Ethical and privacy concerns surrounding patient data need to be rigorously addressed to build trust and compliance with data protection regulations. Engaging healthcare providers and patients through education and training will be essential to ensure acceptance and the proper use of these technologies.
Ethical and privacy concerns surrounding patient data also need to be rigorously addressed to build trust and compliance with data protection regulations. Engaging healthcare providers and patients through education and training will be essential to ensure acceptance and proper use of these technologies.
In conclusion, while the potential for in-bed patient monitoring technologies is substantial, overcoming these barriers will require coordinated efforts from researchers, industry stakeholders, regulatory bodies, and healthcare providers. By addressing these challenges, we can pave the way for more personalized and proactive healthcare solutions that significantly improve patient care and outcomes.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/informatics11040076/s1, Table S1: Search strategy for PubMed, Table S2: Summary of the 79 studies included according to the outcome, Table S3: Characteristics of the machine learning models, Table S4: Characteristics of the study considering smart beds.

Author Contributions

D.G., C.L., and H.O. conceived the study. C.L. and H.O. defined the search string. H.O. researched the literature. C.B., A.C., and N.M.K. conducted the paper screening, while H.O. and C.L. discussed and resolved all the screening conflicts. H.O., C.L., C.B., A.C., and N.M.K. defined and composed the tables and wrote the first draft of the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the SMACT (Repertorio 157/2021 Prot. 3001 07/09/2021).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available on request from the corresponding author, as most of the information is already provided in this article.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Kristoffersson, A.; Lindén, M. A Systematic Review of Wearable Sensors for Monitoring Physical Activity. Sensors 2022, 22, 573. [Google Scholar] [CrossRef] [PubMed]
  2. Majumder, S.; Mondal, T.; Deen, M.J. Wearable Sensors for Remote Health Monitoring. Sensors 2017, 17, 130. [Google Scholar] [CrossRef] [PubMed]
  3. Ghersi, I.; Mariño, M.; Miralles, M.T. Smart Medical Beds in Patient-Care Environments of the Twenty-First Century: A State-of-Art Survey. BMC Med. Inform. Decis. Mak. 2018, 18, 63. [Google Scholar] [CrossRef] [PubMed]
  4. Karvounis, E.; Polymeni, S.; Tsipouras, M.; Koritsoglou, K.; Tzovaras, D. Smart Beds and Bedding Surfaces for Personalized Patient Care: A Review. In Proceedings of the 2021 6th South-East Europe Design Automation, Computer Engineering, Computer Networks and Social Media Conference (SEEDA-CECNSM), Preveza, Greece, 24–26 September 2021; pp. 1–8. [Google Scholar]
  5. Au-Yeung, W.-T.M.; Miller, L.; Dodge, H.H.; Reynolds, C.; Beattie, Z.; Vahia, I.; Kaye, J. Sensing a Problem: Proof of Concept for Characterizing and Predicting Agitation. Alzheimer’s Dement. Transl. Res. Clin. Interv. 2020, 6, e12079. [Google Scholar] [CrossRef] [PubMed]
  6. Subermaniam, K.; Welfred, R.; Subramanian, P.; Chinna, K.; Ibrahim, F.; Mohktar, M.S.; Tan, M.P. The Effectiveness of a Wireless Modular Bed Absence Sensor Device for Fall Prevention among Older Inpatients. Front. Public Health 2017, 4, 292. [Google Scholar] [CrossRef]
  7. Sakai, K.; Sanada, H.; Matsui, N.; Nakagami, G.; Sugama, J.; Komiyama, C.; Yahagi, N. Continuous Monitoring of Interface Pressure Distribution in Intensive Care Patients for Pressure Ulcer Prevention. J. Adv. Nurs. 2009, 65, 809–817. [Google Scholar] [CrossRef]
  8. Waltisberg, D.; Amft, O.; Brunner, D.P.; Troster, G. Detecting Disordered Breathing and Limb Movement Using In-Bed Force Sensors. IEEE J. Biomed. Health Inform. 2017, 21, 930–938. [Google Scholar] [CrossRef]
  9. Pino, E.J.; Dörner De la Paz, A.; Aqueveque, P.; Chávez, J.A.P.; Morán, A.A. Contact pressure monitoring device for sleep studies. In Proceedings of the 2013 35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Osaka, Japan, 3–7 July 2013; pp. 4160–4163. [Google Scholar] [CrossRef]
  10. Johnson, K.B.; Wei, W.; Weeraratne, D.; Frisse, M.E.; Misulis, K.; Rhee, K.; Zhao, J.; Snowdon, J.L. Precision Medicine, AI, and the Future of Personalized Health Care. Clin. Transl. Sci. 2021, 14, 86–93. [Google Scholar] [CrossRef]
  11. Ajami, S.; Khaleghi, L. A Review on Equipped Hospital Beds with Wireless Sensor Networks for Reducing Bedsores. J. Res. Med. Sci. 2015, 20, 1007–1015. [Google Scholar] [CrossRef]
  12. Viriyavit, W.; Sornlertlamvanich, V. Bed Position Classification by a Neural Network and Bayesian Network Using Noninvasive Sensors for Fall Prevention. J. Sens. 2020, 2020, e5689860. [Google Scholar] [CrossRef]
