Skip to main content
Download PDF
  • Systematic Review
  • Open access
  • Published:

Applications of digital health technologies and artificial intelligence algorithms in COPD: systematic review

BMC Medical Informatics and Decision Making volume 25, Article number: 77 (2025) Cite this article

Abstract

Background

Chronic Obstructive Pulmonary Disease (COPD) represents a significant global health challenge, placing considerable burdens on healthcare systems. The rise of digital health technologies (DHTs) and artificial intelligence (AI) algorithms offers new opportunities to improve COPD predictive capabilities, diagnostic accuracy, and patient management. This systematic review explores the types of data in COPD under DHTs, the AI algorithms employed for data analysis, and identifies key application areas reported in the literature.

Methods

A systematic search was conducted in PubMed and Web of Science for studies published up to December 2024 that applied AI algorithms in digital health for COPD management. Inclusion criteria focused on original research utilizing AI algorithms and digital health technologies for COPD, while review articles were excluded. Two independent reviewers screened the studies, resolving discrepancies through consensus.

Results

From an initial pool of 265 studies, 41 met the inclusion criteria. Analysis of these studies highlighted a diverse range of data types and modalities collected from DHTs in the COPD context, including clinical data, patient-reported outcomes, and environmental/lifestyle data. Machine learning (ML) algorithms were employed in 34 studies, and deep learning (DL) algorithms in 16. Support vector machines and boosting were the most frequently used ML models, while deep neural networks (DNN) and convolutional neural networks (CNN) were the most commonly used DL models. The review identified three key application domains for AI in COPD: screening and diagnosis (10 studies), exacerbation prediction (22 studies), and patient monitoring (9 studies). Disease progression prediction was a prevalent focus across three domains, with promising accuracy and performance metrics reported.

Conclusions

Digital health technologies and AI algorithms have a wide range of applications and promise for COPD management. ML models, in particularly, show great potential in improving digital health solutions for COPD. Future research should focus on enhancing global collaboration to explore the cost-effectiveness and data-sharing capabilities of DHTs, enhancing the interpretability of AI models, and validating these algorithms through clinical trials to facilitate their safe integration into the routine COPD management.

Peer Review reports

Introduction

Background

Chronic obstructive pulmonary disease (COPD) represents a significant public health concern globally [1] ranking as the third leading cause of mortality and seventh in terms of health risks [2]. In 2019, approximately 392 million people worldwide were affected by COPD, resulting in over 3 million deaths annually [3]. The complex nature of COPD, characterized by progressive and irreversible deterioration, emphasizes the urgent need for early diagnosis, prompt detection of exacerbations, and continuous monitoring to ensure effective management [4, 5].

Digital health is defined as the utilization of digital, mobile, and wireless technologies to facilitate the realization of health objectives [6, 7]. The scope of digital health technologies (DHTs) encompasses e-health, mHealth, remote monitoring of medical devices, public health alerts and so on, are pivotal in transforming COPD care, moving from traditional self-management strategies, like written action plans, to more dynamic, data-driven approaches [8,9,10,11]. For example, DHTs can facilitate personalized communication by sending reminders and health promotion messages, thereby stimulating demand for services and enhancing access to health information [12]. However, the extensive and complex data generated by these technologies present substantial analytical challenges [13]. Advanced computational tools are essential for providing deeper insights into the progression of COPD [14].

AI algorithms, particularly machine learning (ML) and deep learning (DL), are playing an important role across various health-related domains [15]. These algorithms are increasingly applied to recognize key patterns in healthcare data, enabling intelligent clinical systems and improving the diagnosis and treatment of complex diseases [16,17,18]. Research has shown that AI can enhance the management of chronic diseases and improve diagnostic capabilities [14]. ML, a subset of AI algorithms, employs statistical and mathematical models to identify patterns in data [19], and its application of ML in diagnostics has demonstrated promising results. For instance, Aldhyani et al. [20], illustrated that integrating soft clustering techniques with ML can improve predictive accuracy for chronic diseases by managing ambiguous data more effectively.

DL, a more advanced branch of AI, can directly learn from raw data without the need for manual feature engineering [21]. Its ability to process unstructured data, such as speech, images, and videos, makes it a valuable tool for analyzing COPD patients [22, 23]. In a pioneering study, Tang et al. [24] utilized DL to train and validate residual networks on low-dose computed tomography (CT) scans for COPD detection. This innovative approach has demonstrated that DL can enhance diagnostic precision while reducing the radiation exposure associated with higher dose scans.

Despite the promising role of AI algorithms in improving COPD management through DHTs, there is a lack of comprehensive reviews that specifically address this area. Much of the existing literature tends to either take a broad view of AI in healthcare [4, 25] or focus narrowly on specific tasks within COPD care [26, 27]. This review aims to fill that gap by systematically exploring the primary applications of DHTs and AI algorithms in COPD, with a particular emphasis on the types and sources of data access from DHTs, the algorithms employed and their applications to COPD management.

Objectives

This systematic review aims to: (1) summarizing the type and source of data access from DHTs for COPD; (2) identifying the specific AI algorithms used within the digital health framework for COPD management; (3) and examining the key applications of AI algorithms in advancing digital health for COPD. Through this exploration, this review intends to offer a comprehensive understanding of the current landscape and suggest future directions of AI-driven digital health in COPD care.

Methods

Search strategy

This review was conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [28], with the completed PRISMA checklist available in Supplementary Material 1. The search was performed in December 2024 using the Web of Science and PubMed databases, aiming to gather studies that incorporate AI into COPD care via digital health platforms. A broad range of keywords, detailed in Table 1, was used in combination with the “OR” and “AND” logical operators to refine the search across article topics, titles, abstracts, and full texts. These keywords were carefully selected to capture various aspects of AI’s role in COPD-related digital health. Specific search strings for each database are provided in Supplementary Material 2.

Table 1 Search strategy
Full size table

Eligibility criteria

Articles were included if they were full-text, published in English, and focused on the direct application of DHTs and AI algorithms for data accessing and processing in the context of COPD. The studies had to utilize DHTs for practical COPD care, addressing aspects such as diagnosis, prognosis, management, or treatment of the disease, rather than merely describe the potential mechanisms or development process of DHTs. Only original research articles were considered. Review articles, editorials, and non-original research papers were excluded. The primary exclusion criterion was the removal of duplicates found across the two databases.

Study selection

The search initially yielded 224 articles, which were reduced to 152 after deduplication. Two independent reviewers screened the remaining articles in a two-stage process. In the first stage, the titles and abstracts of the articles were assessed for relevance. This step resulted in 51 articles that met the inclusion criteria. In the second stage, a full-text review was conducted for these 51 articles, and 32 studies were determined to meet al.l eligibility criteria. Any disagreements between the two reviewers during the screening process were resolved through consensus. In order to guarantee that the existing literature was both comprehensive and up-to-date, a further search was conducted. This search covered the period from 2024/03/07 to 2024/12/24, and resulted in the inclusion of a further 9 articles. The selection process is outlined in Fig. 1, in accordance with PRISMA guidelines [28].

Fig. 1
figure 1

Selection process

Full size image

Data extraction

Data extraction was performed across four main domains: (1) publication details (such as the first author and year of publication, (2) study data (including the number and characteristics of participants, types of data captured by digital health device, and the use of public database), (3) algorithms (the types of AI algorithms used, specific models implemented, their performance metrics, and their applications in COPD digital health), and (4) DHTs’ applications (the DHTs deployed and their role in COPD care). Two independent reviewers conducted the data extraction, and any discrepancies were resolved by discussion.

