Feasibility and Acceptability of Artificial Intelligence-Driven Video and Digital Auscultation Tools for Identifying Increased Work of Breathing in Young Children with Acute Lower Respiratory Infections in Rural Bangladesh

Feasibility and Acceptability of Artificial Intelligence-Driven Video and Digital Auscultation Tools for Identifying Increased Work of Breathing in Young Children with Acute Lower Respiratory Infections in Rural Bangladesh | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Feasibility and Acceptability of Artificial Intelligence-Driven Video and Digital Auscultation Tools for Identifying Increased Work of Breathing in Young Children with Acute Lower Respiratory Infections in Rural Bangladesh Tamanna Sharmin, Salahuddin Ahmed, Ahad Mahmud Khan, Tajkia Rumman, and 9 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7648403/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 4 You are reading this latest preprint version Abstract Background Acute lower respiratory infections (ALRIs) are a leading cause of child mortality globally, with timely recognition of increased work of breathing (WOB) critical for early intervention. In low-resource settings, WOB assessment is often subjective and inconsistent, contributing to delays in care. Digital health tools using artificial intelligence (AI) offer promising solutions to standardize detection of increased WOB through video and audio analysis. We are developing an AI-based tool to support assessment of increased WOB in children. This pilot study explores its feasibility and acceptability before its prototype is developed. Methods This qualitative study was conducted in March–April 2025 in Zakiganj, Sylhet, Bangladesh. Data were collected through four focus group discussions and four in-depth interviews with caregivers, community health care providers (CHCPs), community leaders, policymakers and health administrators. They followed a semi-structured guide tailored to participant roles. Participants were verbally introduced to the concept of an AI-based tool for assessing WOB in children to guide discussion; no demonstrations or recordings were used. Results The findings demonstrate broad support for the feasibility of implementing AI-supported tools - digital stethoscopes and video-based respiratory assessments—for identifying increased WOB in children, particularly in rural and resource-constrained settings. Stakeholders emphasized that these tools offer practical solutions to address critical gaps in skilled personnel and diagnostic infrastructure, with strong endorsement for their use by CHCPs to facilitate task-shifting. Prior experience with similar technologies, such as digital stethoscopes for pneumonia, further reinforced their confidence in feasibility. Policymakers noted its alignment with national digital health strategies and child mortality reduction goals, indicating potential for scale-up through pilot initiatives and public-private partnerships. Caregivers also expressed positive perceptions, highlighting improved diagnostic accuracy, enhanced understanding through visual and audio feedback, and greater accessibility at local clinics. Nevertheless, concerns were raised about child cooperation, need for trained operators, time constraints, equitable access, and trust in technology. Conclusion AI-driven video and digital auscultation tools are considered feasible and acceptable for assessing increased WOB in children with ALRIs in rural Bangladesh. Their adoption will require training, technical support, community trust-building, and equitable access, with potential for scale-up to strengthen child health services. Figures Figure 1 Figure 2 Figure 3 Background Acute lower respiratory infections (ALRIs) represent a major global health challenge in children under five years of age. Pneumonia and bronchiolitis are the most common ALRIs and are among the leading causes of morbidity, hospitalization, and mortality in this age group 1 . Each year, approximately 22 million cases of severe ALRIs occur globally, resulting in around 16.4 million hospital admissions 2 . These infections cause nearly 700,000 deaths annually among children under five, accounting for approximately 13% of global child mortality 3 . ALRIs are caused by a variety of viral and bacterial pathogens. Respiratory syncytial virus (RSV), influenza viruses, and parainfluenza viruses are among the most frequent viral etiologies 4 . RSV itself is responsible for 50–80% of bronchiolitis cases and 30–60% of pneumonia cases in younger children. RSV is also the leading cause of hospital admissions for respiratory illness in children. Each year, RSV is associated with over 118,000 deaths among children under five years old, with 99% of these deaths occurring in low- and middle-income countries (LMICs), largely due to limited access to timely and appropriate healthcare 4 . Severe ALRIs during infancy can also have long-term effects on lung health, increasing the risk of developing chronic conditions such as asthma, chronic obstructive pulmonary disease (COPD), and even early death later in life 5 . Several risk factors contribute to the high burden of ALRIs in young children in LMICs. These include malnutrition, exposure to indoor air pollution, particularly from cooking with solid fuels, overcrowded living environments, poor sanitation, and the lack of exclusive breastfeeding during the first six months of life 1 . In Bangladesh, ALRIs remain one of the biggest threats to young children’s health, responsible for approximately 18% of all deaths in children under five years old 6 . This burden is especially pronounced in rural areas, where healthcare access is limited and delays in care-seeking is common. Although overall care-seeking behavior among families is relatively high, only approximately 42–46% of children with ALRI symptoms receive timely treatment from qualified healthcare providers, which significantly increases the risk of severe illness and death 7 . One of the major clinical features of ALRIs is increased work of breathing (WOB), which serves as a key indicator for the severity of the infection. 8 Recognizing and accurately assessing increased WOB is crucial for timely intervention and treatment. However, the subjective nature of clinical signs reflecting increased WOB and its severity (e.g., does the child have mild or significant indrawing of skin at the intercostal or subcostal areas), along with the varying levels of expertise among healthcare providers, often leads to inconsistencies in identification and delays in treatment. 9 Emerging technologies like artificial intelligence (AI) offer advances in improving the identification of increased WOB in young children. AI-driven tools using video recordings can objectively analyze chest movements to detect abnormal breathing patterns. 10 Similarly, digital stethoscopes equipped with an AI algorithm can identify adventitious lung sounds such as wheeze and crackles. 11 – 13 These tools can make it easier to identify breathing problems, reduce human error, and support faster and more reliable identification and treatment of ALRIs. To address existing identification challenges, a novel AI-based tool is being developed to help standardize the identification of increased WOB in young children. This tool will use video and digital stethoscope recordings to assess increased WOB and can potentially be expanded to older children. Our pilot study explored how feasible and acceptable the idea of this AI tool (which is still being developed) is to frontline health workers, carers and stakeholders when being used in community settings in Bangladesh. Methods Study Design This qualitative study used a narrative research approach, as described by the Clandinin and Connelly (2000), where narrative inquiry is viewed as “a way of understanding experience” and involves studying the lived experiences of individuals as they are expressed in stories. The researcher and participant often engage collaboratively in co-constructing the narrative. 14 Data were collected through Focus Group Discussions (FGDs) and In-Depth Interviews (IDIs). FGDs provided insights into shared perceptions and social dynamics influencing participants’ acceptance of new diagnostic tools, while IDIs allowed for detailed exploration of individual experiences, professional opinions, and policy-level considerations. Study Site and Duration This study was conducted at the Projahnmo surveillance area in Zakiganj sub-district, Sylhet district, Bangladesh. The site was established in 2001 through a partnership between Johns Hopkins University, the Bangladesh Ministry of Health and Family Welfare (MOHFW), Bangladeshi academic institutions and non-government organizations (NGOs) including Projahnmo Research Foundation. The study was conducted from March to April 2025. The location map of the study site is shown in Fig. 1 . Study Population For the FGDs, the study population were the caregivers (male and female) of children under 2 years old, community health care providers (CHCPs), and community leaders. CHCPs are trained health workers, and staff of Community Clinics (CCs)—small, government-established health centers at the village level. Each community clinic typically serves about 6,000 people. CHCPs are responsible for providing basic health services. Community leaders are individuals who hold influence, authority, or respect within their local communities and often play critical roles in social mobilization, dispute resolution, and local development. Under-2 children’s caregivers were divided into male and female FGD groups separately to encourage more open and honest discussion. In many cultural contexts, especially those with traditional gender roles, participants may feel uncomfortable or restrained discussing sensitive topics—such as child health, caregiving practices, or household dynamics—in mixed-gender settings. By conducting gender-specific groups, the study aimed to create a safe and comfortable environment where participants could express their views freely without fear of judgment, social pressure, or embarrassment. Four IDIs were conducted with key stakeholders, including a national-level policymaker, a national-level policymaker and implementer, a division-level health administrator, and an upazila (subdistrict)-level health administrator. The study used purposive sampling to select participants with relevant knowledge and experience regarding child health and healthcare service delivery. Data Collection Procedures Data collection was conducted for the four FGDs and four IDIs. Each FGD lasted between 1.5 to 2 hours, while IDIs lasted approximately 30–45 minutes. A semi-structured interview guide was used across all FGDs and IDIs to ensure consistency in topics while allowing flexibility to explore emerging themes. The