  13. Chollet, F.; Allaire, J.J. Deep Learning with R; Manning Publications: New York, NY, USA, 2018. [Google Scholar]
  14. Jiang, M.; Ma, Y.; Guo, S.; Jin, L.; Lv, L.; Han, L.; An, N. Using Machine Learning Technologies in Pressure Injury Management: Systematic Review. JMIR Med. Inform. 2021, 9, e25704. [Google Scholar] [CrossRef] [PubMed]
  15. Marchione, F.G.; Araújo, L.M.Q.; Araújo, L.V. Approaches That Use Software to Support the Prevention of Pressure Ulcer: A Systematic Review. Int. J. Med. Inform. 2015, 84, 725–736. [Google Scholar] [CrossRef] [PubMed]
  16. Ribeiro, F.; Fidalgo, F.; Silva, A.; Metrôlho, J.; Santos, O.; Dionisio, R. Literature Review of Machine-Learning Algorithms for Pressure Ulcer Prevention: Challenges and Opportunities. Informatics 2021, 8, 76. [Google Scholar] [CrossRef]
  17. Page, M.J.; McKenzie, J.E.; Bossuyt, P.M.; Boutron, I.; Hoffmann, T.C.; Mulrow, C.D.; Shamseer, L.; Tetzlaff, J.M.; Akl, E.A.; Brennan, S.E.; et al. The PRISMA 2020 Statement: An Updated Guideline for Reporting Systematic Reviews. BMJ 2021, 372, n71. [Google Scholar] [CrossRef]
  18. Ouzzani, M.; Hammady, H.; Fedorowicz, Z.; Elmagarmid, A. Rayyan—A Web and Mobile App for Systematic Reviews. Syst. Rev. 2016, 5, 210. [Google Scholar] [CrossRef]
  19. Moons, K.G.M.; Wolff, R.F.; Riley, R.D.; Whiting, P.F.; Westwood, M.; Collins, G.S.; Reitsma, J.B.; Kleijnen, J.; Mallett, S. PROBAST: A Tool to Assess Risk of Bias and Applicability of Prediction Model Studies: Explanation and Elaboration. Ann. Intern. Med. 2019, 170, W1–W33. [Google Scholar] [CrossRef]
  20. Moher, D.; Liberati, A.; Tetzlaff, J.; Altman, D.G.; PRISMA Group. Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. Ann. Intern. Med. 2009, 151, 264–269. [Google Scholar] [CrossRef]
  21. Albukhari, A.; Lima, F.; Mescheder, U. Bed-Embedded Heart and Respiration Rates Detection by Longitudinal Ballistocardiography and Pattern Recognition. Sensors 2019, 19, 1451. [Google Scholar] [CrossRef]
  22. Alinia, P.; Samadani, A.; Milosevic, M.; Ghasemzadeh, H.; Parvaneh, S. Pervasive Lying Posture Tracking. Sensors 2020, 20, 5953. [Google Scholar] [CrossRef]
  23. Olguın, D.O.; Pentland, A.S. Human Activity Recognition: Accuracy across Common Locations for Wearable Sensors. In Proceedings of the 2006 10th IEEE International Symposium on Wearable Computers, Montreux, Switzerland, 11–14 October 2006; pp. 11–14. [Google Scholar]
  24. Altun, K.; Barshan, B.; Tunçel, O. Comparative Study on Classifying Human Activities with Miniature Inertial and Magnetic Sensors. Pattern Recognit. 2010, 43, 3605–3620. [Google Scholar] [CrossRef]
  25. Arora, A.; Chakraborty, P.; Bhatia, M. Analysis of Data from Wearable Sensors for Sleep Quality Estimation and Prediction Using Deep Learning. Arab. J. Sci. Eng. 2020, 45, 10793–10812. [Google Scholar] [CrossRef]
  26. de Arriba-Pérez, F.; Caeiro-Rodríguez, M.; Santos-Gago, J.M. Collection and Processing of Data from Wrist Wearable Devices in Heterogeneous and Multiple-User Scenarios. Sensors 2016, 16, 1538. [Google Scholar] [CrossRef] [PubMed]
  27. Azimi, H.; Bouchard, M.; Goubran, R.; Knoefel, F. Unobtrusive Screening of Central Sleep Apnea From Pressure Sensors Measurements: A Patient-Specific Longitudinal Study. IEEE Trans. Instrum. Meas. 2020, 69, 3282–3296. [Google Scholar] [CrossRef]
  28. Bai, D.; Ho, M.C.; Mathunjwa, B.M.; Hsu, Y.L. Deriving Multiple-Layer Information from a Motion-Sensing Mattress for Precision Care. Sensors 2023, 23, 1736. [Google Scholar] [CrossRef]
  29. Breuss, A.; Gnarra, O.; Fujs, M.; Schmidt, M.; Riener, R. Sleep Position Classification with a Compact Pressure Textile Sensor Mattress Using Convolutional Neural Networks. Biomed. Signal Process. Control 2024, 90, 105775. [Google Scholar] [CrossRef]
  30. Bruser, C.; Diesel, J.; Zink, M.; Winter, S.; Schauerte, P.; Leonhardt, S. Automatic Detection of Atrial Fibrillation in Cardiac Vibration Signals. IEEE J. Biomed. Health Inform. 2013, 17, 162–171. [Google Scholar] [CrossRef]
  31. Casas, L.; Navab, N.; Demirci, S. Patient 3D Body Pose Estimation from Pressure Imaging. Int. J. Comput. Assist. Radiol. Surg. 2019, 14, 517–524. [Google Scholar] [CrossRef]
  32. Chica, M.; Campoy, P.; Pérez, M.A.; Rodríguez, T.; Rodríguez, R.; Valdemoros, Ó. Real-Time Recognition of Patient Intentions from Sequences of Pressure Maps Using Artificial Neural Networks. Comput. Biol. Med. 2012, 42, 364–375. [Google Scholar] [CrossRef]
  33. Cho, T.; Sunarya, U.; Yeo, M.; Hwang, B.; Koo, Y.; Park, C. Deep-ACTINet: End-to-End Deep Learning Architecture for Automatic Sleep-Wake Detection Using Wrist Actigraphy. Electronics 2019, 8, 1461. [Google Scholar] [CrossRef]
  34. Costello, F.J.; Kim, M.G.; Kim, C.; Lee, K.C. Exploring a Fuzzy Rule Inferred ConvLSTM for Discovering and Adjusting the Optimal Posture of Patients with a Smart Medical Bed. Int. J. Environ. Res. Public Health 2021, 18, 6341. [Google Scholar] [CrossRef]