The extracted data were further analyzed to categorize the digital health applications according to the lifecycle stages of COPD management. This analysis allowed for the identification of key advancements in AI algorithms applied to COPD digital health management and provided insights into how these technologies are being integrated into different aspects of COPD care.

Result

To summarize the findings of this review, we present them in three parts. The first part summarizes the characteristics of data used by AI algorithms within digital health settings for COPD patients. The second part outlines the types of algorithms used in digital health for COPD and the associated data they utilize. The third part focuses on the primary applications of AI-driven digital health solutions for COPD. Figure 2 provides an overview of these key findings.

Fig. 2
figure 2

Overview of key findings from our systematic review

Full size image

Data types and source of COPD digital health

DHTs capture a variety of data from COPD patients, such as symptoms, daily living capacity, sleep, and vital signs, across different environments (home, hospital, work, outdoors) [8]. We classified the data into three categories based on the included articles: clinical data, patient-reported outcomes, and environmental and lifestyle data [29]. Figure 3 presents the data types, features, and sources in detail.

Fig. 3
figure 3

Data types and features captured from digital health devices. (Created with BioRender.com)

Full size image

Clinical data

Clinical data, including clinical records, vital signs, and medical examination results, reflects patients’ health status. Vital signs, particularly blood oxygen saturation (SpO2), heart rate, and respiratory rate [30], are crucial in assessing the lung function and oxygenation status of COPD patients. To facilitate regular monitoring of these parameters, digital health devices like pulse oximeters and blood pressure monitors are employed [31,32,33,34]. Additionally, devices such as smart masks [35], electronic stethoscopes [36], sensors [37], and electret condenser microphone (ECM) [38] are used to track respiratory sounds. Lung function tests, considered the gold standard for COPD diagnosis, necessitate consistent monitoring. Portable spirometers [39] are instrumental in measuring key physiological indicators like Forced Expiratory Volume in the first second (FEV1), Forced Vital Capacity (FVC). Furthermore, research incorporating image data from capnographic signals and photoplethysmograms (PPGs) has been advancing, with these signals obtained through specific sensors [40] and pulse oximeters [32], respectively. Additionally, the examination of saliva and sputum, collected via clinical questionnaires [41] or extracted using permittivity biosensors [42], has been explored in a limited number of studies, adding to the breadth of clinical data available for COPD care.

Patient-reported outcomes

Patient-reported outcomes are essential in documenting symptoms such as breathlessness, coughing, fever, and chest pain [43], which are vital for assessing the severity of COPD and formulating treatment strategies. This category includes information such as patient demographics and medical history, typically stored in electronic medical records. In the management of chronic diseases, patient self-reporting is a critical component for acquiring precise and prompt data. Digital platforms facilitate this process; e-questionnaires [44] and dedicated applications like myCOPD [45], Re-Admit [46], and DmD [47] allow patients to easily record and communicate their symptoms from the comfort of their home. Moreover, innovative solutions such as digital pens [48] and daily diaries [49] provide patients with varied options for documenting their health status, enhancing the versatility and accessibility of patient-reported data collection.

Environmental and lifestyle factors

Environmental and lifestyle factors, such as air quality, temperature, smoking status, sleep patterns, and physical activity [50], significantly influence the lung health and respiratory function of COPD patients. The measurement of air quality, for example, is facilitated through environmental sensors and weather data interfaces, including air quality-sensing devices [41, 51] and platforms like the World Air Quality Index (WAQI) data platform [52], Weatherbit Application Programming Interface (API), and OpenWeather API [33]. Conversely, wearable devices and accelerometers are predominantly employed to capture lifestyle-related data. Brands like Fitbit, Garmin, Apple, Oura Ring, and Asus provide wearables that monitor sleep, walking speed, and distance, while accelerometers, such as Stayhealthy [53] and ActiGraph GT3X [54], specifically track limb movement during physical activities. Furthermore, the development of mobile applications for assessing natural walking speed and distance [55] enables the execution of standardized tests like the six-minute walk test in a patient-friendly manner.

The integration of these DHTs in daily monitoring emphasizes their vital role in assessing and managing COPD. Additionally, the utilization of public databases for gathering audio data, such as coughs and breathing sounds, has expanded the scope of data available for COPD research and analysis [49, 56, 57], as outlined in Table 2. These databases complement digital health data to help classify and predict COPD.

Table 2 Introduction of public datasets
Full size table

Summary of AI algorithms used in digital health for COPD

An in-depth analysis of the included studies revealed that AI algorithms applied in digital health for COPD primarily focused on conventional ML algorithms and DL algorithms. Out of the 41 studies analyzed, 34 utilized conventional ML algorithms, and 16 employed DL algorithms. For clarity, the term “ML” in this review refers specifically to conventional ML algorithms that do not include neural networks. A detailed breakdown of specific algorithms used in these studies is provided in Supplementary Material 3, with Table 3 showing the top five most used algorithms in each category.

Table 3 The top five ML and DL algorithms used in Digital Health for COPD
Full size table

ML algorithms

As DHT advanced and data complexity increased, ML algorithms became central to COPD digital health analytics. The five most commonly used algorithms were Support Vector Machine (SVM), Boosting, Random Forest (RF), Logistic Regression (LR), and Decision Tree (DT), applied 14, 14, 13, 11, and 9 times, respectively, across the 41 included studies.

SVM [62, 63] emerged as the preferred technique, used in 14 studies. SVM’s versatility in handling both classification and regression tasks, along with its ability to support various kernel functions (such as radial basis functions), makes it highly adaptable for analyzing diverse data types, including breathing sounds. For instance, researchers used SVM to process multidimensional motion data from mobile devices, effectively extracting relevant information from high-dimensional data [55]. In other studies, SVM processed breathing sounds using kernel tricks [38], demonstrating strong generalization abilities, particularly in small datasets like home-based physiological monitoring data, such as oxygen saturation and respiratory rate [29].

Boosting algorithms, including XGBoost [64], AdaBoost [52], LightGBM [33], and CatBoost [37], were used in 14 of the 41 included studies. These algorithms, known for combining multiple weak models into a single strong model, were frequently applied to analyze complex physiological and clinical data, such as pulmonary function test results, blood gas analyses, symptom scores, and respiratory signals. XGBoost consistently demonstrated strong performance across diverse datasets, particularly when analyzing combinations of demographic, symptomatic, and respiratory data. Meanwhile, CatBoost excelled in handling categorical data from medical records, providing high accuracy in analyzing patient characteristics and clinical histories.

Tree-based algorithms, such as RF [65] and DT [41], were also widely used, appearing 13 and 9 times, respectively. These algorithms are robust against overfitting and can handle large datasets, which is crucial for analyzing clinical and audio data. RF was particularly effective in handling noisy data, such as coughing sounds from COPD patients [66], while DT provided simpler models for feature selection and prediction.

LR [67], used 11 times, was widely adapted due to its simplicity and interpretability, especially in predicting COPD presence. LR efficiently processed data from mobile devices, such as demographic information, clinical symptoms, and pulmonary function test results, providing accurate screening outcomes [36]. Additionally, LR proved effective in processing time-series data and identifying disease patterns [32]. Several semi-supervised learning algorithms, such as probabilistic models like Bayesian networks [44] and naïve Bayes (NB) [51], were also employed in COPD digital health studies. These models helped manage uncertainty and predict exacerbations using COPD-related digital health data.