guides were tailored to each participant group (e.g., we would ask carers about their experiences in bringing their children with respiratory symptoms to the community clinics and ask CHCPs about their experiences in treating these children). Prior to the discussions, participants received a verbal explanation (detailed below) of the proposed idea of AI-based tool to ensure conceptual understanding. These explanations were not part of the data collection, and the AI tool hasn’t been developed yet. The proposed AI tool included: Digital stethoscope : To record lung sounds at four chest positions (two front, two back), each lasting ~ 10 seconds (total ~ 60 seconds), capturing 3–4 full breathing cycles (inspiration and expiration) per site. An example of a digital stethoscope is shown in Fig. 2 . Video image recording : A 60-second video of the child’s breathing from nose to lower chest wall, potentially captured via mobile phone. An example of video-based recording of chest movements is shown in Fig. 3 . Research Team and Training A qualitative researcher led data collection with a background in anthropology and experience in conducting qualitative methods, including FGDs, IDIs, data analysis, and reporting. A Research Assistant (RA) with a social science background and some research experience supported data collection. The RA received specific training and was supervised by the qualitative researcher. TS and AMK conducted an orientation session on the study protocol for the research team. Ethical Considerations All participants provided written informed consent before participation. All discussions and interviews were conducted in Bangla, and audio-recorded with participant permission. The study received ethical approval from Edinburgh Medical School Research Ethics Committee (EMREC) with reference number 24-EMREC-078 and Projahnmo Research Foundation Institutional Review Board (PRF IRB) with reference number PR-25001. Data Management and Analysis Audio recordings were transcribed verbatim and translated into English. Data were analyzed using thematic analysis, identifying key themes and sub-themes. The process began with a thorough, line-by-line reading of the data, during which initial codes were assigned to segments that appeared meaningful or relevant. These codes were then reviewed to identify connections and group similar ideas together. Through this process, broader patterns emerged, leading to the development of potential themes that captured significant aspects of the data in relation to the research question. This approach facilitated the transition from raw data to more organized and interpretable findings. A matrix table was developed to organize, compare, and analyze data across participant groups. Grounded Theory was used for this study as it enables the inductive development of a conceptual model grounded in the perspectives and experiences of healthcare providers, caregivers, and stakeholders regarding the implementation of AI-driven video and digital auscultation tools in diagnosing respiratory distress in young children. Given the limited existing research on the adoption of such technologies in low-resource settings like Zakiganj, Sylhet, this approach allows for the identification of emergent themes and processes related to feasibility and acceptability. By systematically analyzing qualitative data, Grounded Theory facilitates a nuanced understanding of contextual factors and user interactions, thereby informing strategies for effective integration and broader application of these innovative diagnostic tools. Results The baseline characteristics of the included participants were available in Table 1 . FGDs included participants of caregivers (20 participants, 9 females and 11 males), community healthcare providers (10 participants) and community leaders (10 participants). IDIs included participants of key policymakers and senior health officials (4 participants in total). Table 1 Summary of study participants Participant type Method Number of FGDs/IDIs Total participants Characteristics Caregiver (female) FGD 1 9 Age: -Range 20–27 years - Mean 23 years Education: -Secondary (n = 7) -Higher secondary (n = 1) -Graduate (n = 1) Occupation: -Housewife (n = 9) Caregiver (male) FGD 1 10 Age: -Range 25–32 years -Mean 29 years Education: -Illiterate (n = 1) -Primary (n = 4) -Secondary (n = 3) -Higher secondary (n = 1) -Post-graduation (n = 1) Occupation: - Mostly farmers with small business - Others include rickshaw puller, small tea stall owners, small vegetable sellers CHCPs FGD 1 10 Education: Completed Higher Secondary Certificate or an equivalent qualification Community leaders FGD 1 10 Education: Higher Secondary (n = 4) Graduation (n = 4) Post-graduation (n = 2) Policymakers and senior health officials IDI 4 4 Education: MBBS (n = 3), Post-graduation in Pediatrics (n = 1) FGD – focus group discussion; IDI – In-depth interview Theme 1: Supplement for Systemic Gaps Participants of IDIs consistently emphasized that shortages in skilled personnel and diagnostic infrastructure pose major challenges in rural healthcare delivery. In such contexts, AI tools were considered feasible solutions that could support clinical decision-making where radiographic imaging, laboratory testing, or pediatric specialists are unavailable. These tools were seen not as replacements for physicians, but as supplementary aids to bridge critical service gaps. "In remote clinics, doctors aren’t always present, and we lack tools like X-rays. If AI can help detect danger signs early, it could make a real difference." — Health administrators at Upazila-level Theme 2: Task-Sharing and Human Resource Utilization The potential for task-sharing was highlighted as a practical advantage. Division-level Health Administrator noted that with minimal but targeted training, nurses, CHCPs, and paramedics could effectively operate the tools and interpret outputs. This redistribution of responsibilities was seen as a feasible and scalable strategy to alleviate pressure on physicians, especially during peak patient loads. “If experienced nurses or CHCPs can be trained, they can handle the tool. It’s better than waiting for a doctor who may not be there.” — Division-level Health Administrator Theme 3: Operational Feasibility in Existing Infrastructure CHCPs described how minor logistical adjustments—such as creating quiet corners or designated rooms in busy clinics—could allow effective use of the AI tools. The simplicity of the proposed technology (e.g., mobile-based video, portable stethoscopes) was considered an advantage, aligning with existing infrastructure constraints. “We don’t need a separate room. A quiet space in the clinic would be enough for recording.” — CHCP Theme 4: Previous Experience Increases Confidence The Division-level Health Administrator referred to earlier successful initiatives using digital stethoscopes in community settings to detect pneumonia. This prior familiarity with similar technologies gave them confidence in the feasibility of implementing an AI-enhanced system for identifying increased WOB. “We’ve seen this work before in pneumonia studies. This is not entirely new for us.” — Health Administrator at Upazila (subdistrict) level Theme 5: Scalability with Structured Training and Supervision CHCPs emphasized that while the tools are technologically feasible, effective rollout would require structured training modules, ongoing technical support, and clear protocols for use. With these systems in place, participants believed the tools could be sustainably integrated into routine service delivery. “Training is key. If CHCPs are supported and supervised, they can do it well.” — Health Administrator at Upazila (subdistrict) level Theme 6: Alignment with Policy Goals Policymakers noted that AI-supported tools aligned with national goals to strengthen digital health, improve early detection, and reduce child mortality. The feasibility of implementation was seen as high, particularly if supported by donor funding, public-private partnerships, or pilot programs to test scalability. “This fits with our digital health strategy. If it proves effective, we can advocate for wider adoption.” — National-level Health Policy makers Perceived Feasibility and Potential Benefits Improved Identification Accuracy and Early Detection Caregivers viewed the AI-supported tool as a promising addition to existing methods of identifying increased WOB, particularly for respiratory illnesses like pneumonia. They hoped that the device’s combination of audio (lung sounds) and video (visual assessment of breathing patterns) would enhance identification accuracy and support timely detection of illness, especially in settings where access to experienced doctors is limited. "With this device, maybe they can catch the illness early—even before it becomes severe." — Female caregiver Visual and Audio Feedback Enhancing Understanding and Trust Participants appreciated the tool’s potential to provide visual and audio evidence of a child’s condition. This was seen as a key strength that could help bridge communication gaps between healthcare providers and caregivers, who often struggle to understand the severity of respiratory conditions. Seeing or hearing a problem directly was expected to build trust and encourage adherence to treatment or referral advice. "Sometimes we don’t understand what the doctor says. But if we see the problem ourselves, we will take it seriously." — Male caregiver Support for Local Clinics and Health Workers Caregivers supported the use of AI tools at community clinics and by local health workers, particularly CHCPs. They saw the tool as a way to extend identification capabilities to the local level, potentially reducing unnecessary travel and promoting early care-seeking. However, they emphasized the need for trained personnel to operate the device to ensure correct use. "If it’s available in the village clinic and they know how to use it, it will save us time and money." — Female caregiver Perceptions of the Proposed Identification Device The female caretakers expressed a clear preference for the proposed device over traditional methods, indicating that the new technology would provide more accurate and reliable assessments of a child’s respiratory condition. They described limitations with the current approach, where village doctors often rely on stethoscopes and subjective judgment, which may lead to inappropriate treatment such as unnecessary antibiotic prescriptions or failure to recognize severity. The device’s capability to objectively analyze and provide instant feedback on increased WOB, assisting in clinical decision-making on whether a child requires hospital referral or home treatment was considered highly beneficial. Concerns and Constraints Affecting Feasibility Child