Supplemental oxygen may be prescribed to correct severe hypoxemia at rest, with exertion, and/or during sleep. Ambulatory oxygen therapy is defined as the use of supplemental oxygen during exertion [1]. The American Thoracic Society guideline conditionally recommends prescribing ambulatory oxygen therapy for patients who have severe exertional room air hypoxemia [1], based on acute improvements in exercise capacity and modest evidence of improvements in health-related quality of life [2,3]. Portable oxygen concentrators (POCs), oxygen cylinders, and liquid oxygen (LOX) can be used to deliver ambulatory oxygen therapy [1]. Oxygen cylinders for ambulatory use can deliver continuous oxygen flow up to 6 L per minute but run out quickly, especially when used at high flow rates, and need to be refilled. POCs supply concentrated oxygen by removing nitrogen from the air as long as they have a power source; however, they may provide oxygen only intermittently (“pulse flow” triggered by inspiration) and may have limited battery life. In continuous-flow settings, oxygen is delivered throughout both inhalation and exhalation, resulting in substantial oxygen waste during exhalation. Pulse-flow settings were developed to minimize this waste by supplying oxygen only during inspiration, thereby conserving both oxygen and battery power [4]. However, if the pulse is not synchronized well with inspiration, a portion of the oxygen bolus may be exhaled before reaching the alveoli [5]. LOX supplies oxygen by gradually evaporating LOX at cryogenic temperatures and can deliver higher flow rates for longer periods. However, LOX devices are costly, and availability may be limited [6].

In previous studies, users of portable oxygen devices reported a lack of physically manageable portable systems, a lack of devices capable of delivering higher oxygen-flow rates (>3 L/min), and an inability to leave their homes for more than 2-4 hours due to lack of reliable and enduring ambulatory oxygen supply [7-9]. In addition, users may face large out-of-pocket expenses for the ongoing costs of oxygen equipment [7]. Given these limitations, people using ambulatory oxygen reported that it was important to have a variety of device options that allowed a choice based on their individual needs [9]. However, there is little information on the types of devices available, the performance of each device, and the costs of ambulatory oxygen therapy devices [7]. It is possible that technologies to deliver ambulatory oxygen therapy are available in other fields (eg, military medicine) that are not yet available as commercial ambulatory devices.

A scoping review was deemed most appropriate to investigate the current status of portable oxygen devices, as the concepts of interest (technology, performance characteristics, and cost) extend beyond the medical field and may be published outside traditional peer-reviewed medical literature (eg, patent documents, technical reports) [10]. This scoping review aimed to identify the range of available portable oxygen devices, the technologies that are used to generate oxygen by the devices, the performance of each device, and the costs associated with its use. In addition, given the limitations of current portable oxygen devices, it is clinically important to clarify the status of technological innovation. Some patients require high oxygen flow rates that exceed the capacity of the currently available portable oxygen devices [11].

Furthermore, it is recognized that the weight of portable oxygen systems may limit their usability in clinical practice [12]. This may reduce adherence and contribute to inconsistent evidence for efficacy in exertional hypoxemia [11,13]. Therefore, we also aimed to identify innovations in the design of ambulatory oxygen therapy devices, such as improvements in weight or flow rates, and pulse oximeter oxygen saturation (SpO₂)–targeted automatic titration of oxygen delivery, which may not be available commercially [14].

This scoping review aimed to identify the range of available portable oxygen devices, the technologies that are used to generate oxygen by the devices, the performance of each device, and the costs associated with its use. We also aimed to identify innovations in the design of ambulatory oxygen therapy devices, such as improvements in weight or flow rates and SpO₂-targeted automatic titration of oxygen delivery, which may not be available commercially.

This scoping review was conducted according to the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guideline (Multimedia Appendices 1-3; [15]), and the Joanna Briggs Institute (JBI) Manual for Evidence Synthesis [16]. The protocol was registered prospectively with the Open Science Framework on May 29, 2024.