DL algorithms

The growth of digital health data, such as patient monitoring and medical imaging, has facilitated the adoption of DL algorithms in COPD management. Techniques such as Convolutional Neural Network (CNN) [68, 69], Deep Neural Network (DNN) [46,47,48,49,50,51,52,53] Recurrent Neural Networks (RNN) [57], Long Short-Term Memory (LSTM) [70], and Gated Recurrent Units (GRU) [33] were frequently used due to their ability to process complex and varied data types.

CNNs were employed in 10 of the 41 included studies and were particularly effective in image classification, directly capturing spatial features for the remote monitoring of COPD patients using visual data. For example, OpenPose [71], a CNN-based system, was utilized to remotely monitor physical activity and frailty in COPD patients through video data. This approach enabled accurate tracking of patient mobility and overall health status.

LSTM [45] (used in 6 studies) and RNN [52] (used in 5 studies) were favored for their capacity to process sequential data, such as breath sounds and time-series health monitoring. These algorithms were effective in tasks like breath sound detection and classification, as well as remote health monitoring. For example, RNNoise [57], a DL method for breath sound analysis, consistently captured variations in breath sounds while performing noise removal, making it possible for continuous monitoring in COPD patients.

In some studies, researchers integrated multiple DL algorithms to target specific data challenges. For instance, innovative AI models like the Spatio-Temporal Artificial Intelligence Network (STAIN) demonstrated the integration of CNN with RNN to analyze sequential images over time, providing a comprehensive analysis of COPD patient data in a temporal and spatial context. Additionally, some isolated studies [47, 49, 52, 54, 56] explored non-mainstream DL algorithms, enriching the diversity of approaches in COPD digital health research. These innovative approaches have led to the development of sophisticated models that combine multiple algorithms for enhanced analytical capabilities, further advancing the potential of DL in managing COPD.

Applications of AI-based digital health for COPD

The integration of AI algorithms into digital health for COPD has significantly advanced the field by enabling more precise and timely interventions. Through the analysis of the 32 included studies, AI applications have been grouped into three key areas: COPD Screening and Diagnosis (10 studies), Identification and Response to COPD Exacerbations (22 studies), and COPD Patient Monitoring (9 studies). These AI algorithms, applied across a variety of digital health platforms, primarily focus on processing complex and high-dimensional data generated by sensors, wearable devices, and other health monitoring technologies. AI algorithms embedded in these applications highlight their ability to transform raw health data into actionable insights, thus enhancing COPD management by improving early detection, predicting disease progression, and ensuring continuous patient monitoring. Table 4 provides a summary of the three application areas, including the data sizes, algorithms, features used for specific tasks, and performance metrics reported in the literature.

Table 4 A summary of the three application areas of digital health in COPD
Full size table

COPD screening and diagnosis

In exploring diagnostic techniques for COPD, researchers utilized different levels of algorithmic strategies, from the application of a single ML algorithm (2/10), to the integration of multiple ML algorithms (3/10), to the introduction of DL algorithms (3/10), and finally, comparative studies between ML and DL algorithms (2/10), each step contributing to improved diagnostic accuracy and efficiency.

Initially, Zhang et al. [35] developed a smart mask integrated with self-powered respiratory sensors and a portable readout circuit, with a bagged DT model distinguishing between five healthy individuals and twenty those with chronic respiratory diseases, including COPD, with up to 95.5% accuracy. The development of such smart masks represents a significant advancement in respiratory health monitoring, particularly in a time of increasing respiratory-related epidemics.

To further enhance diagnostic accuracy, multiple ML algorithms were applied in some studies. For instance, pulse oximetry data (oxygen saturation and heart rate) was analyzed using five ML algorithms, such as XGBoost, RF, and SVM, with XGBoost performing best, achieving 91.04% accuracy in identifying COPD patients from a diverse dataset [71]. Another system, TussisWatch [66], a smartphone-based AI system, monitored real-time cough patterns using sensors and applied multiple ML algorithms to diagnose conditions such as COPD, which the RF model achieving an accuracy of 78.05%.

As DL algorithms gained popularity in medical diagnostics, it was increasingly applied to COPD detection [73]. For instance, a 1-D deformable CNN-based system was developed for quantitative capnographic sensor analysis, achieving classification accuracies of 92.9% and 92.16% in laboratory and real-world environments, respectively [40]. DL’s capacity to process capnographic signals demonstrates its potential in diagnosing COPD and related conditions.

Comparative studies between ML and DL approaches were also conducted. In one study [69], SVMs, CNNs, and LSTMs processed airflow sensor data that mimicked respiratory patterns of COPD and ILD patients. CNNs achieved 100% accuracy in identifying severe COPD cases, with an overall average accuracy of 85%. The ConvLSTM model, which combines CNN and LSTM, enhanced diagnostic accuracy to 97%, showcasing DL’s superior capacity in processing complex respiratory data and improving the recognition of respiratory dysfunction.

Despite heterogeneity among the studies, the median accuracy across the algorithms for diagnosing COPD was 91.25%, with even the lowest-performing studies demonstrating superior accuracy compared to traditional methods. These findings suggest that AI-powered diagnostic tools could significantly improve early detection and screening for COPD in the future.

Identification and response to COPD exacerbations

COPD is characterized as a progressive and incurable condition with frequent acute exacerbations that lead to reduced lung function, diminished quality of life, and increased mortality risk [73]. Early detection and timely intervention are critical to mitigating these severe consequences [74]. A large portion of the studies (22/41) focused on identifying COPD exacerbations. Of these, four employed single ML algorithms, one used multiple DL algorithms, six applied multiple ML algorithms, and seven compared ML and DL algorithms.

For example, in four studies employing single ML algorithms, models like DTs [37], k-means clustering [75], Bayesian networks [44], and LR [76] were used. One study employing DTs successfully predicted early acute exacerbations with 11 out of 18 wavelet features in a total of 16 COPD patients with an accuracy of 78%, detecting exacerbations 4.4 days before their onset by analyzing respiratory audio data [37].

Other studies incorporated more advanced AI approaches. One examined accelerometer data to monitor patients’ physical activity, transmitted to a smartphone for analysis using logarithmic regression, SVM, and neural networks [53]. The SVM achieved over 90% accuracy in filtering 17 relevant features from 90, improving assessment accuracy in 58 COPD patients.

Interactive digital health technologies have also advanced, focusing on improving doctor-patient communication. The Smart Clinical Decision Support System (CIDSS) [33] integrated biometric abnormalities and environmental data, using LSTM, GRU, and boosting algorithms to assess COPD exacerbation risks. Another system, EDGE [32], analyzed pulse oximetry data from 110 moderate-to-severe COPD patients, finding that oxygen saturation, respiratory rate, and pulse rate were critical predictors of exacerbation events. Combining these vital signs with ML algorithms significantly improved prediction accuracy and demonstrated the reliability of pulse oximetry in home monitoring.

COPD patient monitoring

Continuous monitoring is essential for COPD patients due to the variability in their symptoms and condition [4]. In the included studies, DHTs monitored various real-time parameters, such as gait, walking speed, posture, respiratory rate, blood oxygen level, heart rate, and blood pressure [28, 60]. A variety of AI algorithms were applied to these data to improve monitoring accuracy and efficiency.

For instance, Juen et al. [54] used smartphone sensors to collect walking data and predict patients’ walking speed and distance using ML algorithms like SVMs. The study demonstrated high prediction accuracy during a controlled 6-minute walk test, with only a 3.23% error rate, suggesting the potential application of this method in natural walking conditions. In another study, Sanchez-Morillo et al. [31] developed an intelligent portable oxygen concentrator (iPOC) that automatically adjusted oxygen flow based on the patient’s physical activity levels. In a preliminary trial, the system achieved 91.1% weighted accuracy, leading to enhanced patient oxygenation and higher satisfaction compared to traditional oxygen concentrators.