Discomfort and Cooperation Challenges A common concern raised by caregivers was the practical difficulty of keeping young children calm during the test. They noted that unfamiliar devices, especially if cold or metallic, might distress the child. Caregivers suggested breastfeeding or soothing the child beforehand as potential strategies. "Babies don’t sit still. If the device is cold or scary, they’ll cry and move too much." — Female caregiver Need for Skilled Operation and Interpretation Caregivers strongly felt that trained healthcare workers—not family members—should operate the device. They believed that improper use by untrained individuals could lead to misinterpretation or misuse, reducing the tool’s effectiveness and possibly causing harm or unnecessary worry. "Only trained people should use it. We don’t understand how it works." — Male caregiver Time and Patience Required During Use Some participants noted that the test process might take longer than usual checkups, requiring patience from both caregivers and health workers, especially if multiple children are being assessed. This could be challenging in crowded clinics or for caregivers in a hurry. "If the process takes time, it should be explained. Otherwise, people may get frustrated or leave." — Female caregiver Concerns About Availability and Equity While caregivers supported the tool’s use, they expressed concerns that it may not be equitably accessible, especially in remote or under-resourced clinics. Participants emphasized the importance of widespread distribution and sufficient training across facilities, so all children—regardless of location—could benefit. "We hope it’s not only in big clinics. Even the small ones in our villages should have it." — Male caregiver Trust and Cultural Acceptance Although many caregivers were optimistic, a few raised concerns about trust in technology replacing traditional physical examinations. They valued the human touch and reassurance that comes from a provider physically examining the child and cautioned that community education and demonstration would be needed to build comfort and trust in the tool. "We believe what the doctor says when he checks with his hand. A machine might not feel the same." — Female caregiver Discussion This qualitative study explored the perceived feasibility and acceptability of AI-supported tools—specifically a digital stethoscope and video-based respiratory assessment—for the early detection of increased WOB in children under two years old in rural Bangladesh. Drawing on perspectives from caregivers, healthcare providers, and policymakers, the findings indicate broad optimism regarding the potential of such tools to enhance identification accuracy, improve early detection, and strengthen decentralized care. However, several important contextual and operational factors must be addressed to ensure successful implementation. Across all respondent groups, participants recognized the potential of AI-assisted tools to support early identification of increased WOB, particularly in settings where healthcare infrastructure and skilled personnel are limited 15 . Caregivers believed that visual and audio outputs could improve their understanding of a child's condition and encourage adherence to treatment or referral advice. These perceptions reflect growing trust in digital health solutions and echo findings from earlier studies in Bangladesh demonstrating caregiver receptiveness to digital tools when linked to better outcomes and clearer communication 13 . Healthcare providers emphasized the value of AI tools in bridging critical human resource gaps and reducing dependence on clinical expertise alone. By supporting task-shifting to community health workers and paramedics, these tools were seen as a means to decentralize care and reduce unnecessary referrals, thereby improving system efficiency. Policymakers shared this view, highlighting alignment with national digital health priorities and suggesting that such innovations could support ongoing efforts to reduce child mortality and improve health equity. Importantly, the visual nature of the tool—particularly the video capture of breathing—was described by caregivers and providers as a unique strength. It was viewed not only as an identification aid but also as a communication tool that empowers caregivers to better engage in their child’s care. This finding suggests that digital tools that offer tangible, understandable feedback may enhance patient-provider interactions and shared decision-making in low-literacy settings. Despite strong enthusiasm, the study also revealed concerns that must be addressed to ensure feasibility. These include logistical challenges such as child discomfort during testing, environmental noise in busy clinics, and the need for dedicated quiet spaces. Moreover, while the tools themselves are non-invasive and relatively easy to operate, participants emphasized the importance of proper and ongoing training, technical supervision, and support systems for health workers. A one-time training intervention was viewed as insufficient for maintaining quality over time, especially in rural clinics already facing staff shortages and high turnover. A recurring theme was the tension between technological innovation and the perceived need for human touch. Some caregivers expressed concern that reliance on devices may undermine trust traditionally built through direct physical examination by a physician. This underscores the need for careful integration of AI tools within existing care models, ensuring that technology augments rather than replaces the human elements of care. Furthermore, participants raised critical issues around scalability and sustainability. Policymakers noted that widespread adoption would require robust monitoring systems, validation studies in real-world conditions, and strong evidence of cost-effectiveness. Infrastructure constraints such as limited electricity and unreliable internet connectivity in some clinics may also limit short-term scalability. Moreover, while the study explored the potential of an AI-supported tool to strengthen community health service delivery, its actual implementation revealed a range of significant challenges. Although the tool showed promise in theory—offering efficiencies in data analysis, decision support, and resource allocation—financial constraints quickly surfaced as a major barrier. Many upazila- and union-level facilities lack digital literacy, and consistent internet access needed to support such technology. Moreover, the introduction of AI-based decision-making tools altered existing workflows and responsibilities, potentially leading to unintended tensions among health workers. For example, task shifting—where AI tools assisted CHCPs in triaging or clinical decision-making—was met with resistance or confusion from other cadres like Health Assistants or Family Welfare Assistants, who perceived the tool as either undermining their roles or increasing their workload. This underscores a broader issue: while AI solutions may appear effective in the abstract, successful implementation requires addressing ground-level limitations, ensuring buy-in from frontline workers, and carefully managing changes to established practices and hierarchies. However, most of the participating caregivers, CHCPs and community leaders have lower educational qualifications, and all participating caregivers are relatively young which might have biased the result. Implications for Policy and Practice The insights generated in this study have several important implications for policy and practice. First, the introduction of AI tools should be accompanied by structured training programs and long-term technical support. Second, implementation efforts should prioritize community engagement, including sensitization campaigns and trust-building through transparent communication. Third, health system readiness—including infrastructure, staffing, and support mechanisms—must be assessed prior to deployment. To foster sustainability, integration with existing government health systems, rather than parallel implementation by external actors, is critical. Additionally, policies ensuring equitable access, data privacy, and ethical use must be developed in consultation with frontline users and affected communities. Declarations Competing interests The authors declare no competing interests. Funding TS acknowledged the Chancellor’s Fellowship Starting Fund from University of Edinburgh. Author Contribution TS, SA, AMK conceptualised and designed this study. Tamanna S collected and analysed data. Tamanna S wrote the first draft of the manuscript. SA, TS, AMK provided critical feedback on the manuscript structure and content. All authors contributed to the interpretation of the results, reviewed the manuscript, and approved the submitted version. Acknowledgement The authors extend their deepest gratitude to the Ministry of Health and Family Welfare, Government of Bangladesh, National Newborn Health Program, and CHCPs of the respective community clinics in Zakiganj, Sylhet, for their invaluable assistance in supporting activities for this research. References Walker CLF, Rudan I, Liu L et al (2013) Global burden of childhood pneumonia and diarrhoea. Lancet 381(9875):1405–1416 Nair H, Simões EA, Rudan I et al (2013) Global and regional burden of hospital admissions for severe acute lower respiratory infections in young children in 2010: a systematic analysis. Lancet 381(9875):1380–1390 United Nations Inter-agency Group for Child Mortality Estimation (UN IGME) (2023) Levels & Trends in Child Mortality 2023. UNICEF, New York: Shi T, McAllister DA, O'Brien KL et al (2017) Global, regional, and national disease burden estimates of acute lower respiratory infections due to respiratory syncytial virus in young children in 2015: a systematic review and modelling study. Lancet 390(10098):946–958 Sebastin, Kenmoe Etienne Atenguena Okobalemba Association between early viral lower respiratory tract infections and subsequent asthma development. World J Crit Care Med 2022 Jul 9; 11(4): 298–310 World Health Organization. Pneumonia (2023) https://www.who.int/news-room/fact-sheets/detail/pneumonia (accessed July 31 2025) UNICEF, Pneumonia (2023) The forgotten child killer. https://www.unicef.org/stories/pneumonia-forgotten-child-killer (accessed 12 August 2025) Rambaud-Althaus C, Althaus F, Genton B, D'Acremont V (2015) Clinical features for diagnosis of pneumonia in children younger than 5 years: a systematic review and meta-analysis. Lancet Infect Dis 15(4):439–450 Wallihan R, Ramilo O (2014) Community-acquired pneumonia in children: current challenges and future directions. J Infect 69(Suppl 1):S87–90 Khan AM, Ahmed S, Chowdhury NH et al (2022) Developing a video expert panel as a