We included English language documents that described a device that can deliver oxygen in an ambulatory context, including scientific papers (including review articles) and gray literature (conference proceedings, patent documents, company websites, technical reports, and government documents). We excluded documents that only described stationary equipment, high-flow nasal oxygen therapy, hyperbaric oxygen therapy, positive pressure ventilation systems, heliotherapy, short-burst oxygen therapy, and wall-based high-flow nasal oxygen therapy, as well as those limited to over-the-counter medical devices.

An initial search was performed on May 29, 2024, in MEDLINE (Ovid) and IEEE Xplore. After the initial search, the text words in the titles and abstracts of the retrieved articles and the index terms used to describe the articles were analyzed to refine the search terms. For the second search using updated terms, we searched scientific papers in 5 databases (MEDLINE [Ovid], Embase [Ovid], SCOPUS, CENTRAL [Wiley], and IEEE Xplore) using the search strings created with index terms (MeSH [Medical Subject Headings] terms or IEEE terms). Search strings are shown in Table S1 in Multimedia Appendix 4.

For searching gray literature, the following three strategies were used based on the guideline for gray literature and a previous study [17,18]: (1) targeted website browsing and searching, (2) gray literature database search, and (3) search engine searching. Specifically, technical reports, white papers including military medicine, and other gray literature on portable oxygen devices (manuals, company websites on product performance, etc) were searched using the appropriate websites respectively (National Technical Reports Library, World Health Organization, Defense Technical Information Center, International Health technology Assessment Database, and Google Advanced). As complex search strings were not allowed in some gray literature databases, we adjusted the search strings to match the database functionality. Gray literature search strings are also shown in Table S1 in Multimedia Appendix 4. Google Advanced searches were run on corporate, government, and military domains (.com, .gov, and .mil) to review the top 100 results from each. The latest patents of inventions and developments for the period January 1, 2022, to June 25, 2024, were searched in the World Intellectual Property Organization (WIPO) database using adjusted search strings. The search was repeated on September 29, 2025, for CENTRAL, MEDLINE (Ovid), and Embase (Ovid), on September 25, 2025, for all other databases including Scopus, IEEE Xplore, and gray literature sources, and on September 29, 2025, for WIPO.

A third search consisted of hand searching reference lists of all selected articles and review papers to identify any additional articles not found by the first two search methods. Additionally, Google Advanced searches were performed to obtain missing product performance information. The search period in each database was 2004 to the present to capture devices that were currently or recently available. We reported database searches and literature selection according to PRISMA-S (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Search extension) [19].

Scientific papers were imported from the databases into the Covidence platform [20], and gray literature search results were imported into a spreadsheet. In the Covidence platform, duplicate records were automatically removed. Two independent reviewers (SK and MH) screened titles and abstracts of scientific papers for eligibility. Studies that met the inclusion criteria, or for which eligibility was unclear, underwent full-text review. Gray literature records were also screened for eligibility by 2 independent reviewers (SK and MH). Disagreements in study selection were resolved by consensus or by consulting a third reviewer (AEH). The reasons for exclusion at the full-text stage are reported.

Outcome selection was guided by patient priorities for oxygen devices, identified in previous studies and are shown in Table 1 [7-9].

Data from the included sources of evidence were charted using a custom-designed form (SK and MH). Two review authors (SK and MH) independently charted the data from the eligible studies. Disagreements regarding the data charting between authors were resolved by discussion. If consensus could not be reached, a third author (AEH) reviewed the study and arbitrated.

The data chart included: types of publication (scientific paper, patent documents, technical report, etc), types of content (products, performances, costs, etc), author, year of publication, country where the information originated, information related to the products and performances outlined, and information related to the costs.

Scoping reviews are conducted to provide an overview of the existing evidence regardless of methodological quality or risk of bias [22]. Therefore, the included sources of evidence are typically not critically appraised for scoping reviews. As such, we did not undertake a quality assessment of the included sources of evidence.