Additionally, AI algorithms were applied to monitor speech patterns in COPD patients, focusing on the rhythmic aspects of speech, which could be influenced by physical activity and medication intake. Using ML classification techniques, researchers successfully predicted lung function (FEV1), demonstrating the importance of recording patient conditions and medication information in developing an automated COPD monitoring system [77].

Furthermore, some research proposed frameworks integrating environmental sensors, spirometers, and peak flow meters with algorithms like probabilistic latent component analysis (PLCA) and linear dynamical systems (LDS) [47] for multivariate tracking and prediction. These systems were capable of detecting changes in respiratory symptoms during periods of stability, providing a more comprehensive approach to COPD monitoring [47].

Discussion

This systematic review provides an in-depth analysis of the application of AI algorithms in digital health for COPD. Our findings underscore three key areas: the types and sources of data utilized in digital health, the types of AI algorithms employed in digital health and their application domains within COPD care. The primary finding is the significant role of ML and DL algorithms in advancing digital health, particularly in the areas of screening and diagnosis, exacerbation prediction, and patient monitoring for COPD. This highlights the critical role of AI algorithms plays in improving the precision, efficiency, and scope of DHTs aimed at enhancing COPD care.

Potential and challenges of DHTs in data accessing

In the context of DHTs, data is multimodal and high dimensional [78], which is also confirmed in our review. The health status of patients with COPD can be characterized by various signals, such as clinical variables, symptoms, lifestyle-related data, and continuous signals from various digital health devices. Of these, clinical and patient-reported data are the most commonly used forms, particularly SpO2, respiratory rate, heart rate, body weight, FEV1 and peak expiratory flow, providing a dual perspective of objective health status and subjective symptom experience.

In contrast, environmental and lifestyle data remain relatively underutilized in digital health frameworks for COPD, likely due to the complexity of accessing and analyzing these data. For example, the collection of physical activity data may require the integration of various DHTs, such as accelerometers and smart devices, and the pre-processing of collected images or videos [72]. However, studies have shown a 10% improvement in accuracy and a 20% improvement in AUROC for lifestyle and environmental data compared to using clinical questionnaire data alone [39]. Therefore, the collection and integrated use of this type of data should receive more attention in the future.

While DHTs have facilitated data collection, data sharing remains a significant challenge, particularly across national and institutional boundaries. We advocate for enhanced global collaboration to establish a comprehensive COPD dataset that enables the exchange of patient data worldwide. Such efforts are critical to improving the accuracy and utility of COPD monitoring systems, deepening our understanding of the disease’s epidemiology, and developing more precise and personalized treatment strategies for patients.

Comparison of ML and DL algorithms within DHTs

ML algorithms, used in 34 out of 41 studies, dominate AI applications in COPD digital health, with SVM and boosting being the most commonly used. On the other hand, DL algorithms, applied in 16 studies, such as CNNs and LSTMs, are increasingly gaining counts. ML algorithms achieved accuracy levels exceeding 78.05%, while DL algorithms surpassed 72%. Interestingly, about half of studies comparing ML and DL found DL to outperform ML, suggesting that with larger datasets and more diverse data types, DL could become a more robust option in the future [79]. For example, studies combining clinical test data, respiratory signals, and image data like CT imaging demonstrated improved predictions of COPD progression using multimodal approaches [80].

Another important observation is the prevalence of small sample sizes in many of the reviewed studies, which likely explains the preference for ML algorithms over more complex DL models. Small datasets pose significant challenges for DL, as they increase the risk of overfitting and hinder generalization [81]. For instance, Wu et al. [36] employed a DNN combined with cross-validation to predict AECOPD in a dataset of only 67 patients. While this approach achieved the best performance compared to other ML algorithms, it highlights the need for larger datasets and methodological innovations to ensure robust and generalizable outcomes. This presents an important opportunity for future research to improve the reliability and clinical applicability of DL models in COPD management.

The lack of interpretability in DL algorithms, as suggested in previous literature [19], remains a significant barrier to their adoption in clinical practice. Researchers have proposed several approaches to address this issue, including ensemble methods to enhance robustness [82], Bayesian networks [42] to improve interpretability, and techniques such as bootstrapping [31] and data augmentation [83] to improve data quality. These advancements underscore the ongoing need for research aimed at making DL models more transparent and interpretable, facilitating their integration into clinical workflows.

Practicality and cost-effectiveness of DHTs

Of the three major applications areas, exacerbation prediction is most frequently studied (22/41 studies). This trend mirrors other chronic disease applications [84], as exacerbations have a significant impact on patient outcomes, increasing hospitalizations, mortality risk, and healthcare costs. Accurate exacerbation prediction allows for timely intervention, improving prognosis and reducing the burden on healthcare systems.

While AI-based DHTs offer promising solutions [85], a key challenge is ensuring equitable access for all patients with COPD [86]. Disparities in access to technology due to socioeconomic and geographical factors can lead to unequal benefits from these innovations [79]. Moreover, AI algorithms must be designed to account for demographic differences, such as ethnicity, to ensure accurate predictions across diverse populations [87]. For instance, the majority of included studies employed smartphones as a medium due to their cost-effectiveness and portability, enabling remote counselling, health education, and self-management support [49, 62]. However, this approach poses significant challenges for older patients with COPD, who may face difficulties in navigating such technology. Additionally, cost-effectiveness remains a crucial consideration, as the cost of these tools often exceeds that of routine care, making them less accessible in low-resource settings [88]. These observations highlight the need for digital tools that are affordable, accessible, and user-friendly tools, especially for populations in low-resource settings or those with limited technological proficiency.

Limitations

This review has several limitations. The search strategy utilized broad terms like AI, ML, and DL but did not include specific algorithm names, which may have led to the exclusion of relevant studies focusing on specific or novel AI techniques. Furthermore, the literature search was restricted to PubMed and Web of Science, potentially overlooking important studies from other databases. Additionally, most of the included studies were small-scale trials, limiting the generalizability of the findings.

Implications for future research

Despite the challenges faced in the application of AI-driven DHTs in COPD management, several areas warrant further exploration to maximize their potential. First, future research should extend beyond traditional ML and DL algorithms to explore advanced techniques such as transfer learning and pre-trained models, including those based on foundation models like GPT-4 [89]. These models have shown significant promise in improving the adaptability and precision of AI in healthcare and could offer considerable benefits when applied to COPD management [90]. Their ability to adapt to various contexts with few samples data retraining could be especially useful in current situation with limited patient data.

Another critical area for future research is improving the interpretability of AI models. While DL algorithms have shown superior performance in some cases, their “black-box” nature remains a major obstacle to clinical adoption. Enhancing the transparency and interpretability of these models is essential for integrating AI into clinical workflows. Healthcare professionals need to trust and understand the rationale behind AI-generated insights to confidently incorporate them into patient care.

Additionally, future research should prioritize addressing issues related to technology equity and accessibility, optimising multi-sensor fusion technologies, and developing high-precision wearables and mobile apps. A review by Masanneck et al. [91] found that a marked increase in the frequency of clinical using DHTs, rising from 0.7% in 2010 to 11.4% in 2020. The field of chronic diseases, including COPD, similarly requires a stronger focus on DHTs in clinical trials and an advancement of regulatory science and standardization. It is critical that fairness issues are addressed to ensure that all COPD patients can benefit from DHTs, regardless of socioeconomic or geographic background. This requires designing technologies that are accessible to diverse populations and ensuring that these tools do not exacerbate existing health disparities [84]. Inclusivity must be prioritized in both the development and deployment of AI technologies, with careful consideration of socioeconomic, geographic, and ethnic diversity.