reference standard to evaluate respiratory rate counting in paediatric pneumonia diagnosis: protocol for a cross-sectional study. BMJ open 12(11):e067389 Ahmed S, Mitra DK, Nair H et al (2022) Digital auscultation as a novel childhood pneumonia diagnostic tool for community clinics in Sylhet, Bangladesh: protocol for a cross-sectional study. BMJ open 12(2):e059630 Ahmed S, Sultana S, Khan AM et al (2022) Digital auscultation as a diagnostic aid to detect childhood pneumonia: A systematic review. J global health 12:04033 Joarder T, Tune SNBK, Islam AA et al (2023) End-user acceptability of a prototype digital stethoscope to diagnose childhood pneumonia-a qualitative exploration from Sylhet, Bangladesh. BMC Digit Health 1(1):26 Clandinin DJ, Connelly FM (2000) Narrative Inquiry: Experience and Story in Qualitative Research. Jossey-Bass., San Francisco, CA Karthika M, Sreedharan JK, Shevade M, Mathew CS, Ray S (2024) Artificial intelligence in respiratory care. Front Digit Health 6:1502434 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 26 Sep, 2025 Editor assigned by journal 19 Sep, 2025 Submission checks completed at journal 19 Sep, 2025 First submitted to journal 18 Sep, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7648403","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":517635065,"identity":"9d4df05b-94e1-4e3b-9bea-1725edee4024","order_by":0,"name":"Tamanna Sharmin","email":"","orcid":"","institution":"Projahnmo Research Foundation","correspondingAuthor":false,"prefix":"","firstName":"Tamanna","middleName":"","lastName":"Sharmin","suffix":""},{"id":517635068,"identity":"bd2e76ac-1569-4373-bdac-2b4b7b169cba","order_by":1,"name":"Salahuddin Ahmed","email":"","orcid":"","institution":"Projahnmo Research Foundation","correspondingAuthor":false,"prefix":"","firstName":"Salahuddin","middleName":"","lastName":"Ahmed","suffix":""},{"id":517635069,"identity":"93dc0a84-6c1a-40e1-a5fd-02a888a8fbd6","order_by":2,"name":"Ahad Mahmud Khan","email":"","orcid":"","institution":"Projahnmo Research Foundation","correspondingAuthor":false,"prefix":"","firstName":"Ahad","middleName":"Mahmud","lastName":"Khan","suffix":""},{"id":517635073,"identity":"ef04c74a-ac7c-45fe-91ca-9fbbe4aaaf97","order_by":3,"name":"Tajkia Rumman","email":"","orcid":"","institution":"Projahnmo Research Foundation","correspondingAuthor":false,"prefix":"","firstName":"Tajkia","middleName":"","lastName":"Rumman","suffix":""},{"id":517635074,"identity":"56d69bee-ff21-4ffe-87d6-5e4ba68fc6bb","order_by":4,"name":"Nighat Sultana","email":"","orcid":"","institution":"Projahnmo Research Foundation","correspondingAuthor":false,"prefix":"","firstName":"Nighat","middleName":"","lastName":"Sultana","suffix":""},{"id":517635075,"identity":"3bc6cd7b-79d5-43f5-88ec-303d11aecfd0","order_by":5,"name":"Rezwana Tabassum","email":"","orcid":"","institution":"Projahnmo Research Foundation","correspondingAuthor":false,"prefix":"","firstName":"Rezwana","middleName":"","lastName":"Tabassum","suffix":""},{"id":517635076,"identity":"2e45e6b7-544c-45e2-b0f4-16df2bd79622","order_by":6,"name":"Shohana Shahreen","email":"","orcid":"","institution":"Projahnmo Research Foundation","correspondingAuthor":false,"prefix":"","firstName":"Shohana","middleName":"","lastName":"Shahreen","suffix":""},{"id":517635077,"identity":"eae4dcbe-27a5-4a54-8e71-9128730488f0","order_by":7,"name":"Sumyta Rahman","email":"","orcid":"","institution":"Projahnmo Research Foundation","correspondingAuthor":false,"prefix":"","firstName":"Sumyta","middleName":"","lastName":"Rahman","suffix":""},{"id":517635078,"identity":"67b1ea4b-972d-4679-a632-53efc1329737","order_by":8,"name":"Jaime Garcia Iglesias","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Jaime","middleName":"Garcia","lastName":"Iglesias","suffix":""},{"id":517635079,"identity":"01450ef6-0b5f-43cf-b275-b492b6c46ee1","order_by":9,"name":"Mohsen Khadem","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Mohsen","middleName":"","lastName":"Khadem","suffix":""},{"id":517635081,"identity":"1e0929eb-e653-49c0-baf3-3966591b82c7","order_by":10,"name":"Chun Lin","email":"","orcid":"","institution":"University of Edinburgh","correspondingAuthor":false,"prefix":"","firstName":"Chun","middleName":"","lastName":"Lin","suffix":""},{"id":517635082,"identity":"95ddb5d6-fd13-4d46-8618-e32448b1d523","order_by":11,"name":"Eric D McCollum","email":"","orcid":"","institution":"Johns Hopkins University","correspondingAuthor":false,"prefix":"","firstName":"Eric","middleName":"D","lastName":"McCollum","suffix":""},{"id":517635084,"identity":"e5056326-c23a-4280-b003-92baa0ce51f8","order_by":12,"name":"Ting Shi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAy0lEQVRIiWNgGAWjYBAC+8Ng6gAPPwODAZgpQUgLGzNUi2QD0Vog1AEGgwNEa2HnPfyC4c8dGePjzRsYftQwJM5sIOgwvjQLxrZnPGZnjhUw9hxjSJxN0BZmHjMDxobDPGY3cgwYeBsYEucRpYXhz2Ee4/lvDBj/EqnF+AED22EeAwkeA2aQLUQ5jCGx7TCPxJm0gsMyxySMCXuf/4zxhw9/Dtvztx/e+PBNjY3sjAOErAHqkkiAsg4QEStgwPyBKGWjYBSMglEwcgEACr45JafMXUsAAAAASUVORK5CYII=","orcid":"","institution":"University of Edinburgh","correspondingAuthor":true,"prefix":"","firstName":"Ting","middleName":"","lastName":"Shi","suffix":""}],"badges":[],"createdAt":"2025-09-18 10:23:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7648403/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7648403/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":92069232,"identity":"04034fec-57ac-4c0c-aac0-f8e2bccb581d","added_by":"auto","created_at":"2025-09-24 09:30:45","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3444170,"visible":true,"origin":"","legend":"","description":"","filename":"FeasibilitystudyPRFv5170925clean.docx","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/fd08cef131d73b51aab12d0e.docx"},{"id":92069214,"identity":"a061e57a-87ea-42eb-aa3f-60521374692c","added_by":"auto","created_at":"2025-09-24 09:30:44","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14272,"visible":true,"origin":"","legend":"","description":"","filename":"cc0763ae800c40129a5ca74f909c0e17.json","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/8d5e60364e19b3f9aaeb57cf.json"},{"id":92069186,"identity":"48be9246-a338-48ed-af10-4c4c5e9fbf34","added_by":"auto","created_at":"2025-09-24 09:30:41","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":68791,"visible":true,"origin":"","legend":"","description":"","filename":"cc0763ae800c40129a5ca74f909c0e171enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/4cf37c88000e89f92a4f8d96.xml"},{"id":92069339,"identity":"83b597b2-2c2b-4e72-b801-3efb172e1a4b","added_by":"auto","created_at":"2025-09-24 09:30:51","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":100926,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/f5ae99558dd6f2874ae2a27a.png"},{"id":92069226,"identity":"b0dd6f03-9d56-46e9-969a-f85e47d755a7","added_by":"auto","created_at":"2025-09-24 09:30:44","extension":"jpeg","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55726,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/676b6a30588d4c85464f597a.jpeg"},{"id":92069388,"identity":"78ecc21f-b037-4d2a-8799-5586bd2e2dcc","added_by":"auto","created_at":"2025-09-24 09:30:52","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3226096,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/d576bc28261b6a3a9aa3914c.png"},{"id":92069295,"identity":"57d68369-c273-4bd3-acf0-c47fefcd8aa1","added_by":"auto","created_at":"2025-09-24 09:30:50","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":38645,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/2f91d43caac200099c1b169d.png"},{"id":92069291,"identity":"9f6a1a77-da85-4c38-b12d-b12cec32d75d","added_by":"auto","created_at":"2025-09-24 09:30:49","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":37041,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/42b3bc6594fd84fec3077689.png"},{"id":92069342,"identity":"56bbce5b-52eb-4965-9a91-0fa830ae2f86","added_by":"auto","created_at":"2025-09-24 09:30:51","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":312203,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/50a92ac44ce0c10904636d25.png"},{"id":92069340,"identity":"224f3a4c-ede3-4e9a-a981-c56c10f85967","added_by":"auto","created_at":"2025-09-24 09:30:51","extension":"xml","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":66586,"visible":true,"origin":"","legend":"","description":"","filename":"cc0763ae800c40129a5ca74f909c0e171structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/25dd204ddaa9768ddaccc3b0.xml"},{"id":92069385,"identity":"700bebc2-0728-411c-a634-6f6d4b289e09","added_by":"auto","created_at":"2025-09-24 09:30:52","extension":"html","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":77258,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/e219c94da391a0b5e74d5664.html"},{"id":92069175,"identity":"bc897178-440e-4e5a-b493-5295e9915096","added_by":"auto","created_at":"2025-09-24 09:30:40","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":192662,"visible":true,"origin":"","legend":"\u003cp\u003eStudy site in Zakiganj Upazila, Sylhet District, Bangladesh.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/1b3db1f4d9d4bcfeb674db38.png"},{"id":92069199,"identity":"de443475-80fc-4ee5-92ca-8886f23e6c31","added_by":"auto","created_at":"2025-09-24 09:30:42","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":29530,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAn example of digital stethoscope\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/5e6ff682176fc36abc6c1cc0.jpg"},{"id":92069161,"identity":"b5d98480-8923-49d7-96ac-4a315b8308f3","added_by":"auto","created_at":"2025-09-24 09:30:38","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":144494,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAn example of video-based recording of chest movements in a child\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/8dafe281349d8773db8e2014.png"},{"id":92070075,"identity":"72d3b289-4e99-4643-a842-8819bb4fb390","added_by":"auto","created_at":"2025-09-24 09:38:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1396897,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7648403/v1/5e655154-695c-4496-9059-7500c6a43c18.