To synthesize data on products and their performance, we prioritized information published by the product company (including user manual), scientific papers, white papers, technical reports, and websites, in that order, if there was different information on the same item regarding one product. If multiple products were identified as having the same product name but differing only in numbers, the newest product was checked at the official website, and only the newest one was extracted (eg, Eclipse 1/2/3/4/5 Caire, Ball Ground, United States). As stationary oxygen concentrators are generally considered to weigh 10 kg or more, and POCs range from 1-9 kg [4], we defined ambulatory oxygen devices in this study as those weighing less than 10 kg and excluded heavier devices. Performance characteristics, the technology used, and the current status of development were reported descriptively. Costs for ambulatory oxygen therapy were mapped descriptively by country and type of device (portable oxygen cylinders, POCs, and portable LOX). Descriptive statistics were used to describe all data. Categorical data were presented as frequency and percentages. In addition to descriptive synthesis, we developed a gap map to describe areas in need of future development. The gap map addressed the degree to which key patient needs for ambulatory oxygen devices (oxygen flow rate, device weight, operating time, and auto-titration) are met by characteristics of current products or those in development. Patients’ needs extracted from previous studies were used as the framework [7-9].

A total of 9702 records were identified from scientific databases (medical database: n=3720; engineering database: n=133; multidisciplinary database: n=1842), and 4007 records from gray literature. After removing duplicates, a total of 3028 records from scientific databases and 4007 records from gray literature sources were screened (Figure 1). Of these, a total of 166 records related to ambulatory oxygen therapy devices were finally included in this review (n=106 from scientific databases; n=60 from gray literature sources). Among the 166 included records, a total of 81 (48.8%) originated from North America, 37 (22.3%) from Europe, 31 (18.7%) from Asia, 11 (6.6%) from Oceania, 2 (1.2%) from Africa, and 2 (1.2%) from South America (Table S2 in Multimedia Appendix 4). The scientific literature included 85 original research articles, 20 review articles (of which 1 was a systematic review and 19 were narrative reviews), and 1 clinical guideline. The gray literature comprised 7 technical reports, 3 white papers, 3 clinical trial registrations, 14 government-related websites (eg, military, energy, and space exploration), 4 company websites, 1 charitable organization website, and 28 patents (Tables 2 and 3). Although we identified 3 chemical oxygen generators in the scientific literature, they were excluded because chemical oxygen generators exhaust in 30 minutes or less and their output cannot be adjusted, making these devices unsuitable for delivery of ambulatory oxygen in clinical care [23].

We identified 33 different POCs in this study, of which 29 were on the market, 1 had discontinued production, and 3 were prototypes under development. The summary of identified POCs is shown in Table 4. The weight of the POCs products ranged from 1.0 to 9.1 kg, and size ranged from 15.7 × 11.7 × 6.1 cm (756 cm³) to 51.3 × 27.7 × 20.3 cm (28,847 cm³). Oxygen delivery capacity varied widely, with maximum flow rate in the continuous flow setting ranging from 2.0 to 6.0 L/min and maximum pulse-dose bolus volume of pulse flow setting ranging from 17.3 to 90 mL of oxygen. The pulse-dose bolus volume varies widely across products, even within the same pulse flow setting (Figure 2). A trade-off between portability and oxygen delivery was shown in the identified POCs (Figure 3, Table S3 in Multimedia Appendix 4). Smaller POC products tended to be transported using bags, while larger POC products were transported using a dedicated cart (Table S3 in Multimedia Appendix 4). Smaller POC products also tended to have only pulse flow settings. The most recent POCs were the OxLife Liberty2 (25.4 × 22.9 × 8.9 cm), released in 2024, and the DISCOV-R (23.6 × 10.4 × 25.2 cm), released in 2023, which were capable of delivering oxygen in the continuous flow setting. This contrasts with earlier POCs of the same size (eg, Inogen Rove 6, 18.3 × 8.3 × 20.5 cm with single battery), which only had a pulse flow setting (Table S3 in Multimedia Appendix 4). The mean maximum continuous operating time in pulse flow setting was 3.8 hours, with a range of 1.3 to 6.3 hours, depending on flow settings and battery option (Table 4 and Figure 4). The maximum operating time tended to be shortened by half in the continuous flow setting compared to the same flow rate in the pulse flow setting in terms of equivalent volume (Table 4 and Table S3 in Multimedia Appendix 4). There were also POC products, such as the Simply Go mini, which extended the continuous operating time from 2 hours 40 minutes to 5 hours with an additional battery (Table S3 in Multimedia Appendix 4).