  35. Pouyan, M.; Birjandtalab, J.; Heydarzadeh, M.; Nourani, M.; Ostadabbas, S. A Pressure Map Dataset for Posture and Subject Analytics; IEEE: Orlando, FL, USA, 2017; pp. 65–68. [Google Scholar]
  36. Davoodnia, V.; Ghorbani, S.; Etemad, A. Estimating Pose from Pressure Data for Smart Beds with Deep Image-Based Pose Estimators. Appl. Intell. 2022, 52, 2119–2133. [Google Scholar] [CrossRef]
  37. Ostadabbas, S.; Baran Pouyan, M.; Nourani, M.; Kehtarnavaz, N. In-Bed Posture Classification and Limb Identification. In Proceedings of the 2014 IEEE Biomedical Circuits and Systems Conference (BioCAS) Proceedings, Lausanne, Switzerland, 22–24 October 2014; pp. 133–136. [Google Scholar]
  38. Clever, H.M.; Kapusta, A.; Park, D.; Erickson, Z.; Chitalia, Y.; Kemp, C.C. 3D Human Pose Estimation on a Configurable Bed from a Pressure Image; IEEE: Madrid, Spain, 2018. [Google Scholar]
  39. Diao, H.; Chen, C.; Yuan, W.; Amara, A.; Tamura, T.; Fan, J.; Meng, L.; Liu, X.; Chen, W. Deep Residual Networks for Sleep Posture Recognition with Unobtrusive Miniature Scale Smart Mat System. IEEE Trans. Biomed. Circuits Syst. 2021, 15, 111–121. [Google Scholar] [CrossRef] [PubMed]
  40. Duvall, J.; Karg, P.; Brienza, D.; Pearlman, J. Detection and Classification Methodology for Movements in the Bed That Supports Continuous Pressure Injury Risk Assessment and Repositioning Compliance. J. Tissue Viability 2019, 28, 7–13. [Google Scholar] [CrossRef] [PubMed]
  41. Fonseca, L.; Ribeiro, F.; Metrôlho, J. Effects of the Number of Classes and Pressure Map Resolution on Fine-Grained In-Bed Posture Classification. Computation 2023, 11, 239. [Google Scholar] [CrossRef]
  42. Gabison, S.; Pupic, N.; Evans, G.; Dolatabadi, E.; Dutta, T. Measuring Repositioning in Home Care for Pressure Injury Prevention and Management. Sensors 2022, 22, 7013. [Google Scholar] [CrossRef]
  43. Garcia-Molina, G.; Winger, T.; Makaram, N.; Rao, M.R.; Chernega, P.; Shcherbakov, Y.; McGhee, L.; Chellamuthu, V.; Veneros, E.; Mills, R.; et al. Unobtrusive Skin Temperature Estimation on a Smart Bed. Sensors 2024, 24, 4882. [Google Scholar] [CrossRef]
  44. Gargees, R.; Keller, J.; Popescu, M.; Skubic, M. Non-Invasive Classification of Sleep Stages with a Hydraulic Bed Sensor Using Deep Learning; Pagan, J., Mokhtari, M., Aloulou, H., Abdulrazak, B., Cabrera, M., Eds.; Springer: Cham, Switzerland, 2019; Volume 11862, pp. 73–82. [Google Scholar]
  45. Hagihara, A.; Harada, F.; Shimakawa, H. Estimation of Posture and Prediction of the Elderly Getting out of Bed Using a Body Pressure Sensor. Int. J. Electr. Comput. Eng. (IJECE) 2021, 11, 1208–1222. [Google Scholar] [CrossRef]
  46. Hsiao, R.-S.; Mi, Z.; Yang, B.-R.; Kau, L.-J.; Bitew, M.A.; Li, T.-Y. Body Posture Recognition and Turning Recording System for the Care of Bed Bound Patients. Technol. Health Care 2015, 24 (Suppl. 1), S307–S312. [Google Scholar] [CrossRef]
  47. Hu, Q.; Tang, X.; Tang, W. A Real-Time Patient-Specific Sleeping Posture Recognition System Using Pressure Sensitive Conductive Sheet and Transfer Learning. IEEE Sens. J. 2021, 21, 6869–6879. [Google Scholar] [CrossRef]
  48. Hu, D.; Gao, W.; Ang, K.K.; Hu, M.; Chuai, G.; Huang, R. Smart Sleep Monitoring: Sparse Sensor-Based Spatiotemporal CNN for Sleep Posture Detection. Sensors 2024, 24, 4833. [Google Scholar] [CrossRef]
  49. Jung, H.; Kimball, J.; Receveur, T.; Gazi, A.; Agdeppa, E.; Inan, O. Estimation of Tidal Volume Using Load Cells on a Hospital Bed. IEEE J. Biomed. Health Inform. 2022, 26, 3330–3341. [Google Scholar] [CrossRef] [PubMed]
  50. Kawakami, M.; Toba, S.; Fukuda, K.; Hori, S.; Abe, Y.; Ozaki, K. Application of Deep Learning to Develop a Safety Confirmation System for the Elderly in a Nursing Home. J. Robot. Mechatron. 2017, 29, 338–345. [Google Scholar] [CrossRef]
  51. Kusmakar, S.; Karmakar, C.; Zhu, Y.; Shelyag, S.; Drummond, S.; Ellis, J.; Angelova, M. A Machine Learning Model for Multi-Night Actigraphic Detection of Chronic Insomnia: Development and Validation of a Pre-Screening Tool. R. Soc. Open Sci. 2021, 8, 202264. [Google Scholar] [CrossRef] [PubMed]
  52. Kuwahara, N.; Wada, K. Bed-Leaving Prediction Using a Sheet-Type Pressure-Sensitive Sensor Base with Deep-Learning. J. Fiber Sci. Technol. 2017, 73, 343–347. [Google Scholar] [CrossRef]
  53. Liu, Q.; Kenny, M.; Nilforooshan, R.; Barnaghi, P. An Intelligent Bed Sensor System for Non-Contact Respiratory Rate Monitoring. arXiv 2021, arXiv:2103.13792. [Google Scholar]
  54. Manners, J.; Kemps, E.; Guyett, A.; Stuart, N.; Lechat, B.; Catcheside, P.; Scott, H. Estimating Vigilance from the Pre-Work Shift Sleep Using an under-Mattress Sleep Sensor. J. Sleep. Res. 2024, 33, e14138. [Google Scholar] [CrossRef]