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Feasibility and Acceptability of Artificial Intelligence-Driven Video and Digital Auscultation Tools for Identifying Increased Work of Breathing in Young Children with Acute Lower Respiratory Infections in Rural Bangladesh","fulltext":[{"header":"Background","content":"\u003cp\u003eAcute lower respiratory infections (ALRIs) represent a major global health challenge in children under five years of age. Pneumonia and bronchiolitis are the most common ALRIs and are among the leading causes of morbidity, hospitalization, and mortality in this age group\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Each year, approximately 22\u0026nbsp;million cases of severe ALRIs occur globally, resulting in around 16.4\u0026nbsp;million hospital admissions\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. These infections cause nearly 700,000 deaths annually among children under five, accounting for approximately 13% of global child mortality\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eALRIs are caused by a variety of viral and bacterial pathogens. Respiratory syncytial virus (RSV), influenza viruses, and parainfluenza viruses are among the most frequent viral etiologies\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. RSV itself is responsible for 50\u0026ndash;80% of bronchiolitis cases and 30\u0026ndash;60% of pneumonia cases in younger children. RSV is also the leading cause of hospital admissions for respiratory illness in children. Each year, RSV is associated with over 118,000 deaths among children under five years old, with 99% of these deaths occurring in low- and middle-income countries (LMICs), largely due to limited access to timely and appropriate healthcare\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. Severe ALRIs during infancy can also have long-term effects on lung health, increasing the risk of developing chronic conditions such as asthma, chronic obstructive pulmonary disease (COPD), and even early death later in life\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eSeveral risk factors contribute to the high burden of ALRIs in young children in LMICs. These include malnutrition, exposure to indoor air pollution, particularly from cooking with solid fuels, overcrowded living environments, poor sanitation, and the lack of exclusive breastfeeding during the first six months of life\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. In Bangladesh, ALRIs remain one of the biggest threats to young children\u0026rsquo;s health, responsible for approximately 18% of all deaths in children under five years old \u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. This burden is especially pronounced in rural areas, where healthcare access is limited and delays in care-seeking is common. Although overall care-seeking behavior among families is relatively high, only approximately 42\u0026ndash;46% of children with ALRI symptoms receive timely treatment from qualified healthcare providers, which significantly increases the risk of severe illness and death \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eOne of the major clinical features of ALRIs is increased work of breathing (WOB), which serves as a key indicator for the severity of the infection.\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e Recognizing and accurately assessing increased WOB is crucial for timely intervention and treatment. However, the subjective nature of clinical signs reflecting increased WOB and its severity (e.g., does the child have mild or significant indrawing of skin at the intercostal or subcostal areas), along with the varying levels of expertise among healthcare providers, often leads to inconsistencies in identification and delays in treatment.\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eEmerging technologies like artificial intelligence (AI) offer advances in improving the identification of increased WOB in young children. AI-driven tools using video recordings can objectively analyze chest movements to detect abnormal breathing patterns.\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e Similarly, digital stethoscopes equipped with an AI algorithm can identify adventitious lung sounds such as wheeze and crackles.\u003csup\u003e\u003cspan additionalcitationids=\"CR12\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e These tools can make it easier to identify breathing problems, reduce human error, and support faster and more reliable identification and treatment of ALRIs.\u003c/p\u003e\u003cp\u003eTo address existing identification challenges, a novel AI-based tool is being developed to help standardize the identification of increased WOB in young children. This tool will use video and digital stethoscope recordings to assess increased WOB and can potentially be expanded to older children. Our pilot study explored how feasible and acceptable the idea of this AI tool (which is still being developed) is to frontline health workers, carers and stakeholders when being used in community settings in Bangladesh.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eStudy Design\u003c/h2\u003e\u003cp\u003eThis qualitative study used a narrative research approach, as described by the Clandinin and Connelly (2000), where narrative inquiry is viewed as \u0026ldquo;a way of understanding experience\u0026rdquo; and involves studying the lived experiences of individuals as they are expressed in stories. The researcher and participant often engage collaboratively in co-constructing the narrative.\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eData were collected through Focus Group Discussions (FGDs) and In-Depth Interviews (IDIs). FGDs provided insights into shared perceptions and social dynamics influencing participants\u0026rsquo; acceptance of new diagnostic tools, while IDIs allowed for detailed exploration of individual experiences, professional opinions, and policy-level considerations.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eStudy Site and Duration\u003c/h3\u003e\n\u003cp\u003eThis study was conducted at the Projahnmo surveillance area in Zakiganj sub-district, Sylhet district, Bangladesh. The site was established in 2001 through a partnership between Johns Hopkins University, the Bangladesh Ministry of Health and Family Welfare (MOHFW), Bangladeshi academic institutions and non-government organizations (NGOs) including Projahnmo Research Foundation. The study was conducted from March to April 2025. The location map of the study site is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\n\u003ch3\u003eStudy Population\u003c/h3\u003e\n\u003cp\u003eFor the FGDs, the study population were the caregivers (male and female) of children under 2 years old, community health care providers (CHCPs), and community leaders. CHCPs are trained health workers, and staff of Community Clinics (CCs)\u0026mdash;small, government-established health centers at the village level. Each community clinic typically serves about 6,000 people. CHCPs are responsible for providing basic health services. Community leaders are individuals who hold influence, authority, or respect within their local communities and often play critical roles in social mobilization, dispute resolution, and local development.\u003c/p\u003e\u003cp\u003eUnder-2 children\u0026rsquo;s caregivers were divided into male and female FGD groups separately to encourage more open and honest discussion. In many cultural contexts, especially those with traditional gender roles, participants may feel uncomfortable or restrained discussing sensitive topics\u0026mdash;such as child health, caregiving practices, or household dynamics\u0026mdash;in mixed-gender settings. By conducting gender-specific groups, the study aimed to create a safe and comfortable environment where participants could express their views freely without fear of judgment, social pressure, or embarrassment.\u003c/p\u003e\u003cp\u003eFour IDIs were conducted with key stakeholders, including a national-level policymaker, a national-level policymaker and implementer, a division-level health administrator, and an upazila (subdistrict)-level health administrator.\u003c/p\u003e\u003cp\u003eThe study used purposive sampling to select participants with relevant knowledge and experience regarding child health and healthcare service delivery.\u003c/p\u003e\n\u003ch3\u003eData Collection Procedures\u003c/h3\u003e\n\u003cp\u003eData collection was conducted for the four FGDs and four IDIs. Each FGD lasted between 1.5 to 2 hours, while IDIs lasted approximately 30\u0026ndash;45 minutes.\u003c/p\u003e\u003cp\u003eA semi-structured interview guide was used across all FGDs and IDIs to ensure consistency in topics while allowing flexibility to explore emerging themes. The guides were tailored to each participant group (e.g., we would ask carers about their experiences in bringing their children with respiratory symptoms to the community clinics and ask CHCPs about their experiences in treating these children). Prior to the discussions, participants received a verbal explanation (detailed below) of the proposed idea of AI-based tool to ensure conceptual understanding. These explanations were not part of the data collection, and the AI tool hasn\u0026rsquo;t been developed yet.\u003c/p\u003e\u003cp\u003eThe proposed AI tool included:\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eDigital stethoscope\u003c/b\u003e: To record lung sounds at four chest positions (two front, two back), each lasting\u0026thinsp;~\u0026thinsp;10 seconds (total\u0026thinsp;~\u0026thinsp;60 seconds), capturing 3\u0026ndash;4 full breathing cycles (inspiration and expiration) per site. An example of a digital stethoscope is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003e\u003cb\u003eVideo image recording\u003c/b\u003e: A 60-second video of the child\u0026rsquo;s breathing from nose to lower chest wall, potentially captured via mobile phone. An example of video-based recording of chest movements is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\n\u003ch3\u003eResearch Team and Training\u003c/h3\u003e\n\u003cp\u003eA qualitative researcher led data collection with a background in anthropology and experience in conducting qualitative methods, including FGDs, IDIs, data analysis, and reporting. A Research Assistant (RA) with a social science background and some research experience supported data collection. The RA received specific training and was supervised by the qualitative researcher. TS and AMK conducted an orientation session on the study protocol for the research team.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eEthical Considerations\u003c/h2\u003e\u003cp\u003e All participants provided written informed consent before participation. All discussions and interviews were conducted in Bangla, and audio-recorded with participant permission. The study received ethical approval from Edinburgh Medical School Research Ethics Committee (EMREC) with reference number 24-EMREC-078 and Projahnmo Research Foundation Institutional Review Board (PRF IRB) with reference number PR-25001.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eData Management and Analysis\u003c/h3\u003e\n\u003cp\u003eAudio recordings were transcribed verbatim and translated into English. Data were analyzed using thematic analysis, identifying key themes and sub-themes. The process began with a thorough, line-by-line reading of the data, during which initial codes were assigned to segments that appeared meaningful or relevant. These codes were then reviewed to identify connections and group similar ideas together. Through this process, broader patterns emerged, leading to the development of potential themes that captured significant aspects of the data in relation to the research question. This approach facilitated the transition from raw data to more organized and interpretable findings.