Pressure swing adsorption (PSA), vacuum pressure swing adsorption (VPSA), and vacuum pressure cycle (VPC) were the technologies used for oxygen generation (Table 4, Table S4 in Multimedia Appendix 4). Zeolites were identified as the adsorbent used in POCs. Air is composed of approximately 80% nitrogen, and PSA is a method to generate high concentrations of oxygen by selectively adsorbing nitrogen from air. In PSA, air pressurized by a compressor is sent through an adsorption column with zeolite to adsorb nitrogen and increase oxygen concentration. Then, the pressure is reduced to release the nitrogen. This adsorption and desorption cycle process is alternated in the multiple adsorption columns to continuously produce highly concentrated oxygen [24]. VPSA and VPC are adsorption methods that use a vacuum pump in combination with or instead of a compressor to reduce the pressure in the column [25]. We identified 2 POC prototypes and 1 POC product that were under development. The 4-SLPM prototype developed by TDA Research (Wheat Ridge, United States) weighed approximately 7.8 kg and measured 28 × 25 × 18 cm (Table S3 in Multimedia Appendix 4). The 4-SLPM has a maximum oxygen flow rate of 4 L/min in continuous flow setting, greater than most POCs on the market of comparable weight. The 4-SLPM used high lithium-exchanged X (LiLSX) as the adsorbent. The other PSA-based prototype, co-developed by NASA and Chart Industries (Ball Ground, United States), weighed less than 8.2 kg with a size of 35.6 × 30.5 × 20.3 cm and was capable of delivering up to 6 L/min in the continuous flow setting. There was no available information on commercially available POC innovations, such as remote control of flow settings in POCs. A POC product under development named JUNO (Roam Technologies Pty Ltd, Carlton NSW, Australia) was identified. According to the company’s product information, JUNO is an ultraportable, tankless oxygen system currently under clinical development. Despite being designed to be carried with one hand, it has a continuous flow setting ranging from 1-3 L/min with a concentration of 91% [26]. Although the measurement conditions and detailed specification are not publicly disclosed (Table S3 in Multimedia Appendix 4), the recent patent on downsizing of POCs by Roam Technologies Pty Ltd, has been identified. The patent adopts a miniaturized PSA architecture with a compact arrangement of the adsorbent layer and internal gas pathway structure to shorten flow paths and minimize dead space [27].

Oxygen cylinders and LOX are summarized in Tables 5 and 6. Table 5 and Tables S3, S4, and S5 (shown in Multimedia Appendix 4) are also provided as xlsx files in Multimedia Appendix 5. The weight of cylinders ranged from 0.3-6.5 kg, and the size ranged from 14.9 × 6.4 cm to 86.5 × 10.2 cm. The maximum continuous operating time varied from 0.3 to 5.0 hours at 2 L/min in the continuous flow setting. The continuous operating time per continuous flow setting was comparable to the POC. (Figure 4). Smaller cylinders were transported using a shoulder bag, whereas larger cylinders required trolleys or hand-drawn carts for transport. Conventional cylinders were generally filled at approximately 150 bar, whereas the Ultra Lightweight Cylinder Oxygen System (IOSVR) uses a high-pressure filling of 300 bar. This allows nearly double the amount of oxygen to be filled compared with a comparably sized type M7 cylinder (Table 6). IOSVR is made possible by the high-strength aluminum alloy (L7X), which is reinforced with carbon fiber wrapping technology [28]. LOX products weighed from 1.6-3.7 kg (filled) and had gaseous oxygen capacities ranging from 275.0-1058 L (Table 6). Many oxygen cylinders require a pressure regulator, which is attached to the cylinder’s top and works like a tap, allowing the safe adjustment of oxygen flow rate provided, in L/min [4]. In addition, some regulators support a pulse-dose delivery mode, which can extend the operating duration compared with continuous flow [4,29].