  55. Matar, G.; Lina, J.-M.; Kaddoum, G. Artificial Neural Network for In-Bed Posture Classification Using Bed-Sheet Pressure Sensors. IEEE J. Biomed. Health Inform. 2020, 24, 101–110. [Google Scholar] [CrossRef]
  56. Bernal Monroy, E.; Polo Rodríguez, A.; Espinilla Estevez, M.; Medina Quero, J. Fuzzy Monitoring of In-Bed Postural Changes for the Prevention of Pressure Ulcers Using Inertial Sensors Attached to Clothing. J. Biomed. Inform. 2020, 107, 103476. [Google Scholar] [CrossRef]
  57. Mosquera-Lopez, C.; Leitschuh, J.; Condon, J.; Hagen, C.; Rajhbeharrysingh, U.; Hanks, C.; Jacobs, P. Design and Evaluation of a Non-Contact Bed-Mounted Sensing Device for Automated In-Home Detection of Obstructive Sleep Apnea: A Pilot Study. Biosensors 2019, 9, 90. [Google Scholar] [CrossRef]
  58. Pornpreedawan, P.; Puengsiricharoen, K.; Tantisatirapong, S.; Taweerutchana, R.; Sueaseenak, D. A Bed-Fall Notification System Using Pressure and Ultrasonic Sensors. Curr. Appl. Sci. Technol. 2022, 22, 10-55003. [Google Scholar] [CrossRef]
  59. Pupic, N.; Gabison, S.; Evans, G.; Fernie, G.; Dolatabadi, E.; Dutta, T. Detecting Patient Position Using Bed-Reaction Forces and Monitoring Skin-Bed Interface Forces for Pressure Injury Prevention and Management. medRxiv 2022, 22272323. [Google Scholar] [CrossRef]
  60. Rosales, L.; Su, B.; Skubic, M.; Ho, K. Heart Rate Monitoring Using Hydraulic Bed Sensor Ballistocardiogram. J. Ambient. Intell. Smart Environ. 2017, 9, 193–207. [Google Scholar] [CrossRef]
  61. Stern, L.; Fernie, G.; Roshan Fekr, A. A Novel In-Bed Body Posture Monitoring for Decubitus Ulcer Prevention Using Body Pressure Distribution Mapping. Biomed. Eng. Online 2024, 23, 34. [Google Scholar] [CrossRef] [PubMed]
  62. Tandon, A.; Goyal, A.; Clever, H.M.; Erickson, Z. BodyMAP—Jointly Predicting Body Mesh and 3D Applied Pressure Map for People in Bed. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, USA, 16–22 June 2024. [Google Scholar]
  63. Tapwal, R.; Misra, S.; Deb, P.K. I-Sheet: A Low-Cost Bedsheet Sensor for Remote Diagnosis of Isolated Individuals. IEEE Sens. J. 2023, 23, 906–913. [Google Scholar] [CrossRef] [PubMed]
  64. Walsh, L.; McLoone, S.; Ronda, J.; Duffy, J.; Czeisler, C. Noncontact Pressure-Based Sleep/Wake Discrimination. IEEE Trans. Biomed. Eng. 2017, 64, 1750–1760. [Google Scholar] [CrossRef] [PubMed]
  65. Willemen, T.; Van Deun, D.; Verhaert, V.; Pirrera, S.; Exadaktylos, V.; Verbraecken, J.; Haex, B.; Vander Sloten, J. Automatic Sleep Stage Classification Based on Easy to Register Signals as a Validation Tool for Ergonomic Steering in Smart Bedding Systems. Work 2012, 41 (Suppl. 1), 1985–1989. [Google Scholar] [CrossRef]
  66. Austin, D.; Beattie, Z.T.; Riley, T.; Adami, A.M.; Hagen, C.C.; Hayes, T.L. Unobtrusive Classification of Sleep and Wakefulness Using Load Cells under the Bed. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. 2012, 2012, 5254–5257. [Google Scholar] [CrossRef]
  67. Bajkowski, T.M.; Marchal, N.; Saied-Walker, J.; Gupta, P.; Keller, J.M.; Skubic, M.; Scott, G.J. Cohort Discovery from Bed Sensor Data with Fuzzy Evidence Accumulation Clustering; Institute of Electrical and Electronics Engineers Inc.: Piscataway, NJ, USA, 2023. [Google Scholar]
  68. Belay, S.S.; Prince, T.; Beyene, T.B.; Marew, W.Y.; Kebede, A.D. Implementing a Robust IoT Solution to Monitor Prone Position Using Machine Learning Techniques; Institute of Electrical and Electronics Engineers Inc.: Piscataway, NJ, USA, 2022; pp. 263–269. [Google Scholar]
  69. Breuss, A.; Sferrazza, C.; Pleisch, J.; D’Andrea, R.; Riener, R. Unobtrusive Sleep Position Classification Using a Novel Optical Tactile Sensor. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. 2023, 2023, 1–5. [Google Scholar] [CrossRef]
  70. Channa, A.; Yousuf, M.; Popescu, N. Machine Learning Algorithms for Posture Identification of Obstructive Sleep Apnea Patients Using IoT Solutions; IEEE: Iasi, Romania, 2020. [Google Scholar]
  71. Davoodnia, V.; Etemad, A. Identity and Posture Recognition in Smart Beds with Deep Multitask Learning; IEEE: Bari, Italy, 2019; pp. 3054–3059. [Google Scholar]
  72. Goldberger, A.L.; Amaral, L.A.; Glass, L.; Hausdorff, J.M.; Ivanov, P.C.; Mark, R.G.; Mietus, J.E.; Moody, G.B.; Peng, C.K.; Stanley, H.E. PhysioBank, PhysioToolkit, and PhysioNet: Components of a New Research Resource for Complex Physiologic Signals. Circulation 2000, 101, E215–E220. [Google Scholar] [CrossRef]
  73. Duan, B.; Zhao, D.; Yang, J.; Wang, S. A Novel Posture Recognition Based on Time Series Supervised Learning Algorithm. In Proceedings of the 2021 IEEE International Conference on Intelligence and Safety for Robotics (ISR), Nagoya, Japan, 4–6 March 2021; pp. 394–398. [Google Scholar]
  74. Enayati, M.; Skubic, M.; Keller, J.M.; Popescu, M.; Farahani, N.Z. Sleep Posture Classification Using Bed Sensor Data and Neural Networks. In Proceedings of the 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Honolulu, HI, USA, 17–21 July 2018; pp. 461–465. [Google Scholar]