\u003c/p\u003e\u003cp\u003e A matrix table was developed to organize, compare, and analyze data across participant groups.\u003c/p\u003e\u003cp\u003e Grounded Theory was used for this study as it enables the inductive development of a conceptual model grounded in the perspectives and experiences of healthcare providers, caregivers, and stakeholders regarding the implementation of AI-driven video and digital auscultation tools in diagnosing respiratory distress in young children. Given the limited existing research on the adoption of such technologies in low-resource settings like Zakiganj, Sylhet, this approach allows for the identification of emergent themes and processes related to feasibility and acceptability. By systematically analyzing qualitative data, Grounded Theory facilitates a nuanced understanding of contextual factors and user interactions, thereby informing strategies for effective integration and broader application of these innovative diagnostic tools.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eThe baseline characteristics of the included participants were available in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. FGDs included participants of caregivers (20 participants, 9 females and 11 males), community healthcare providers (10 participants) and community leaders (10 participants). IDIs included participants of key policymakers and senior health officials (4 participants in total).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSummary of study participants\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParticipant type\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMethod\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNumber of FGDs/IDIs\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eTotal participants\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eCharacteristics\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCaregiver (female)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFGD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e9\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eAge:\u003c/p\u003e\u003cp\u003e-Range 20\u0026ndash;27 years\u003c/p\u003e \u003cp\u003e- Mean 23 years\u003c/p\u003e\u003cp\u003eEducation:\u003c/p\u003e\u003cp\u003e-Secondary (n\u0026thinsp;=\u0026thinsp;7)\u003c/p\u003e\u003cp\u003e-Higher secondary (n\u0026thinsp;=\u0026thinsp;1)\u003c/p\u003e\u003cp\u003e-Graduate (n\u0026thinsp;=\u0026thinsp;1)\u003c/p\u003e\u003cp\u003eOccupation:\u003c/p\u003e\u003cp\u003e-Housewife (n\u0026thinsp;=\u0026thinsp;9)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCaregiver (male)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFGD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eAge:\u003c/p\u003e\u003cp\u003e-Range 25\u0026ndash;32 years\u003c/p\u003e\u003cp\u003e-Mean 29 years\u003c/p\u003e\u003cp\u003eEducation:\u003c/p\u003e\u003cp\u003e-Illiterate (n\u0026thinsp;=\u0026thinsp;1)\u003c/p\u003e\u003cp\u003e-Primary (n\u0026thinsp;=\u0026thinsp;4)\u003c/p\u003e\u003cp\u003e-Secondary (n\u0026thinsp;=\u0026thinsp;3)\u003c/p\u003e\u003cp\u003e-Higher secondary (n\u0026thinsp;=\u0026thinsp;1)\u003c/p\u003e\u003cp\u003e-Post-graduation (n\u0026thinsp;=\u0026thinsp;1)\u003c/p\u003e\u003cp\u003eOccupation:\u003c/p\u003e\u003cp\u003e- Mostly farmers with small business\u003c/p\u003e\u003cp\u003e- Others include rickshaw puller, small tea stall owners, small vegetable sellers\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCHCPs\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFGD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eEducation:\u003c/p\u003e\u003cp\u003eCompleted Higher Secondary Certificate or an equivalent qualification\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCommunity leaders\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFGD\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eEducation:\u003c/p\u003e\u003cp\u003eHigher Secondary (n\u0026thinsp;=\u0026thinsp;4)\u003c/p\u003e\u003cp\u003eGraduation (n\u0026thinsp;=\u0026thinsp;4)\u003c/p\u003e\u003cp\u003ePost-graduation (n\u0026thinsp;=\u0026thinsp;2)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePolicymakers and senior health officials\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eIDI\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eEducation:\u003c/p\u003e\u003cp\u003eMBBS (n\u0026thinsp;=\u0026thinsp;3), Post-graduation in Pediatrics (n\u0026thinsp;=\u0026thinsp;1)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eFGD \u0026ndash; focus group discussion; IDI \u0026ndash; In-depth interview\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eTheme 1: Supplement for Systemic Gaps\u003c/h2\u003e\u003cp\u003eParticipants of IDIs consistently emphasized that shortages in skilled personnel and diagnostic infrastructure pose major challenges in rural healthcare delivery. In such contexts, AI tools were considered feasible solutions that could support clinical decision-making where radiographic imaging, laboratory testing, or pediatric specialists are unavailable. These tools were seen not as replacements for physicians, but as supplementary aids to bridge critical service gaps.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\"In remote clinics, doctors aren\u0026rsquo;t always present, and we lack tools like X-rays. If AI can help detect danger signs early, it could make a real difference.\"\u003c/em\u003e \u0026mdash; Health administrators at Upazila-level\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003eTheme 2: Task-Sharing and Human Resource Utilization\u003c/h2\u003e\u003cp\u003eThe potential for task-sharing was highlighted as a practical advantage. Division-level Health Administrator noted that with minimal but targeted training, nurses, CHCPs, and paramedics could effectively operate the tools and interpret outputs. This redistribution of responsibilities was seen as a feasible and scalable strategy to alleviate pressure on physicians, especially during peak patient loads.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\u0026ldquo;If experienced nurses or CHCPs can be trained, they can handle the tool. It\u0026rsquo;s better than waiting for a doctor who may not be there.\u0026rdquo;\u003c/em\u003e \u0026mdash; Division-level Health Administrator\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eTheme 3: Operational Feasibility in Existing Infrastructure\u003c/h2\u003e\u003cp\u003eCHCPs described how minor logistical adjustments\u0026mdash;such as creating quiet corners or designated rooms in busy clinics\u0026mdash;could allow effective use of the AI tools. The simplicity of the proposed technology (e.g., mobile-based video, portable stethoscopes) was considered an advantage, aligning with existing infrastructure constraints.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\u0026ldquo;We don\u0026rsquo;t need a separate room. A quiet space in the clinic would be enough for recording.\u0026rdquo;\u003c/em\u003e \u0026mdash; CHCP\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003eTheme 4: Previous Experience Increases Confidence\u003c/h2\u003e\u003cp\u003eThe Division-level Health Administrator referred to earlier successful initiatives using digital stethoscopes in community settings to detect pneumonia. This prior familiarity with similar technologies gave them confidence in the feasibility of implementing an AI-enhanced system for identifying increased WOB.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\u0026ldquo;We\u0026rsquo;ve seen this work before in pneumonia studies. This is not entirely new for us.\u0026rdquo;\u003c/em\u003e \u0026mdash; Health Administrator at Upazila (subdistrict) level\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003eTheme 5: Scalability with Structured Training and Supervision\u003c/h2\u003e\u003cp\u003eCHCPs emphasized that while the tools are technologically feasible, effective rollout would require structured training modules, ongoing technical support, and clear protocols for use. With these systems in place, participants believed the tools could be sustainably integrated into routine service delivery.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\u0026ldquo;Training is key. If CHCPs are supported and supervised, they can do it well.\u0026rdquo;\u003c/em\u003e \u0026mdash; Health Administrator at Upazila (subdistrict) level\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003eTheme 6: Alignment with Policy Goals\u003c/h2\u003e\u003cp\u003ePolicymakers noted that AI-supported tools aligned with national goals to strengthen digital health, improve early detection, and reduce child mortality. The feasibility of implementation was seen as high, particularly if supported by donor funding, public-private partnerships, or pilot programs to test scalability.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\u0026ldquo;This fits with our digital health strategy. If it proves effective, we can advocate for wider adoption.\u0026rdquo;\u003c/em\u003e \u0026mdash; National-level Health Policy makers\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003ePerceived Feasibility and Potential Benefits\u003c/h2\u003e\u003cdiv id=\"Sec18\" class=\"Section3\"\u003e\u003ch2\u003eImproved Identification Accuracy and Early Detection\u003c/h2\u003e\u003cp\u003eCaregivers viewed the AI-supported tool as a promising addition to existing methods of identifying increased WOB, particularly for respiratory illnesses like pneumonia. They hoped that the device\u0026rsquo;s combination of audio (lung sounds) and video (visual assessment of breathing patterns) would enhance identification accuracy and support timely detection of illness, especially in settings where access to experienced doctors is limited.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\"With this device, maybe they can catch the illness early\u0026mdash;even before it becomes severe.\"\u003c/em\u003e \u0026mdash; Female caregiver\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\u003ch2\u003eVisual and Audio Feedback Enhancing Understanding and Trust\u003c/h2\u003e\u003cp\u003eParticipants appreciated the tool\u0026rsquo;s potential to provide visual and audio evidence of a child\u0026rsquo;s condition. This was seen as a key strength that could help bridge communication gaps between healthcare providers and caregivers, who often struggle to understand the severity of respiratory conditions. Seeing or hearing a problem directly was expected to build trust and encourage adherence to treatment or referral advice.