LOX had a longer continuous operating time compared to POCs and cylinders, with a mean maximum continuous operating time of approximately 7.4 hours even at 2 L/min of continuous flow setting (Figure 4). There was no available information on the pulse-dose bolus volume of pulse flow setting in LOX.

There were no available records of the cost of POCs officially disclosed by POC companies. Two identified scientific papers referred to POC prices of approximately US $2000 and US $3000 as retail costs to end users [43,44]. There were also no records of specific costs for LOX and oxygen cylinders. However, an identified article reported that LOX was approximately 4 times as expensive as standard oxygen therapy using POCs or portable cylinders [45]. Another identified article indicated that LOX systems entail higher service costs due to the need for regular home refills [46].

The gap map (Figure 5) showed that only a limited range of oxygen cylinders met patient needs for high flow rates (>3 L/min) and light weight (<2.5 kg). Although current POCs did not meet this need, a one-hand carry device capable of delivering 3 L/min was under clinical development. In all device types, few products met patient needs for continuous operating time (>5-6 hours). For auto-titration, SpO₂-targeted automatic titration was available for POCs and cylinders as an interface, and the development of automatic titration capabilities embedded into the device itself has been reported only in POCs.

This scoping review aimed to identify the range of portable oxygen devices, their oxygen generation technologies, performance characteristics, innovations in the design of ambulatory oxygen therapy devices, and cost. We identified 33 POCs, 10 oxygen cylinders, and 6 portable LOX systems. There was a trade-off between the portability (weight) and oxygen delivery capacity of POCs. The mean maximum continuous operating time of POCs was 3.8 hours. PSA with zeolite was the most common oxygen generation technology in POCs. Two POC prototypes were identified that had better oxygen delivery capacity despite being the same size as POCs on the market. LOX was the most portable and had the best continuous operating time among the devices. In terms of development and innovation, the downsizing of POCs using nanozeolite as the adsorbent, improving flow rate using RPSA and SpO₂-targeted automated oxygen titration based on to patient’s SpO₂ and physical activity were identified. Costs were rarely disclosed by manufacturers, and published data indicated that LOX is substantially more expensive than POCs or cylinders.

This study showed that high-flow oxygen delivery of 3 L/min in the continuous flow setting in POCs is only possible from commercially available POCs weighing more than 7.2 kg (eg eQuinox, Table S3 in Multimedia Appendix 4), which need to be transported on a cart. This may be a burden for patients requiring high flow rates. Furthermore, the short continuous operating time of the POCs identified in this study may also limit daily activities and community engagement. Patients have previously identified a desirable operating time for portable oxygen devices of 5-6 hours [9]. In the identified POC products, the mean maximum continuous operating time was up to 3.8 hours in the pulse flow setting. As higher flow rates shorten the continuous operating time, it is likely that fewer devices will meet patients’ needs. Although an additional battery can be used to double the operating time (eg, Simply Go), carrying the additional battery may be a burden for the patient. Furthermore, it should be noted that the numerical setting of a pulse flow does not correspond to the continuous flow rate and that the bolus volume varies between products even within the same pulse flow rate setting [47].