Conclusion

This systematic review highlights the transformative potential of DHTs and AI algorithms in COPD management. These technologies enable the integration and processing of multimodal data, transforming complex health information into actionable insights. AI-driven DHTs offer promising opportunities for monitoring and analyzing patient conditions, with ML algorithms emerging as the most commonly applied tools, particularly in exacerbation prediction, a central focus of many studies. To fully realize the potential of DHTs and AI in COPD care, future research is needed to address key challenges, including improving accessibility and ensuring fairness to prevent exacerbating existing health disparities. Additionally, global collaboration is essential to facilitate data collection and sharing, enabling a more comprehensive understanding of COPD epidemiology and enhancing the performance of AI algorithms with diverse datasets. By overcoming these challenges, AI-driven DHTs can more effectively address the diverse needs of COPD patients and contribute to better long-term outcomes.

Data availability

No datasets were generated or analysed during the current study.

Abbreviations

COPD:

Chronic Obstructive Pulmonary Disease

AI:

Artificial Intelligence

ML:

Machine Learning

DL:

Deep Learning

DNN:

Deep Neural Networks

CNN:

Convolutional Neural Networks

DHTs:

Digital Health Technologies

CT:

Computed Tomography

SVM:

Support Vector Machine

RF:

Random Forest

DT:

Decision Tree

LR:

Logistic Regression

NB:

Naïve Bayes

RNN:

Recurrent Neural Networks

LSTM:

Long Short-Term Memory

GRU:

Gated Recurrent Units

iPOC:

Intelligent Portable Oxygen Concentrator

PLCA:

Probabilistic Latent Component Analysis

LDS:

Linear Dynamical Systems

References

  1. Global Initiative for Chronic Obstructive Lung Disease. Global strategy for the diagnosis, management and prevention of chronic obstructive pulmonary disease (2023 report). Global Initiative for Chronic Obstructive Lung Disease - GOLD. https://goldcopd.org/2023-gold-report-2/. Accessed 31 Mar 2024.

  2. Guan Z, Li H, Liu R, et al. Artificial intelligence in diabetes management: advancements, opportunities, and challenges. Cell Rep Med. 2023;4(10):101213.

    Article  PubMed  PubMed Central  Google Scholar 

  3. Smoking is the leading cause of chronic obstructive pulmonary disease. https://www.who.int/news/item/15-11-2023-smoking-is-the-leading-cause-of-chronic-obstructive-pulmonary-disease. Accessed 3 Mar 2024.

  4. Long H, Li S, Chen Y. Digital health in chronic obstructive pulmonary disease. Chronic Dis Transl Med. 2023;9(2):90–103.

    PubMed  PubMed Central  Google Scholar 

  5. Albiges T, Sabeur Z, Arbab-Zavar B. Compressed sensing data with performing audio signal reconstruction for the intelligent classification of chronic respiratory diseases. SENSORS. 2023;23(3):1439.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Mathews SC, McShea MJ, Hanley CL, Ravitz A, Labrique AB, Cohen AB. Digital health: a path to validation. NPJ Digit Med. 2019;2:38.

    Article  PubMed  PubMed Central  Google Scholar 

  7. Cuff A. The evolution of digital health and its continuing challenges. BMC Digit Health. 2023;1(1):3.

    Article  PubMed  PubMed Central  Google Scholar 

  8. Bashi N, Fatehi F, Mosadeghi-Nik M, Askari MS, Karunanithi M. Digital health interventions for chronic diseases: a scoping review of evaluation frameworks. BMJ Health Care Inf. 2020;27(1):e100066. https://doiorg.publicaciones.saludcastillayleon.es/10.1136/bmjhci-2019-1000669.

    Article  Google Scholar 

  9. Digital health technologies for remote data acquisition in clinical investigations. 2023. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/digital-health-technologies-remote-data-acquisition-clinical-investigations Accessed 16 Mar 2024.

  10. Boer L, Bischoff E, van der Heijden M, et al. A smart mobile health tool versus a paper action plan to support self-management of chronic obstructive pulmonary disease exacerbations: randomized controlled trial. JMIR MHealth UHealth. 2019;7(10):e14408.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Ding H, Fatehi F, Maiorana A, Bashi N, Hu W, Edwards I. Digital health for COPD care: the current state of play. J Thorac Dis. 2019;11:S2210-20.

    Article  PubMed  PubMed Central  Google Scholar 

  12. WHO guideline recommendations on digital interventions for health system strengthening. Geneva: World Health Organization; 2019. 1, Introduction. Available from: https://www.ncbi.nlm.nih.gov/books/NBK541905/.

  13. Liu Y, Wu M. Deep learning in precision medicine and focus on glioma. Bioeng Transl Med. 2023;8(5):e10553.

    Article  PubMed  PubMed Central  Google Scholar 

  14. Chan J. Exploring digital health care: eHealth, mHealth, and librarian opportunities. J Med Libr Assoc. 2021;109(3):376–81.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Kitsios F, Kamariotou M, Syngelakis AI, Talias MA. Recent advances of Artificial Intelligence in healthcare: a systematic literature review. Appl Sci. 2023;13(13):7479.

    Article  Google Scholar 

  16. Kumar Y, Koul A, Singla R, Ijaz MF. Artificial intelligence in disease diagnosis: a systematic literature review, synthesizing framework and future research agenda. J Ambient Intell Humaniz Comput. 2023;14(7):8459–86.

    Article  PubMed  Google Scholar 

  17. Uysal G, Ozturk M. Hippocampal atrophy based Alzheimer’s disease diagnosis via machine learning methods. J Neurosci Methods. 2020;337:108669.

    Article  PubMed  Google Scholar 

  18. Huynh E, Hosny A, Guthier C, et al. Artificial intelligence in radiation oncology. Nat Rev Clin Oncol. 2020;17(12):771–81.

    Article  PubMed  Google Scholar 

  19. Soori M, Arezoo B, Dastres R. Artificial intelligence, machine learning and deep learning in advanced robotics, a review. Cogn Robot. 2023;3:54–70.

    Article  Google Scholar 

  20. Aldhyani THH, Alshebami AS, Alzahrani MY. Soft clustering for enhancing the diagnosis of chronic diseases over Machine Learning algorithms. J Healthc Eng. 2020;2020:4984967.

    Article  PubMed  PubMed Central  Google Scholar 

  21. Triantafyllidis A, Kondylakis H, Katehakis D, et al. Deep learning in mHealth for cardiovascular disease, diabetes, and cancer. Syst Rev JMIR MHealth UHealth. 2022;10(4):e32344. https://doiorg.publicaciones.saludcastillayleon.es/10.2196/32344.

    Article  Google Scholar 

  22. Srivastava A, Jain S, Miranda R, Patil S, Pandya S, Kotecha K. Deep learning based respiratory sound analysis for detection of chronic obstructive pulmonary disease. PeerJ Comput Sci. 2021;7:e369.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Zhuang Y, Xing F, Ghosh D, et al. Deep learning on graphs for multi-omics classification of COPD. PLoS ONE. 2023;18(4):e0284563.