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\"Sometimes we don\u0026rsquo;t understand what the doctor says. But if we see the problem ourselves, we will take it seriously.\"\u003c/em\u003e \u0026mdash; Male caregiver\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\u003ch2\u003eSupport for Local Clinics and Health Workers\u003c/h2\u003e\u003cp\u003eCaregivers supported the use of AI tools at community clinics and by local health workers, particularly CHCPs. They saw the tool as a way to extend identification capabilities to the local level, potentially reducing unnecessary travel and promoting early care-seeking. However, they emphasized the need for trained personnel to operate the device to ensure correct use.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\"If it\u0026rsquo;s available in the village clinic and they know how to use it, it will save us time and money.\"\u003c/em\u003e \u0026mdash; Female caregiver\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e\u003ch2\u003ePerceptions of the Proposed Identification Device\u003c/h2\u003e\u003cp\u003eThe female caretakers expressed a clear preference for the proposed device over traditional methods, indicating that the new technology would provide more accurate and reliable assessments of a child\u0026rsquo;s respiratory condition. They described limitations with the current approach, where village doctors often rely on stethoscopes and subjective judgment, which may lead to inappropriate treatment such as unnecessary antibiotic prescriptions or failure to recognize severity. The device\u0026rsquo;s capability to objectively analyze and provide instant feedback on increased WOB, assisting in clinical decision-making on whether a child requires hospital referral or home treatment was considered highly beneficial.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\u003ch2\u003eConcerns and Constraints Affecting Feasibility\u003c/h2\u003e\u003cdiv id=\"Sec23\" class=\"Section3\"\u003e\u003ch2\u003eChild Discomfort and Cooperation Challenges\u003c/h2\u003e\u003cp\u003eA common concern raised by caregivers was the practical difficulty of keeping young children calm during the test. They noted that unfamiliar devices, especially if cold or metallic, might distress the child. Caregivers suggested breastfeeding or soothing the child beforehand as potential strategies.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\"Babies don\u0026rsquo;t sit still. If the device is cold or scary, they\u0026rsquo;ll cry and move too much.\"\u003c/em\u003e \u0026mdash; Female caregiver\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e\u003ch2\u003eNeed for Skilled Operation and Interpretation\u003c/h2\u003e\u003cp\u003eCaregivers strongly felt that trained healthcare workers\u0026mdash;not family members\u0026mdash;should operate the device. They believed that improper use by untrained individuals could lead to misinterpretation or misuse, reducing the tool\u0026rsquo;s effectiveness and possibly causing harm or unnecessary worry.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\"Only trained people should use it. We don\u0026rsquo;t understand how it works.\"\u003c/em\u003e \u0026mdash; Male caregiver\u003c/p\u003e\u003cdiv id=\"Sec25\" class=\"Section3\"\u003e\u003ch2\u003eTime and Patience Required During Use\u003c/h2\u003e\u003cp\u003eSome participants noted that the test process might take longer than usual checkups, requiring patience from both caregivers and health workers, especially if multiple children are being assessed. This could be challenging in crowded clinics or for caregivers in a hurry.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\"If the process takes time, it should be explained. Otherwise, people may get frustrated or leave.\"\u003c/em\u003e \u0026mdash; Female caregiver\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec26\" class=\"Section3\"\u003e\u003ch2\u003eConcerns About Availability and Equity\u003c/h2\u003e\u003cp\u003eWhile caregivers supported the tool\u0026rsquo;s use, they expressed concerns that it may not be equitably accessible, especially in remote or under-resourced clinics. Participants emphasized the importance of widespread distribution and sufficient training across facilities, so all children\u0026mdash;regardless of location\u0026mdash;could benefit.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\"We hope it\u0026rsquo;s not only in big clinics. Even the small ones in our villages should have it.\"\u003c/em\u003e \u0026mdash; Male caregiver\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec27\" class=\"Section3\"\u003e\u003ch2\u003eTrust and Cultural Acceptance\u003c/h2\u003e\u003cp\u003eAlthough many caregivers were optimistic, a few raised concerns about trust in technology replacing traditional physical examinations. They valued the human touch and reassurance that comes from a provider physically examining the child and cautioned that community education and demonstration would be needed to build comfort and trust in the tool.\u003c/p\u003e\u003cp\u003e\u003cem\u003e\"We believe what the doctor says when he checks with his hand. A machine might not feel the same.\"\u003c/em\u003e \u0026mdash; Female caregiver\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis qualitative study explored the perceived feasibility and acceptability of AI-supported tools\u0026mdash;specifically a digital stethoscope and video-based respiratory assessment\u0026mdash;for the early detection of increased WOB in children under two years old in rural Bangladesh. Drawing on perspectives from caregivers, healthcare providers, and policymakers, the findings indicate broad optimism regarding the potential of such tools to enhance identification accuracy, improve early detection, and strengthen decentralized care. However, several important contextual and operational factors must be addressed to ensure successful implementation.\u003c/p\u003e\u003cp\u003eAcross all respondent groups, participants recognized the potential of AI-assisted tools to support early identification of increased WOB, particularly in settings where healthcare infrastructure and skilled personnel are limited \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. Caregivers believed that visual and audio outputs could improve their understanding of a child's condition and encourage adherence to treatment or referral advice. These perceptions reflect growing trust in digital health solutions and echo findings from earlier studies in Bangladesh demonstrating caregiver receptiveness to digital tools when linked to better outcomes and clearer communication \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eHealthcare providers emphasized the value of AI tools in bridging critical human resource gaps and reducing dependence on clinical expertise alone. By supporting task-shifting to community health workers and paramedics, these tools were seen as a means to decentralize care and reduce unnecessary referrals, thereby improving system efficiency. Policymakers shared this view, highlighting alignment with national digital health priorities and suggesting that such innovations could support ongoing efforts to reduce child mortality and improve health equity.\u003c/p\u003e\u003cp\u003eImportantly, the visual nature of the tool\u0026mdash;particularly the video capture of breathing\u0026mdash;was described by caregivers and providers as a unique strength. It was viewed not only as an identification aid but also as a communication tool that empowers caregivers to better engage in their child\u0026rsquo;s care. This finding suggests that digital tools that offer tangible, understandable feedback may enhance patient-provider interactions and shared decision-making in low-literacy settings.\u003c/p\u003e\u003cp\u003eDespite strong enthusiasm, the study also revealed concerns that must be addressed to ensure feasibility. These include logistical challenges such as child discomfort during testing, environmental noise in busy clinics, and the need for dedicated quiet spaces. Moreover, while the tools themselves are non-invasive and relatively easy to operate, participants emphasized the importance of proper and ongoing training, technical supervision, and support systems for health workers. A one-time training intervention was viewed as insufficient for maintaining quality over time, especially in rural clinics already facing staff shortages and high turnover.\u003c/p\u003e\u003cp\u003eA recurring theme was the tension between technological innovation and the perceived need for human touch. Some caregivers expressed concern that reliance on devices may undermine trust traditionally built through direct physical examination by a physician. This underscores the need for careful integration of AI tools within existing care models, ensuring that technology augments rather than replaces the human elements of care.\u003c/p\u003e\u003cp\u003eFurthermore, participants raised critical issues around scalability and sustainability. Policymakers noted that widespread adoption would require robust monitoring systems, validation studies in real-world conditions, and strong evidence of cost-effectiveness. Infrastructure constraints such as limited electricity and unreliable internet connectivity in some clinics may also limit short-term scalability.\u003c/p\u003e\u003cp\u003eMoreover, while the study explored the potential of an AI-supported tool to strengthen community health service delivery, its actual implementation revealed a range of significant challenges. Although the tool showed promise in theory\u0026mdash;offering efficiencies in data analysis, decision support, and resource allocation\u0026mdash;financial constraints quickly surfaced as a major barrier. Many upazila- and union-level facilities lack digital literacy, and consistent internet access needed to support such technology. Moreover, the introduction of AI-based decision-making tools altered existing workflows and responsibilities, potentially leading to unintended tensions among health workers. For example, task shifting\u0026mdash;where AI tools assisted CHCPs in triaging or clinical decision-making\u0026mdash;was met with resistance or confusion from other cadres like Health Assistants or Family Welfare Assistants, who perceived the tool as either undermining their roles or increasing their workload. This underscores a broader issue: while AI solutions may appear effective in the abstract, successful implementation requires addressing ground-level limitations, ensuring buy-in from frontline workers, and carefully managing changes to established practices and hierarchies.