  75. Heydarzadeh, M.; Nourani, M.; Ostadabbas, S. In-Bed Posture Classification Using Deep Autoencoders. In Proceedings of the 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Orlando, FL, USA, 16–20 August 2016; pp. 3839–3842. [Google Scholar]
  76. Husák, M.; Kaczmarczyk, V.; Baštán, O.; Benešl, T. Application of Body Posture Recognition Algorithm in IoT Smart Mattress. In Proceedings of the 2021 Selected Issues of Electrical Engineering and Electronics (WZEE), Rzeszow, Poland, 13–15 September 2021; pp. 1–3. [Google Scholar]
  77. Ibrahim, A.; Chaccour, K.; El Hassani, A.H.; Andres, E. SleepPal: A Novel System for Elderly Sleep Monitoring and Bed Falls Detection; Springer: Cham, Switzerland, 2024; Volume 2087, pp. 76–90. [Google Scholar]
  78. Lei, C.; Li, F.; Chen, X.; Yang, L. Pressure-Based In-Bed Pose Estimation Methods Using the Skinned Multi-Person Linear Model. In Proceedings of the 2024 4th Asia Conference on Information Engineering (ACIE), Singapore, 26–28 January 2024; Institute of Electrical and Electronics Engineers Inc.: New York, NY, USA, 2024; pp. 74–79. [Google Scholar]
  79. Luo, S.; Zhao, C.; Fu, Y. An Intelligent Human Activity Recognition Method with Incremental Learning Capability for Bedridden Patients. In Proceedings of the 2018 15th International Conference on Control, Automation, Robotics and Vision (ICARCV), Singapore, 18–21 November 2018; pp. 1284–1289. [Google Scholar]
  80. Madokoro, H.; Shimoi, N.; Sato, K. Unrestrained Multi-Sensor Systems for Real-Time Prediction of Bed-Leaving Behavior Patterns. In Proceedings of the 2014 SICE Annual Conference (SICE), Sapporo, Japan, 9–12 September 2014; pp. 1946–1953. [Google Scholar]
  81. Matthies, D.J.C.; Haescher, M.; Chodan, W.; Bieber, G. DIY-PressMat: A Smart Sensor Mat for Posture Detection Applicable for Bed-Exit Intention Detection, Pressure Ulcer Prevention, and Sleep Apnea Mitigation. In Proceedings of the Proceedings of the 14th PErvasive Technologies Related to Assistive Environments Conference, Corfu, Greece, 29 June–2 July 2021; Association for Computing Machinery: New York, NY, USA, 2021; pp. 76–80. [Google Scholar]
  82. Mendez, M.O.; Migliorini, M.; Kortelainen, J.M.; Nistico, D.; Arce-Santana, E.; Cerutti, S.; Bianchi, A.M. Evaluation of the Sleep Quality Based on Bed Sensor Signals: Time-Variant Analysis. In Proceedings of the 2010 Annual International Conference of the IEEE Engineering in Medicine and Biology, Buenos Aires, Argentina, 31 August–4 September; pp. 3994–3997. [CrossRef]
  83. Metsis, V.; Galatas, G.; Papangelis, A.; Kosmopoulos, D.; Makedon, F. Recognition of Sleep Patterns Using a Bed Pressure Mat. In Proceedings of the 4th International Conference on Pervasive Technologies Related to Assistive Environments, Heraklion Crete, Greece, 25–27 May 2011; Association for Computing Machinery: New York, NY, USA, 2011; pp. 1–4. [Google Scholar]
  84. Migliorini, M.; Bianchi, A.M.; Nisticò, D.; Kortelainen, J.; Arce-Santana, E.; Cerutti, S.; Mendez, M.O. Automatic Sleep Staging Based on Ballistocardiographic Signals Recorded through Bed Sensors. In Proceedings of the 2010 Annual International Conference of the IEEE Engineering in Medicine and Biology, Buenos Aires, Argentina, 31 August–4 September 2010; pp. 3273–3276. [Google Scholar]
  85. Moon, A.; Son, S.W.; Park, H. Exploration of Lossy Posture Classification Model Using In-Bed Flexible Pressure Sensors; IEEE: Boston, MA, USA, 2023; pp. 1–4. [Google Scholar]
  86. Mukai, T.; Matsuo, K.; Kato, Y.; Shimizu, A.; Guo, S. Lying Posture Detection for Unconstrained Measurement of Respiration and Heartbeat on a Bed. In Proceedings of the IASTED International Conference Biomedical Engineering (BioMed 2014), Zurich, Switzerland, 23–25 June 2014. [Google Scholar]
  87. Obo, T.; Kubota, N. Structured Learning in Fuzzy Spiking Neural Networks for Human State Estimation. In Proceedings of the 2014 World Automation Congress (WAC), Waikoloa, HI, USA, 3–7 August 2014; pp. 707–712. [Google Scholar]
  88. Perez-Macias, J.M.; Adavanne, S.; Viik, J.; Värri, A.; Himanen, S.-L.; Tenhunen, M. Assessment of Support Vector Machines and Convolutional Neural Networks to Detect Snoring Using Emfit Mattress. In Proceedings of the 2017 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Jeju, Republic of Korea, 11–15 July 2017; pp. 2883–2886. [Google Scholar]
  89. Pouyan, M.; Ostadabbas, S.; Nourani, M.; Pompeo, M. Classifying Bed Inclination Using Pressure Images; IEEE: Chicago, IL, USA, 2014; pp. 4663–4666. [Google Scholar]
  90. Pouyan, M.B.; Nourani, M.; Pompeo, M. Sleep State Classification Using Pressure Sensor Mats. In Proceedings of the 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Milan, Italy, 25–29 August 2015; pp. 1207–1210. [Google Scholar]