    Article  PubMed  PubMed Central  Google Scholar 

  24. Tang LYW, Coxson HO, Lam S, Leipsic J, Tam RC, Sin DD. Towards large-scale case-finding: training and validation of residual networks for detection of chronic obstructive pulmonary disease using low-dose CT. Lancet Digit Health. 2020;2(5):e259-67.

    Article  PubMed  Google Scholar 

  25. Smith LA, Oakden-Rayner L, Bird A, et al. Machine learning and deep learning predictive models for long-term prognosis in patients with chronic obstructive pulmonary disease: a systematic review and meta-analysis. Lancet Digit Health. 2023;5(12):e872-81.

    Article  PubMed  Google Scholar 

  26. Tsang KCH, Pinnock H, Wilson AM, Shah SA. Application of machine learning algorithms for asthma management with mhealth: a clinical review. J Asthma Allergy. 2022;15:855–73.

    Article  PubMed  PubMed Central  Google Scholar 

  27. Glyde HMG, Morgan C, Wilkinson TMA, Nabney IT, Dodd JW. Remote patient monitoring and machine learning in acute exacerbations of chronic obstructive pulmonary disease: dual systematic literature review and narrative synthesis. J Med Internet Res. 2024;26: e52143.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71.

    Article  PubMed  PubMed Central  Google Scholar 

  29. Pyper E, McKeown S, Hartmann-Boyce J, Powell J. Digital health technology for real-world clinical outcome measurement using patient-generated data: systematic scoping review. J Med Internet Res. 2023;25: e46992. https://doiorg.publicaciones.saludcastillayleon.es/10.2196/46992.

    Article  PubMed  PubMed Central  Google Scholar 

  30. Al Rajeh A, Hurst J. Monitoring of physiological parameters to predict exacerbations of Chronic Obstructive Pulmonary Disease (COPD): a systematic review. J Clin Med. 2016;5(12):108. https://doiorg.publicaciones.saludcastillayleon.es/10.3390/jcm5120108.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Sanchez-Morillo D, Munoz-Zara P, Lara-Dona A, Leon-Jimenez A. Automated home oxygen delivery for patients with COPD and respiratory failure: a New Approach. Sensors. 2020;20(4):1178. https://doiorg.publicaciones.saludcastillayleon.es/10.3390/s20041178.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Pereira J, Antunes N, Rosa J, Ferreira JC, Mogo S, Pereira M. Intelligent clinical decision support system for managing COPD patients. J Pers Med. 2023;13(9): 1359. https://doiorg.publicaciones.saludcastillayleon.es/10.3390/jpm13091359.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Zhang K, Li Z, Zhang J, et al. Biodegradable Smart Face masks for Machine Learning-assisted chronic respiratory disease diagnosis. ACS Sens. 2022;7(10):3135–43. https://doiorg.publicaciones.saludcastillayleon.es/10.1021/acssensors.2c01628.

    Article  PubMed  Google Scholar 

  34. Shah SA, Velardo C, Farmer A, Tarassenko L. Exacerbations in Chronic Obstructive Pulmonary Disease: Identification and Prediction Using a Digital Health System. J Med Internet Res. 2017;19(3):e69. https://doiorg.publicaciones.saludcastillayleon.es/10.2196/jmir.7207.

  35. Chamberlain DB, Kodgule R, Fletcher RR. A mobile platform for automated screening of asthma and chronic obstructive pulmonary disease. Annu Int Conf IEEE Eng Med Biol Soc IEEE Eng Med Biol Soc Annu Int Conf. 2016;2016:5192–5. https://doiorg.publicaciones.saludcastillayleon.es/10.1109/EMBC.2016.7591897.

    Article  Google Scholar 

  36. Angel Fernandez-Granero M, Sanchez-Morillo D, Leon-Jinnenez A. An artificial intelligence approach to early predict symptom-based exacerbations of COPD. Biotechnol Biotechnol Equip. 2018;32(3):778–84. https://doiorg.publicaciones.saludcastillayleon.es/10.1080/13102818.2018.1437568.

    Article  Google Scholar 

  37. Angel Fernandez-Granero M, Sanchez-Morillo D, Leon-Jimenez A. Computerised analysis of telemonitored respiratory sounds for predicting acute exacerbations of COPD. SENSORS. 2015;15(10):26978–96. https://doiorg.publicaciones.saludcastillayleon.es/10.3390/s151026978.

    Article  Google Scholar 

  38. Yin H, Wang K, Yang R, et al. A machine learning model for predicting acute exacerbation of in-home chronic obstructive pulmonary disease patients. Comput Methods Programs Biomed. 2024;246: 108005. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.cmpb.2023.108005.

    Article  PubMed  Google Scholar 

  39. Bhagya D, Suchetha M. A 1-D deformable convolutional neural network for the quantitative analysis of capnographic sensor. IEEE Sens J. 2021;21(5):6672–8. https://doiorg.publicaciones.saludcastillayleon.es/10.1109/JSEN.2020.3042989.

    Article  Google Scholar 

  40. Wu CT, Li GH, Huang CT, et al. Acute exacerbation of a chronic obstructive pulmonary disease prediction system using wearable device data, machine learning, and deep learning: development and cohort study. JMIR MHealth UHealth. 2021;9(5):e22591 https://mhealth.jmir.org/2021/5/e22591/. Accessed 6 Nov 2023.

    Article  PubMed  PubMed Central  Google Scholar 

  41. Zarrin PS, Roeckendorf N, Wenger C. In-Vitro classification of Saliva samples of COPD patients and healthy controls using machine learning tools. IEEE ACCESS. 2020;8:168053–60. https://doiorg.publicaciones.saludcastillayleon.es/10.1109/ACCESS.2020.3023971.

    Article  Google Scholar 

  42. Wang JM, Han MK, Labaki WW. Chronic obstructive pulmonary disease risk assessment tools: is one better than the others? Curr Opin Pulm Med. 2022;28(2):99–108. https://doiorg.publicaciones.saludcastillayleon.es/10.1097/MCP.0000000000000833.

    Article  PubMed  Google Scholar 

  43. van der Heijden M, Lucas PJF, Lijnse B, Heijdra YF, Schermer TRJ. An autonomous mobile system for the management of COPD. J Biomed Inf. 2013;46(3):458–69. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.jbi.2013.03.003.

    Article  Google Scholar 

  44. Chmiel FP, Burns DK, Pickering JB, Blythin A, Wilkinson TMA, Boniface MJ. Prediction of chronic obstructive pulmonary disease exacerbation events by using patient self-reported data in a Digital Health App: statistical evaluation and machine learning approach. JMIR Med Inf. 2022;10(3):e26499. https://doiorg.publicaciones.saludcastillayleon.es/10.2196/26499.

    Article  Google Scholar 

  45. Wang L, Li G, Ezeana CF, et al. An AI-driven clinical care pathway to reduce 30-day readmission for chronic obstructive pulmonary disease (COPD) patients. Sci Rep. 2022;12(1):20633. https://doiorg.publicaciones.saludcastillayleon.es/10.1038/s41598-022-22434-3.

    Article  PubMed  PubMed Central  Google Scholar 

  46. Sanchez-Morillo D, Angel Fernandez-Granero M, Leon Jimenez A. Detecting COPD exacerbations early using daily telemonitoring of symptoms and k-means clustering: a pilot study. Med Biol Eng Comput. 2015;53(5):441–51. https://doiorg.publicaciones.saludcastillayleon.es/10.1007/s11517-015-1252-4.