\u003c/p\u003e\u003cp\u003eHowever, most of the participating caregivers, CHCPs and community leaders have lower educational qualifications, and all participating caregivers are relatively young which might have biased the result.\u003c/p\u003e\u003cdiv id=\"Sec29\" class=\"Section2\"\u003e\u003ch2\u003eImplications for Policy and Practice\u003c/h2\u003e\u003cp\u003eThe insights generated in this study have several important implications for policy and practice. First, the introduction of AI tools should be accompanied by structured training programs and long-term technical support. Second, implementation efforts should prioritize community engagement, including sensitization campaigns and trust-building through transparent communication. Third, health system readiness\u0026mdash;including infrastructure, staffing, and support mechanisms\u0026mdash;must be assessed prior to deployment.\u003c/p\u003e\u003cp\u003eTo foster sustainability, integration with existing government health systems, rather than parallel implementation by external actors, is critical. Additionally, policies ensuring equitable access, data privacy, and ethical use must be developed in consultation with frontline users and affected communities.\u003c/p\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eCompeting interests\u003c/h2\u003e\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e\u003cp\u003eTS acknowledged the Chancellor\u0026rsquo;s Fellowship Starting Fund from University of Edinburgh.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eTS, SA, AMK conceptualised and designed this study. Tamanna S collected and analysed data. Tamanna S wrote the first draft of the manuscript. SA, TS, AMK provided critical feedback on the manuscript structure and content. All authors contributed to the interpretation of the results, reviewed the manuscript, and approved the submitted version.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThe authors extend their deepest gratitude to the Ministry of Health and Family Welfare, Government of Bangladesh, National Newborn Health Program, and CHCPs of the respective community clinics in Zakiganj, Sylhet, for their invaluable assistance in supporting activities for this research.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eWalker CLF, Rudan I, Liu L et al (2013) Global burden of childhood pneumonia and diarrhoea. Lancet 381(9875):1405\u0026ndash;1416\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNair H, Sim\u0026otilde;es EA, Rudan I et al (2013) Global and regional burden of hospital admissions for severe acute lower respiratory infections in young children in 2010: a systematic analysis. Lancet 381(9875):1380\u0026ndash;1390\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eUnited Nations Inter-agency Group for Child Mortality Estimation (UN IGME) (2023) Levels \u0026amp; Trends in Child Mortality 2023. UNICEF, New York:\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShi T, McAllister DA, O'Brien KL et al (2017) Global, regional, and national disease burden estimates of acute lower respiratory infections due to respiratory syncytial virus in young children in 2015: a systematic review and modelling study. Lancet 390(10098):946\u0026ndash;958\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSebastin, Kenmoe Etienne Atenguena Okobalemba Association between early viral lower respiratory tract infections and subsequent asthma development. World J Crit Care Med 2022 Jul 9; 11(4): 298\u0026ndash;310\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWorld Health Organization. Pneumonia (2023) \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.who.int/news-room/fact-sheets/detail/pneumonia\u003c/span\u003e\u003cspan address=\"https://www.who.int/news-room/fact-sheets/detail/pneumonia\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (accessed July 31 2025)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eUNICEF, Pneumonia (2023) The forgotten child killer. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.unicef.org/stories/pneumonia-forgotten-child-killer\u003c/span\u003e\u003cspan address=\"https://www.unicef.org/stories/pneumonia-forgotten-child-killer\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (accessed 12 August 2025)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRambaud-Althaus C, Althaus F, Genton B, D'Acremont V (2015) Clinical features for diagnosis of pneumonia in children younger than 5 years: a systematic review and meta-analysis. Lancet Infect Dis 15(4):439\u0026ndash;450\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWallihan R, Ramilo O (2014) Community-acquired pneumonia in children: current challenges and future directions. J Infect 69(Suppl 1):S87\u0026ndash;90\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKhan AM, Ahmed S, Chowdhury NH et al (2022) Developing a video expert panel as a reference standard to evaluate respiratory rate counting in paediatric pneumonia diagnosis: protocol for a cross-sectional study. BMJ open 12(11):e067389\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAhmed S, Mitra DK, Nair H et al (2022) Digital auscultation as a novel childhood pneumonia diagnostic tool for community clinics in Sylhet, Bangladesh: protocol for a cross-sectional study. BMJ open 12(2):e059630\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAhmed S, Sultana S, Khan AM et al (2022) Digital auscultation as a diagnostic aid to detect childhood pneumonia: A systematic review. J global health 12:04033\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJoarder T, Tune SNBK, Islam AA et al (2023) End-user acceptability of a prototype digital stethoscope to diagnose childhood pneumonia-a qualitative exploration from Sylhet, Bangladesh. BMC Digit Health 1(1):26\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eClandinin DJ, Connelly FM (2000) Narrative Inquiry: Experience and Story in Qualitative Research. Jossey-Bass., San Francisco, CA\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKarthika M, Sreedharan JK, Shevade M, Mathew CS, Ray S (2024) Artificial intelligence in respiratory care. Front Digit Health 6:1502434\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-digital-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [BMC Digital Health](https://bmcdigitalhealth.biomedcentral.com/)","snPcode":"44247","submissionUrl":"https://submission.nature.com/new-submission/44247/3","title":"BMC Digital Health","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7648403/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7648403/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAcute lower respiratory infections (ALRIs) are a leading cause of child mortality globally, with timely recognition of increased work of breathing (WOB) critical for early intervention. In low-resource settings, WOB assessment is often subjective and inconsistent, contributing to delays in care. Digital health tools using artificial intelligence (AI) offer promising solutions to standardize detection of increased WOB through video and audio analysis. We are developing an AI-based tool to support assessment of increased WOB in children. This pilot study explores its feasibility and acceptability before its prototype is developed.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThis qualitative study was conducted in March\u0026ndash;April 2025 in Zakiganj, Sylhet, Bangladesh. Data were collected through four focus group discussions and four in-depth interviews with caregivers, community health care providers (CHCPs), community leaders, policymakers and health administrators. They followed a semi-structured guide tailored to participant roles. Participants were verbally introduced to the concept of an AI-based tool for assessing WOB in children to guide discussion; no demonstrations or recordings were used.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe findings demonstrate broad support for the feasibility of implementing AI-supported tools - digital stethoscopes and video-based respiratory assessments\u0026mdash;for identifying increased WOB in children, particularly in rural and resource-constrained settings. Stakeholders emphasized that these tools offer practical solutions to address critical gaps in skilled personnel and diagnostic infrastructure, with strong endorsement for their use by CHCPs to facilitate task-shifting. Prior experience with similar technologies, such as digital stethoscopes for pneumonia, further reinforced their confidence in feasibility. Policymakers noted its alignment with national digital health strategies and child mortality reduction goals, indicating potential for scale-up through pilot initiatives and public-private partnerships. Caregivers also expressed positive perceptions, highlighting improved diagnostic accuracy, enhanced understanding through visual and audio feedback, and greater accessibility at local clinics. Nevertheless, concerns were raised about child cooperation, need for trained operators, time constraints, equitable access, and trust in technology.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusion\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAI-driven video and digital auscultation tools are considered feasible and acceptable for assessing increased WOB in children with ALRIs in rural Bangladesh. Their adoption will require training, technical support, community trust-building, and equitable access, with potential for scale-up to strengthen child health services.\u003c/p\u003e","manuscriptTitle":"Feasibility and Acceptability of Artificial Intelligence-Driven Video and Digital Auscultation Tools for Identifying Increased Work of Breathing in Young Children with Acute Lower Respiratory Infections in Rural Bangladesh","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-24 09:24:25","doi":"10.21203/rs.3.rs-7648403/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-09-26T15:00:28+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-19T12:02:15+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-19T12:00:11+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Digital Health","date":"2025-09-18T10:10:56+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-digital-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [BMC Digital Health](https://bmcdigitalhealth.biomedcentral.com/)","snPcode":"44247","submissionUrl":"https://submission.nature.com/new-submission/44247/3","title":"BMC Digital Health","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"d9cadce7-a2f7-4403-81aa-eb6e77932a60","owner":[],"postedDate":"September 24th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-11-27T13:23:10+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-24 09:24:25","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7648403","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7648403","identity":"rs-7648403","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}