  91. Rangarajan, S.; Lee, Y.; Johnson, V.; Schorger, K.; Lee, H.; Nguyen, D.; Behfar, M.H.; Jansson, E.; Rekilä, J.; Hiltunen, J.; et al. A Novel IoT System For Patient-Centric Pressure Ulcer Prevention Using Sensor Embedded Dressings; IEEE: Pisa, Italy, 2022; pp. 42–45. [Google Scholar]
  92. Russo, A.; Licciardo, G.; Di Benedetto, L.; Rubino, A.; Liguori, R.; Naddeo, A.; Cappetti, N. Quantized Fully Convolution Neural Network for HW Implementation of Human Posture Recognition; IEEE: Washington, DC, USA, 2021. [Google Scholar]
  93. Sano, A.; Picard, R.W. Comparison of Sleep-Wake Classification Using Electroencephalogram and Wrist-Worn Multi-Modal Sensor Data. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. 2014, 2014, 930–933. [Google Scholar] [CrossRef]
  94. Sawada, K.; Nakasho, K.; Wasaki, K.; Shimoi, N. Classification of Human Posture on Bed Using Machine Learning; Institute of Electrical and Electronics Engineers Inc.: Piscataway, NJ, USA, 2022; pp. 547–548. [Google Scholar]
  95. Soleimani, R.; Pesch, D. Preventing Pressure Ulcers by Image Based Posture Detection; IEEE: L’Aquila, Italy, 2023; pp. 23–28. [Google Scholar]
  96. Vazquez-Santacruz, E.; Gamboa-Zuniga, M. An Intelligent Device for Assistance in Caring for the Health of Patients with Motor Disabilities; Xiao, Z., Tong, Z., Li, K., Wang, X., Li, K., Eds.; IEEE: Zhangjiajie, China, 2015; pp. 868–874. [Google Scholar]
  97. Vyas, N.; Ryoo, K.; Tesfaye, H.; Yi, R.; Skubic, M. Sleep Stage Classification Using Non-Invasive Bed Sensing and Deep Learning. In Proceedings of the 2021 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Houston, TX, USA, 9–12 December 2021; pp. 3232–3238. [Google Scholar]
  98. Wai, A.A.P.; Yuan-Wei, K.; Fook, F.S.; Jayachandran, M.; Biswas, J.; Cabibihan, J.-J. Sleeping Patterns Observation for Bedsores and Bed-Side Falls Prevention. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. 2009, 2009, 6087–6090. [Google Scholar] [CrossRef] [PubMed]
  99. Wu, Z.; Wan, Q.; Zhao, M.; Ke, Y.; Fang, Y.; Liang, Z.; Xie, F.; Cheng, J. MassNet: A Deep Learning Approach for Body Weight Extraction from A Single Pressure Image; Institute of Electrical and Electronics Engineers Inc.: Piscataway, NJ, USA, 2023; pp. 180–189. [Google Scholar]
  100. Yoon, J.; Moon, A.; Son, S.W. Effective Posture Classification Using Statistically Significant Data from Flexible Pressure Sensors. IEEE J. Flex. Electron. 2024, 3, 173–180. [Google Scholar] [CrossRef]
  101. Youngkong, P.; Panpanyatep, W. A Novel Double Pressure Sensors-Based Monitoring and Alarming System for Fall Detection; IEEE: Bangkok, Thailand, 2021; pp. 39–43. [Google Scholar]
  102. Yousefi, R.; Ostadabbas, S.; Faezipour, M.; Nourani, M.; Ng, V.; Tamil, L.; Bowling, A.; Behan, D.; Pompeo, M. A Smart Bed Platform for Monitoring & Ulcer Prevention. In Proceedings of the 2011 4th International Conference on Biomedical Engineering and Informatics (BMEI), Shanghai, China, 15–17 October 2011; Volume 3, pp. 1362–1366. [Google Scholar]
  103. Stern, L.; Roshan Fekr, A. Recent Advances and Applications of Textile Technology in Patient Monitoring. Healthcare 2023, 11, 3066. [Google Scholar] [CrossRef] [PubMed]
Informatics 11 00076 g001
Figure 1. PRISMA flow chart.
Figure 1. PRISMA flow chart.
Informatics 11 00076 g001
Informatics 11 00076 g002
Figure 2. Type of bed sensors (columns) and their positions (colors). In the y-axis are reported the number of studies, while the x-axis categorizes the types of sensors.
Figure 2. Type of bed sensors (columns) and their positions (colors). In the y-axis are reported the number of studies, while the x-axis categorizes the types of sensors.
Informatics 11 00076 g002
Informatics 11 00076 g003
Figure 3. Bubble chart of the mean minimal accuracy (color) of the best model divided by outcome (vital signs, sleep, multiple output data, in-bed pose estimation, bedsores) and type of input data (acceleration data, multiple input data, other, pressure data, pressure image/map, vital sign data).
Figure 3. Bubble chart of the mean minimal accuracy (color) of the best model divided by outcome (vital signs, sleep, multiple output data, in-bed pose estimation, bedsores) and type of input data (acceleration data, multiple input data, other, pressure data, pressure image/map, vital sign data).
Informatics 11 00076 g003
Informatics 11 00076 g004
Figure 4. Preprocessing techniques in in-bed patient monitoring studies with machine learning approach.
Figure 4. Preprocessing techniques in in-bed patient monitoring studies with machine learning approach.
Informatics 11 00076 g004
Informatics 11 00076 g005
Figure 5. Roadmap for Future Smart-Bed Technologies.
Figure 5. Roadmap for Future Smart-Bed Technologies.
Informatics 11 00076 g005
Table 1. Descriptive characteristics of the included studies.
Table 1. Descriptive characteristics of the included studies.