    Article  PubMed  Google Scholar 

  47. Jacobson PK, Lind L, Persson HL. Unleashing the power of very small data to predict acute exacerbations of chronic obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis. 2023;18:1457–73. https://doiorg.publicaciones.saludcastillayleon.es/10.2147/COPD.S412692.

    Article  PubMed  PubMed Central  Google Scholar 

  48. Kolozali S, Chatzidiakou L, Jones R, Quint JK, Kelly F, Barratt B. Early detection of COPD patients’ symptoms with personal environmental sensors: a remote sensing framework using probabilistic latent component analysis with linear dynamic systems. NEURAL Comput Appl. 2023;35(23):17247–65. https://doiorg.publicaciones.saludcastillayleon.es/10.1007/s00521-023-08554-5.

    Article  PubMed  PubMed Central  Google Scholar 

  49. Jaehyuk C, Mun E. Review of internet of things-based artificial intelligence analysis method through real-time indoor air quality and health effect monitoring: focusing on indoor air pollution that are harmful to the respiratory organ. Tuberc Respir Dis. 2023;86(1):23–32. https://doiorg.publicaciones.saludcastillayleon.es/10.4046/trd.2022.0087.

    Article  Google Scholar 

  50. Wu CT, Wang SM, Su YE, et al. A precision health service for chronic diseases: development and cohort study using wearable device, machine learning, and deep learning. IEEE J Transl Eng Health Med. 2022;10:1–14. https://doiorg.publicaciones.saludcastillayleon.es/10.1109/JTEHM.2022.3207825.

    Article  Google Scholar 

  51. Bhowmik RT, Most SP. A personalized respiratory disease exacerbation prediction technique based on a novel spatio-temporal machine learning architecture and local environmental sensor networks. Electronics. 2022;11(16):2562. https://doiorg.publicaciones.saludcastillayleon.es/10.3390/electronics11162562.

    Article  Google Scholar 

  52. Dias A, Gorzelniak L, Schultz K, et al. Classification of exacerbation episodes in chronic obstructive pulmonary disease patients. Methods Inf Med. 2014;53(2):108–14. https://doiorg.publicaciones.saludcastillayleon.es/10.3414/ME12-01-0108.

    Article  PubMed  Google Scholar 

  53. Chen J, Weldemichael L, Zeng S, et al. Actigraphy informs distinct patient-centered outcomes in Pre-COPD. Respir Med. 2021;187:106543. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.rmed.2021.106543.

    Article  PubMed  Google Scholar 

  54. Juen J, Cheng Q, Schatz B. A natural walking monitor for pulmonary patients using Mobile phones. IEEE J Biomed Health Inf. 2015;19(4):1399–405. https://doiorg.publicaciones.saludcastillayleon.es/10.1109/JBHI.2015.2427511.

    Article  Google Scholar 

  55. Orchard P, Agakova A, Pinnock H, et al. Improving prediction of risk of Hospital Admission in Chronic Obstructive Pulmonary Disease: application of machine learning to Telemonitoring Data. J Med INTERNET Res. 2018;20(9):e263. https://doiorg.publicaciones.saludcastillayleon.es/10.2196/jmir.9227.

    Article  PubMed  PubMed Central  Google Scholar 

  56. Tran-Anh D, Vu NH, Nguyen-Trong K, Pham C. Multi-task learning neural networks for breath sound detection and classification in pervasive healthcare. Pervasive Mob Comput. 2022;86:101685. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.pmcj.2022.101685.

    Article  PubMed  PubMed Central  Google Scholar 

  57. Tang, Y., Turner, M.J. & Baker, A.B. Effects of alveolar dead-space, shunt and V/Q distribution on respiratory deadspace measurements. Br J Anaesth. 2005;95:538–48. https://doiorg.publicaciones.saludcastillayleon.es/10.1093/bja/aei212.

  58. Puppala M, He T, Chen S, et al. METEOR: An Enterprise Health Informatics Environment to support evidence-based medicine. IEEE Trans Biomed Eng. 2015;62(12):2776–86. https://doiorg.publicaciones.saludcastillayleon.es/10.1109/TBME.2015.2450181.

    Article  PubMed  Google Scholar 

  59. Clinical Practice Research Datalink. http://www.cprd.com/. Accessed 1 Apr 2024.

  60. Fonseca E, Plakal M, Font F, et al. General-purpose tagging of freesound audio with AudioSet labels: task description, dataset, and baseline. 2018. http://arxiv.org/abs/1807.09902. Accessed 1 Apr 2024.

  61. Papers with code - ICBHI respiratory sound database dataset. https://paperswithcode.com/dataset/icbhi-respiratory-sound-database. Accessed 1 Apr 2024.

  62. Jia Q, Chen Y, Zen Q, et al. Development and validation of machine learning-based models for prediction of intensive care unit admission and in-hospital mortality in patients with acute exacerbations of chronic obstructive pulmonary disease. Chronic Obstr Pulm Dis. 2024;11(5):460–71. https://doiorg.publicaciones.saludcastillayleon.es/10.15326/jcopdf.2023.0446.

    Article  PubMed  PubMed Central  Google Scholar 

  63. Duckworth C, Cliffe B, Pickering B, et al. Characterising user engagement with mHealth for chronic disease self-management and impact on machine learning performance. NPJ Digit Med. 2024;7(1):66. https://doiorg.publicaciones.saludcastillayleon.es/10.1038/s41746-024-01063-2.

    Article  PubMed  PubMed Central  Google Scholar 

  64. Davies HJ, Bachtiger P, Williams I, Molyneaux PL, Peters NS, Mandic DP. Wearable In-Ear PPG: detailed respiratory variations enable classification of COPD. IEEE Trans Biomed Eng. 2022;69(7):2390–400.

    Article  PubMed  Google Scholar 

  65. Windmon A, Minakshi M, Bharti P, et al. TussisWatch: a smart-phone system to identify cough episodes as early symptoms of chronic obstructive pulmonary disease and congestive heart failure. IEEE J Biomed Health Inf. 2019;23(4):1566–73.

    Article  Google Scholar 

  66. Talker L, Dogan C, Neville D, et al. Diagnosis and Severity Assessment of COPD using a novel fast-response capnometer and interpretable machine learning. COPD. 2024;21(1):2321379. https://doiorg.publicaciones.saludcastillayleon.es/10.1080/15412555.2024.2321379.

    Article  PubMed  Google Scholar 

  67. Lee HK, Park SU, Kong S, et al. Real-time deep learning-assisted mechano-acoustic system for respiratory diagnosis and multifunctional classification. NPJ Flex Electron. 2024;8(1):69. https://doiorg.publicaciones.saludcastillayleon.es/10.1038/s41528-024-00355-7.

    Article  Google Scholar 

  68. Mou X, Wang P, Sun J, et al. A Novel Approach for the detection and severity grading of chronic obstructive pulmonary disease based on transformed volumetric capnography. Bioengineering-basel. 2024;11(6):530. https://doiorg.publicaciones.saludcastillayleon.es/10.3390/bioengineering11060530.

  69. Jhunjhunwala M, Lin HL, Li GY, Chen CS. Recognition of respiratory dysfunctions using algorithm-assisted portable airflow sensors. ECS J SOLID STATE Sci Technol. 2020;9(11):115021. https://doiorg.publicaciones.saludcastillayleon.es/10.1149/2162-8777/abb3b0.

    Article  Google Scholar 

  70. Zahiri M, Wang C, Gardea M, et al. Remote physical Frailty monitoring- the application of deep learning-based image processing in tele-health. IEEE ACCESS. 2020;8:219391–9.