Among the devices, LOX had the best portability and continuous operating time, but the complicated maintenance was an obstacle. Previous studies have recognized LOX as a device with both better portability and continuous operating time [8], and have reported that LOX use is associated with better quality of life and physical activity compared to POCs and cylinders [48,49]. On the other hand, LOX requires refilling of liquid oxygen by a supplier at least 2-3 times a month, which imposes a high cost on the supplier and user [4,45]. As a result, the number of suppliers servicing LOX has been declining in the United States [8].

It was suggested that the trade-off between portability and oxygen delivery capacity is improving based on the latest products and prototypes identified. However, even in the latest POC products (DISCOV-R released in 2023 and OxLife Liberty2 released in 2024), the maximum oxygen flow rate of the continuous flow setting was still 2 L/min (Table S3 in Multimedia Appendix 4). JUNO, which is designed to be carried with one hand, is under clinical development and has a continuous flow setting ranging from 1-3 L/min with a concentration of 91% [26], and may be an innovative device, as the PSA Prototype was published in 2015 and the 4-SLPM in 2018 [50,51]. In terms of technological development, the nanozeolite (LilSX) had been of interest because of its use in 4-SLPM prototypes, patents, and basic research [30,52]. Although nanozeolites are characterized by higher nitrogen adsorption capacity, they are more expensive than conventional zeolites and have challenges such as long-term structural stability and degradation [53,54]. Although RPSA systems can improve oxygen productivity by shortening adsorption–desorption cycles, their application to POCs remains challenging because no clinical application studies have yet been reported [52]. Several recent clinical trials and patents described systems capable of adjusting flow rates based on the user’s SpO₂ and physical activity. In a clinical trial with a small sample size, automated titration improved functional capacity in ADL, dyspnea, and the number of severe hypoxemia events [37]. However, these technologies have barriers, including sensor reliability or inaccuracy even during motion, requiring further development and clinical trials [37,55]. Identified patents in this study should be carefully considered because approximately half of the patents in general are not commercialized and launched on the market [56].

Limitations of this scoping review include the following. First, it is difficult to compare performance such as pulse-dose bolus volumes and continuous operating time between POCs because they depend on the number of breaths and device settings [57]. Second, this scoping review included only information published in English, which may have underestimated information on devices developed and used in non–English-speaking countries. Third, the pulse-dose bolus volume of POC at maximum pulse flow setting (as much as possible at a respiratory rate of 20 breaths per minute) extracted in this study does not exactly indicate the product’s oxygen delivery capacity. POC products vary in their algorithms for converting pulse-bolus volume and flow rate in response to changes in respiratory rate [57]. Fourth, available evidence regarding product performance relied almost on manufacturer-reported specifications, and no available evidence on device lifespan, failure rates, or long-term durability was available. There was no information available on the costs of POC products disclosed by manufacturers in this study. Lastly, the geographical distribution of the searched literature was biased toward the United States, as some gray literature was retrieved using domains such as “.gov,” “.mil,” and “.org” in Google Advanced Search.

This scoping review is the first study to integrate medical, engineering, and gray literature evidence on ambulatory oxygen devices and map current ambulatory oxygen devices and their development status. Although prior literature has narratively described products and technologies for ambulatory oxygen therapy, no previous research has systematically investigated these products and new technologies. We identified 33 POCs, 10 oxygen cylinders, and 6 LOX systems. We showed the performance limitations of current devices and gaps in technological development in ambulatory oxygen therapy and suggested directions for future research and development. Specifically, POCs available to consumers may not meet patients’ needs in terms of oxygen flow rate, portability, and operating time. LOX offered the best performance in terms of operating time and portability among the devices but is restricted by high costs and declining availability. Although POCs are the most widely developed devices, technological innovation to achieve high oxygen flow rates, better portability, and longer continuous operating time has been limited since the POC prototype published in 2015. Collaboration among device developers, researchers, health professionals, and patients is urgently required to develop new lightweight devices with greater oxygen delivery capacity. It is essential to incorporate consumer input from the early stages of design and testing to ensure that future portable oxygen devices meet patients’ needs.