Author, YearStudy DesignPopulation CategoryNumber of ParticipantsNumber of MalesNumber of FemalesAge MeanSettingSensor Type
Research article
Albukhari, 2019 [21] ExperimentalVolunteers761 Controlled environmentBed sensors
Alinia, 2020 [22] Olguın & Pentland, 2006 [23]; Altun, 2010 [24]Healthy volunteers 307; 47; 4 Controlled environmentWearable
Arora, 2020 [25] Observational; Arriba-Pérez, 2016 [26]Volunteers2250 HomeWearable
Azimi, 2020 [27]LongitudinalElders9 Nursing homeBed sensors
Bai, 2023 [28]ExperimentalDementia nursing home residents24 HospitalBed sensors
Breuss, 2024 [29]ExperimentalHealthy volunteers2112928.3 (male) 29.5 (female)Controlled environmentBed sensors
Bruser, 2013 [30]ExperimentalHealthy volunteers109163.1HospitalBed sensors
Casas, 2019 [31]Experimentalhealthy volunteers642 Controlled environmentBed sensors
Chica, 2012 [32]ObservationalPatients201010 Controlled environmentBed sensors
Cho, 2019 [33]ExperimentalHealthy volunteers1073 HomeWearable
Costello, 2021 [34]Pouyan, 2017 [35] Healthy volunteers13 Controlled environmentBed sensors
Davoodnia, 2022 [36]Ostadabbas, 2014 [37]; Pouyan, 2017 [35]; Clever, 2018 [38]Healthy volunteers30 Controlled environmentBed sensors
Diao, 2021 [39]ExperimentalVolunteers1697 Controlled environmentBed sensors
Duvall, 2019 [40]ExperimentalHealthy volunteers10 Controlled environmentBed sensors
Fonseca, 2023 [41]ExperimentalHealthy volunteers60 Controlled environment (sensor sheets over and under a mattress)Bed sensors
Gabison, 2022 [42]ExperimentalHealthy volunteers945 HomeBed sensors
Garcia-Molina, 2024 [43]ExperimentalHealthy volunteers189944.5Controlled environment; homeBoth
Gargees, 2019 [44]ExperimentalHealthy volunteers56421429.27Controlled environmentBed sensors
Hagihara, 2021 [45]ExperimentalHealthy volunteers142; 53; 4 Controlled environmentBed sensors
Hsiao, 2015 [46]ExperimentalVolunteers9 Controlled environmentBed sensors
Hu, 2021 [47]ExperimentalHealthy volunteers53229.2Controlled environmentBed sensors
Hu, 2024 [48]ExperimentalHealthy volunteers22157 HomeBed sensors
Jung, 2022 [49]ExperimentalHealthy volunteers159625.8HospitalBed sensors
Kawakami, 2017 [50]ExperimentalPatients32176.7Nursing homeBed sensors
Kusmakar, 2021 [51]ObservationalPatients; healthy volunteers80384247.6HomeWearable
Kuwahara & Wada, 2017 [52]ObservationalHealthy volunteers7 Controlled environmentBed sensors
Liu, 2021 [53]ObservationalHealthy volunteers42229.75Controlled environmentBed sensors
Manners, 2024 [54]RCTHealthy volunteers24121227.6Controlled environmentBed sensors
Matar, 2020 [55]ObservationalHealthy volunteers1210227.35Controlled environmentBed sensors
Monroy, 2020 [56]ObservationalHealthy volunteers74324.75Controlled environmentWearable
Mosquera-Lopez, 2019 [57]ObservationalPatients1411348Controlled environment; homeBed sensors
Pornpreedawan, 2022 [58]ExperimentalHealthy volunteers105522.5Controlled environmentBed sensors
Pupic, 2022 [59]Observational; simulatedHealthy volunteers1810829.8Controlled environmentBed sensors
Raschella, 2022 [54] ObservationalPatients2619768Controlled environment; homeWearable
Rosales, 2017 [60]ExperimentalElders44091.25Nursing homeBed sensors
Stern, 2024 [61]ExperimentalHealthy volunteers10 Hospital bed, home bed, home bed with foam mattress topperBed sensors
Tandon, 2024 [62]ExperimentalHealthy volunteers Controlled environmentBed sensors
Tapwal, 2023 [63]ExperimentalCOVID-19 patients Controlled environment; homeBed sensors
Walsh, 2017 [64]ExperimentalHealthy volunteers151211 Controlled environmentBed sensors
Waltisberg, 2017 [9]ObservationalPatients96353.6Controlled environmentBed sensors
Willemen, 2012 [65]SimulatedHealthy volunteers10 22.95 Controlled environmentBoth
Conference paper
Austin, 2012 [66]ObservationalPatients2718951Controlled environmentWearable
Bajkowski, 2023 [67]ExperimentalElders19 Nursing homeBed sensors
Belay, 2022 [68]ExperimentalElders7 Controlled environmentBed sensors
Breuss, 2023 [69]MethodologicalHealthy volunteers1 Controlled environmentBed sensors
Channa, 2020 [70]Pouyan, 2017 [35]Healthy volunteers13 26.9Controlled environmentBed sensors
Davoodnia, 2019 [71]Pouyan, 2017 [35]; Goldberger, 2000 [72]Healthy volunteers13 Controlled environmentBed sensors
Duan, 2021 [73]ExperimentalVolunteers8 Controlled environmentBed sensors
Enayati, 2018 [74]ExperimentalHealthy volunteers58 Controlled environmentBed sensors
Heydarzadeh, 2016 [75]ExperimentalVolunteers10 Controlled environmentBed sensors
Husák, 2021 [76]MethodologicalHealthy volunteers Controlled environmentBed sensors
Ibrahim, 2024 [77]MethodologicalHealthy volunteers15 Controlled laboratory environmentWearable
Lei, 2024 [78]MethodologicalHealthy volunteers Controlled environmentBed sensors