    Article  PubMed  PubMed Central  Google Scholar 

  71. Abineza C, Balas VE, Nsengiyumva P. A machine-learning-based prediction method for easy COPD classification based on pulse oximetry clinical use. J Intell FUZZY Syst. 2022;43(2):1683–95.

    Article  Google Scholar 

  72. Farrús M, Codina-Filbà J, Reixach E, Andrés E, Sans M, Garcia N, Vilaseca J. Speech-based support system to supervise chronic obstructive pulmonary disease patient status. Appl Sci. 2021;11(17):7999.

    Article  Google Scholar 

  73. Wu Y, Xia S, Liang Z, Chen R, Qi S. Artificial intelligence in COPD CT images: identification, staging, and quantitation. Respir Res. 2024;25(1):319.

    Article  PubMed  PubMed Central  Google Scholar 

  74. Homma T, Kawahara T, Mikuni H, et al. Beneficial effect of early intervention with garenoxacin for bacterial infection-induced acute exacerbation of bronchial asthma and chronic obstructive pulmonary disease. Int Arch Allergy Immunol. 2019;178(4):355–62.

    Article  PubMed  Google Scholar 

  75. Pericleous P, van Staa TP. The use of wearable technology to monitor physical activity in patients with COPD: a literature review. Int J Chron Obstruct Pulmon Dis. 2019;14:1317–22.

    Article  PubMed  PubMed Central  Google Scholar 

  76. Secher PH, Hangaard S, Kronborg T, et al. Clinical implementation of an algorithm for predicting exacerbations in patients with COPD in telemonitoring: a study protocol for a single-blinded randomized controlled trial. Trials. 2022;23(1):356.

    Article  PubMed  PubMed Central  Google Scholar 

  77. Gupta P, Saied Walker J, Despins L, et al. A semi-supervised approach to unobtrusively predict abnormality in breathing patterns using hydraulic bed sensor data in older adults aging in place. J Biomed Inf. 2023;147:104530.

    Article  Google Scholar 

  78. Berisha V, Krantsevich C, Hahn PR, et al. Digital medicine and the curse of dimensionality. NPJ Digit Med. 2021;4:153. https://doiorg.publicaciones.saludcastillayleon.es/10.1038/s41746-021-00521-5.

    Article  PubMed  PubMed Central  Google Scholar 

  79. Robertson NM, Centner CS, Siddharthan T. Integrating artificial intelligence in the diagnosis of COPD globally: a way forward. Chronic Obstr Pulm Dis J COPD Found. 2024;11(1):114–20.

    Google Scholar 

  80. Watson A, Wilkinson TMA. Digital healthcare in COPD management: a narrative review on the advantages, pitfalls, and need for further research. Ther Adv Respir Dis. 2022;16:17534666221075493.

  81. Amaral JLM, Lopes AJ, Faria ACD, Melo PL. Machine learning algorithms and forced oscillation measurements to categorise the airway obstruction severity in chronic obstructive pulmonary disease. Comput Methods Programs Biomed. 2015;118(2):186–97. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.cmpb.2014.11.002.

    Article  PubMed  Google Scholar 

  82. Glyde HMG, Blythin AM, Wilkinson TMA, Nabney IT, Dodd JW. Exacerbation predictive modelling using real-world data from the myCOPD app. Heliyon. 2024;10(10):e31201. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.heliyon.2024.e31201.

    Article  PubMed  PubMed Central  Google Scholar 

  83. Orani V, Paglialonga A, Cambiaso E, Mongelli M. A generative Adversarial Network (GAN) technique for internet of medical things data. Sensors (Basel, Switzerland). 2021;21(11):3726. https://doiorg.publicaciones.saludcastillayleon.es/10.3390/s21113726.

    Article  PubMed  Google Scholar 

  84. Canali S, Schiaffonati V, Aliverti A. Challenges and recommendations for wearable devices in digital health: data quality, interoperability, health equity, fairness. PLOS Digit Health. 2022;1(10):e0000104.

    Article  PubMed  PubMed Central  Google Scholar 

  85. Castelyn G, Laranjo L, Schreier G, Gallego B. Predictive performance and impact of algorithms in remote monitoring of chronic conditions: a systematic review and meta-analysis. Int J Med Inf. 2021;156:104620.

    Article  Google Scholar 

  86. Badr J, Motulsky A, Denis JL. Digital health technologies and inequalities: a scoping review of potential impacts and policy recommendations. Health Policy. 2024;146:105122.

    Article  PubMed  Google Scholar 

  87. Sezgin E. Redefining virtual assistants in health care: the future with large language models. J Med Internet Res. 2024;26:e53225.

    Article  PubMed  PubMed Central  Google Scholar 

  88. Xu H, Jiang X, Zeng Q, Li R. Application of e-Health tools in the assessment of inhalation therapy adherence in patients with chronic obstructive pulmonary disease: scoping review coupled with bibliometric analysis. Respir Med. 2025;236:107898. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.rmed.2024.107898.

    Article  PubMed  Google Scholar 

  89. Zhang J, Sun K, Jagadeesh A, et al. The potential and pitfalls of using a large language model such as ChatGPT, GPT-4, or LLaMA as a clinical assistant. J Am Med Inf Assoc. 2024;31(9):1884–91. https://doiorg.publicaciones.saludcastillayleon.es/10.1093/jamia/ocae184.

    Article  Google Scholar 

  90. Wilson S, Tolley C, Mc Ardle R, et al. Recommendations to advance digital health equity: a systematic review of qualitative studies. NPJ Digit Med. 2024;7(1):173.

    Article  PubMed  PubMed Central  Google Scholar 

  91. Masanneck L, Gieseler P, Gordon WJ, Meuth SG, Stern AD. Evidence from ClinicalTrials.gov on the growth of digital health technologies in neurology trials. NPJ Digit Med. 2023;6(1):23.

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

Not applicable.

Funding

This work was supported by the National Key Research and Development Program of China (Grant No. 2022YFC3601001), Chinese Academy of Medical Sciences Innovation Fund for Medical Sciences (Grant No. 2021-I2M-1-057), and IMICAMS Youth Talent Development Fund (Grant No. 2024YT01).

Author information

Author notes

  1. Zhenli Chen and Jie Hao contributed equally to this work.

Authors and Affiliations

  1. Institute of Medical Information/Medical Library, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, China

    Zhenli Chen, Jie Hao, Haixia Sun & Qing Qian

  2. Department of Respiratory Medicine, Xiangya Hospital, Central South University, Changsha, China

    Min Li & Yuan Zhang

Authors
  1. Zhenli Chen
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Jie Hao
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Haixia Sun
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Min Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Yuan Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Qing Qian
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

ZC, JH, and HS contributed to the concept and design of the study. ZC, JH, and HS formalized the eligibility criteria. ZC was responsible for data acquisition and data curation. ZC and HS conducted screening and full-text review. ZC, JH, HS, ML, and YZ analyzed the data and interpreted the results. ZC and JH drafted the initial manuscript. JH, HS, and QQ revised the manuscript. QQ supervised the work. All authors contributed to the review of the manuscript and provided approval for the final version of the manuscript.

Corresponding author

Correspondence to Qing Qian.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Chen, Z., Hao, J., Sun, H. et al. Applications of digital health technologies and artificial intelligence algorithms in COPD: systematic review. BMC Med Inform Decis Mak 25, 77 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12911-025-02870-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12911-025-02870-7

Keywords

BMC Medical Informatics and Decision Making

ISSN: 1472-6947

Contact us