Luo, 2018 [79]ExperimentalVolunteers10 Controlled environmentBed sensors
Madokoro, 2014 [80]ExperimentalVolunteers10 Controlled environmentBed sensors
Matthies, 2021 [81]ExperimentalVolunteers118331.45Controlled environmentBed sensors
Mendez, 2010 [82]ExperimentalHealthy volunteers11011 Controlled environmentBed sensors
Metsis, 2011 [83]ExperimentalVolunteers3 Controlled environmentBed sensors
Migliorini, 2010 [84]ExperimentalHealthy volunteers11011 Controlled environmentBed sensors
Moon, 2023 [85]ExperimentalBedridden patients5 HospitalBed sensors
Mukai, 2014 [86]ExperimentalHealthy volunteers1174 Controlled environmentBed sensors
Oboe Kubota, 2014 [87]MethodologicalVolunteers Controlled environmentBed sensors
Perez-Macias, 2017 [88]ExperimentalVolunteers30246 Controlled environmentBed sensors
Pouyan, 2014 [89]ExperimentalVolunteers15 Controlled environmentBed sensors
Pouyan, 2015 [90]ExperimentalVolunteers8 Controlled environmentWearable
Pouyan, 2017 [35]ExperimentalHealthy volunteers13 26.9Controlled environmentBed sensors
Rangarajan, 2022 [91]ExperimentalHealthy volunteers5 Tertiary care, university-affiliated hospitalWearable
Russo, 2021 [92]Pouyan, 2017 [35]Healthy volunteers13 Controlled environmentBed sensors
Sano & Picard, 2014 [93]ExperimentalVolunteers15 Controlled environmentWearable
Sawada, 2022 [94]ExperimentalHealthy volunteers11 Controlled environmentBed sensors
Soleimani & Pesch, 2023 [95]ExperimentalPatients at risk of pressure ulcers13 Controlled environmentBed sensors
Vázquez-Santacruz & Gamboa-Zúñiga, 2016 [96]MethodologicalVolunteers1 Controlled environmentBed sensors
Vyas, 2021 [97]ObservationalPatients52366.2HospitalBed sensors
Wai, 2009 [98]MethodologicalVolunteers Controlled environmentBed sensors
Wu, 2023 [99]ExperimentalHealthy volunteers1073 Controlled environmentBed sensors
Yoon, 2024 [100]ExperimentalHealthy volunteers532 HospitalBed sensors
Youngkong, 2021 [101]ExperimentalVolunteers633 Controlled environmentBed sensors
Yousefi, 2011 [102]MethodologicalVolunteers Controlled environmentBed sensors
Table 2. Summary of the characteristics of the machine learning technique (MLT) used in the included articles.
Table 2. Summary of the characteristics of the machine learning technique (MLT) used in the included articles.
Characteristic OverallOtherPosition EstimationSleep and VigilanceVital Signs
N = 78N = 31N = 481N = 191N = 81
Type of input dataAcceleration data6 (7.7%)1 (33%)2 (4.2%)3 (16%)0 (0%)
Multiple input data12 (15%)0 (0%)7 (15%)5 (26%)0 (0%)
Other5 (6.4%)0 (0%)1 (2.1%)4 (21%)0 (0%)
Pressure data29 (37%)2 (67%)21 (44%)2 (11%)4 (50%)
Pressure image/map1 (1.3%)0 (0%)1 (2.1%)0 (0%)0 (0%)
Vital sign data17 (22%)0 (0%)16 (33%)1 (5.3%)0 (0%)
Input data pre-processingNo8 (10%)0 (0%)0 (0%)4 (21%)4 (50%)
MLT categoryDeep learning29 (37%)2 (67%)22 (46%)3 (16%)2 (25%)
Shallow learning39 (49%)1 (33%)21 (44%)12 (64%)5 (63%)
Both10 (13%)0 (0%)5 (10%)4 (21%)1 (13%)
MLT typeBoosting methods5 (6.4%)0 (0%)3 (6.3%)0 (0%)2 (25%)
Discriminant analysis1 (1.3%)0 (0%)0 (0%)1 (5.3%)0 (0%)
KNN-based6 (7.7%)0 (0%)5 (10%)1 (5.3%)0 (0%)
Linear models1 (1.3%)0 (0%)1 (2.1%)0 (0%)0 (0%)
Naive Bayes1 (1.3%)0 (0%)0 (0%)1 (5.3%)0 (0%)
Other32 (41%)2 (67%)24 (50%)3 (16%)3 (38%)
Random Forest9 (12%)0 (0%)6 (13%)2 (11%)1 (13%)
SVM-based22 (28%)1 (33%)9 (19%)10 (53%)2 (25%)
Unsupervised Learning1 (1.3%)0 (0%)0 (0%)1 (5.3%)0 (0%)
Accuracy more than 95%Yes36 (47%)2 (67%)29 (62%)2 (11%)3 (38%)
Abbreviations: MLT = machine learning technique.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Ocagli, H.; Lanera, C.; Borghini, C.; Khan, N.M.; Casamento, A.; Gregori, D. In-Bed Monitoring: A Systematic Review of the Evaluation of In-Bed Movements Through Bed Sensors. Informatics 2024, 11, 76. https://doi.org/10.3390/informatics11040076

AMA Style

Ocagli H, Lanera C, Borghini C, Khan NM, Casamento A, Gregori D. In-Bed Monitoring: A Systematic Review of the Evaluation of In-Bed Movements Through Bed Sensors. Informatics. 2024; 11(4):76. https://doi.org/10.3390/informatics11040076

Chicago/Turabian Style

Ocagli, Honoria, Corrado Lanera, Carlotta Borghini, Noor Muhammad Khan, Alessandra Casamento, and Dario Gregori. 2024. "In-Bed Monitoring: A Systematic Review of the Evaluation of In-Bed Movements Through Bed Sensors" Informatics 11, no. 4: 76. https://doi.org/10.3390/informatics11040076

APA Style

Ocagli, H., Lanera, C., Borghini, C., Khan, N. M., Casamento, A., & Gregori, D. (2024). In-Bed Monitoring: A Systematic Review of the Evaluation of In-Bed Movements Through Bed Sensors. Informatics, 11(4), 76. https://doi.org/10.3390/informatics11040076

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop