Abstract
Introduction
The global prevalence of heart failure continues to increase, particularly in ageing populations. Many older patients receiving home-based medical care have limited access to standard diagnostic tools, such as chest radiography and echocardiography, which can delay the detection of disease progression. AMI-SSS01, an artificial intelligence (AI)-based portable phonocardiography device, enables rapid and reproducible analysis of heart sounds, potentially allowing earlier identification of heart failure exacerbations in home-based medical settings. This trial aims to evaluate the feasibility and safety of implementing the AMI-SSS01 phonocardiography system in patients receiving home-based medical care.
Methods and analysis
This two-arm, parallel-group, randomised controlled feasibility trial is being conducted at two primary care clinics affiliated with the Research Organisation for Education and Network in Primary Care-Based Research Network in Nagasaki, Japan. Participants aged ≥65 years with symptomatic heart failure receiving home-based medical care are eligible for study inclusion. Participants are randomised in a 1:1 ratio to either the intervention group, in which AI-assisted phonocardiography using the AMI-SSS01 device is performed during physician home visits, or the control group receiving conventional home-based medical care without phonocardiography. Feasibility outcomes include recruitment rate, overall data completeness and completion and protocol-concordant completion rates for phonocardiography examinations. Safety outcomes include adverse events related to device use. The registered primary clinical outcome, collected as an exploratory clinical outcome, is the time from randomisation to hospitalisation for heart failure or cardiovascular death during the 24-week follow-up period. The registered secondary clinical outcome is the number of heart failure exacerbation events.
Ethics and dissemination
Ethical approval was obtained from the Nagasaki University Hospital Clinical Research Ethics Committee (CRB24-028). The findings will be disseminated through peer-reviewed publications and scientific conferences.
Trial registration number
Japan Registry of Clinical Trials (jRCT1072240118).
Keywords: Aging, Hospital to Home Transition, Heart failure, Artificial Intelligence, Feasibility Studies, Randomized Controlled Trial
Strengths and limitations of this study.
This randomised controlled feasibility trial is conducted in a real-world home-based medical care setting involving older adults with symptomatic heart failure.
The study evaluates an artificial intelligence-assisted portable phonocardiography system designed for point-of-care use during physician home visits.
Feasibility outcomes include recruitment rate, overall data completeness and completion and protocol-concordant completion rates for phonocardiography examinations.
The study is not powered to evaluate clinical efficacy because of the feasibility-oriented design and small sample size.
The open-label design and conduct at two clinics within a single geographic region may limit generalisability and introduce ascertainment bias.
Introduction
Heart failure has become a major public health concern, affecting an estimated 56 million individuals worldwide as the global population continues to age.1 It is a major cause of hospitalisation and is associated with poor prognosis, high mortality and frequent rehospitalisation, resulting in a substantial healthcare burden.2 The burden of heart failure is increasing not only in Western countries but also in Asia.3 In Japan, projections based on regional hospital survey data indicate that the number of patients with heart failure will exceed 1.3 million by 2055.4
In Japan, home-based medical care is specifically designed for patients who cannot visit hospitals or clinics. Conventional home-based care for heart failure typically involves assessing congestion based on findings such as dyspnoea, oedema and changes in body weight and considering the initiation or dose escalation of diuretics. When patients develop decompensated clinical conditions and/or abnormal vital signs requiring urgent intervention, home-based care physicians refer them for hospitalisation. However, patients receiving such care have limited access to diagnostic modalities such as chest radiography, electrocardiography and echocardiography.
Heart failure is defined as a clinical syndrome characterised by symptoms and signs caused by structural or functional cardiac abnormalities, supported by elevated N-terminal pro-B-type natriuretic peptide (NT-proBNP) levels or objective evidence of cardiogenic pulmonary or systemic congestion obtained through diagnostic testing.5 However, the interpretation of NT-proBNP levels can be challenging in older patients with renal dysfunction,6 and conventional cardiac auscultation is highly operator-dependent, demonstrating only fair and variable interobserver agreement for murmur classification and wide variability in diagnostic accuracy across studies.7 8 Therefore, complementary diagnostic indices that can be reliably applied in home-based medical care settings are urgently needed.9
The AMI-SSS01 phonocardiography device was developed to address this unmet clinical need.10 This compact, repeatable, palm-sized device records heart sounds in 8 s when placed on the anterior chest wall. AMI-SSS01 was approved by the Japanese Pharmaceutical and Medical Devices Agency in 2023. This device is covered by the national health insurance system in Japan. The recorded data are automatically uploaded to a cloud-based system and analysed using artificial intelligence (AI) algorithms. The device allows the quantitative and reproducible assessment of abnormal heart sounds, such as the third heart sound and gallop rhythm. A validation study11 has demonstrated that this AI-based analysis can identify patients with elevated estimated B-type natriuretic peptide (BNP) levels with high diagnostic accuracy (BNP≥100 pg/mL), achieving an area under the receiver operating characteristic curve of 0.895, with sensitivity and specificity of 84.3% and 82.9%, respectively, among patients attending cardiology outpatient departments at general hospitals. Given its portability and repeatability, the AMI-SSS01 phonocardiography system appears well-suited for the early diagnosis and monitoring of heart failure exacerbations in home-based medical care.
This trial is primarily designed to assess the feasibility and safety of implementing the AI-assisted phonocardiography system in home-based medical care. Feasibility outcomes include recruitment rate, overall data completeness and completion and protocol-concordant completion rates for phonocardiography examinations. Registered primary and secondary outcomes will be collected as exploratory clinical outcomes to estimate event frequency, assess outcome ascertainment procedures and inform the design of future definitive trials.
Methods and analysis
Study design
This trial is a two-arm, parallel-group, randomised controlled feasibility trial. The participants will be followed up for 24 weeks after randomisation. The study period started on 24 February 2025 and will continue until 31 December 2027. Participant enrolment began on 2 June 2025 and is expected to continue until 31 July 2026.
Study sites and participants
Study setting
This study is being conducted at two primary care clinics, Nakayama Naika Clinic and Tanigawa Clinic, which participate in the Research Organisation for Education and Network in Primary Care-Based Research Network (ROEN-PBRN) in Japan. ROEN-PBRN is a Japanese practice-based research network established to promote multicentre primary care research and nationwide research collaboration.12
Participants
The inclusion criteria are as follows: age of ≥65 years, currently receiving home-based medical care, diagnosed with symptomatic heart failure and written informed consent from (online supplemental file 1) the patients or their caregivers. Heart failure at enrolment will be operationally defined based on clinically diagnosed symptomatic heart failure, documented treatment history for heart failure and/or a history of symptomatic heart failure events assessed during routine home-based medical care. Available natriuretic peptide and echocardiographic data obtained during routine clinical care will be collected when accessible.
The exclusion criteria are as follows: undergoing maintenance haemodialysis; expected survival of <1 month due to non-heart failure-related causes, as determined by the treating physician; and participation in interventional clinical trials.
Participants will be randomly assigned to either the intervention group, which will undergo phonocardiographic examinations using the AMI-SSS01 at each home medical visit, or the non-intervention group, which will not undergo phonocardiographic examinations.
Randomisation and blinding
The participants will be allocated in a 1:1 ratio using stratified block randomisation at the study site (figure 1). The principal investigator will generate the allocation sequence in advance using random numbers and seal it in opaque envelopes, which will then be distributed to the participating clinics. Blinding will not be feasible due to the nature of the intervention.
Figure 1. Overview of the study design and participant flow. Participants are randomised to either artificial intelligence (AI)-assisted phonocardiography during physician home visits or conventional home-based medical care. Feasibility and safety outcomes are evaluated together with exploratory clinical outcomes during the 24-week follow-up period.
Interventions
Intervention group
Participants in the intervention group will undergo phonocardiographic examinations using the AMI-SSS01 device during all physician home visits, including scheduled and unscheduled visits (figure 2).
Figure 2. Intervention overview.
The AMI-SSS01 phonocardiography device10 records heart sounds in 8 s when placed at the fourth left sternal border on the anterior chest wall (figure 3). The recorded heart sounds and electrocardiographic data are automatically uploaded to a cloud-based platform developed by the device provider and analysed using AI algorithms to provide rapid and objective diagnostic support. No additional clinical information, laboratory data or imaging findings are uploaded to the AI analysis platform. Traditionally, abnormal findings such as a third heart sound and gallop rhythm have been difficult to quantify; however, this device enables reproducible and quantitative assessment using a categorical indicator of heart failure severity.
Figure 3. AMI-SSS01 phonocardiogram: acquisition, visualisation and artificial intelligence (AI)-assisted assessment. (A) The AMI-SSS01 phonocardiogram, a compact medical device approved by the Pharmaceuticals and Medical Devices Agency in Japan (approval no. 30 400BZX00218000), is capable of synchronously acquiring heart sounds and electrocardiographic data. The device records dynamic physiological signals during a fixed 8-second examination. (B) Real-time display of recorded physiological signals during data acquisition. (C) Visualisation of phonocardiographic findings. (D) Example of AI-derived B-type natriuretic peptide (BNP) category output.
The AI estimates BNP levels and categorises the results into one of five levels: A1 (<35 pg/mL), A2 (35–100 pg/mL), B (100–200 pg/mL), C (200–400 pg/mL) or D (≥ 400 pg/mL). The BNP category results are generated within approximately 1 min of the examination, making them immediately available at the point of care during home visits.
During physician home visits, AI-assisted phonocardiography will be routinely performed, and AI-derived BNP category results will be reviewed together with routine clinical assessments, including symptoms and physical findings, for clinical decision-making. When findings suggestive of heart failure exacerbation are identified, changes in the BNP category compared with previous examinations will be additionally assessed. Significant BNP elevation is defined as an increase in the classification category compared with the patient’s previous result. In the absence of significant BNP elevation, the patient will be managed with careful observation. However, if a significant elevation is confirmed, appropriate medical interventions will be considered, including but not limited to the administration of diuretics or hospital transfer. Following these decisions, the patient will be reassessed at the next regular visit (figure 4).
Figure 4. Clinical management workflow using artificial intelligence (AI)-assisted phonocardiography. AI-derived B-type natriuretic peptide (BNP) category results are available within approximately 1 min during home visits and may support immediate clinical decision-making. Subsequent review by the device provider is conducted separately for quality assurance purposes.
For quality assurance, a final report incorporating a definitive human review will be issued a few days later by the testing laboratory (AMI Inc). This subsequent human review does not determine the immediate clinical management of the participants.
In this feasibility trial, there are no strictly predefined thresholds for specific clinical actions. Instead, attending physicians will integrate phonocardiography findings with the patient’s overall clinical presentation when determining management strategies. The clinical actions informed by the AI-assisted assessment will be documented in the study records.
Control group
Participants in the control group will receive conventional home-based medical care without the use of the AMI-SSS01 phonocardiographic examination. Clinical management, including assessment and treatment of heart failure exacerbations, will be performed according to routine clinical practice at the discretion of the attending physician.
Outcomes
Feasibility and safety outcomes
The feasibility outcomes include recruitment rate, overall data completeness and completion and protocol-concordant completion rates for phonocardiographic examinations. The completion rate of phonocardiographic examinations is evaluated as the proportion of scheduled examinations that are successfully completed. In addition, the protocol-concordant completion rate is assessed as the proportion of completed examinations that are performed according to the protocol-specified procedures. Reasons for non-completion will be recorded and categorised (eg, technical issues, participant-related factors or safety-related discontinuation). Safety outcomes include potential adverse events, such as skin irritation or burns, associated with the application of the AMI-SSS01 device to the anterior chest wall.
Definition of heart failure exacerbation
According to the universal definition of heart failure, heart failure exacerbation events are defined based on a combination of clinical criteria and abnormal diagnostic findings.13 The clinical criteria for a heart failure exacerbation event require the presence of two or more of the following symptoms or abnormal physical findings.14 Symptoms include dyspnoea (shortness of breath), orthopnoea and malaise, and physical findings include the presence of a third heart sound, gallop rhythm and peripheral oedema. Abnormal laboratory findings are defined as elevated serum NT-proBNP levels and abnormal echocardiography or chest radiography findings. Laboratory or imaging criteria for heart failure exacerbation include NT-proBNP levels ≥125 pg/mL and higher than the baseline value, echocardiographic findings of reduced ejection fraction or chest radiographic findings indicating pulmonary congestion or cardiomegaly. Baseline NT-proBNP levels are defined as the value obtained at enrolment.
Registered primary clinical outcome
In this feasibility trial, the registered primary clinical outcome will be collected as an exploratory clinical outcome to estimate event frequency and inform the design of future definitive trials. The registered primary clinical outcome is defined as the time from randomisation to the composite endpoint of hospitalisation for heart failure or cardiovascular death during the 24-week follow-up period.
Based on prior studies,13 14 the composite endpoint components are defined as follows: (1) hospitalisation for heart failure, indicating unscheduled hospital admission with a primary diagnosis of heart failure and (2) cardiovascular death resulting from acute myocardial infarction, sudden cardiac death, heart failure, stroke, cardiovascular procedures, cardiovascular haemorrhage or other cardiovascular causes.13
To ensure standardised outcome ascertainment across the two study sites, clinical information related to potential cardiovascular hospitalisation and heart failure exacerbation events will be documented using predefined criteria at each clinic and reviewed by the clinical research secretariat. The primary diagnosis of hospitalisation or death due to heart failure will be determined through a review of medical records, including the documented clinical diagnosis, post-admission clinical course and treatments provided.
To minimise ascertainment bias in this open-label study, two physicians who are not directly involved in the clinical care of the study participants will independently adjudicate all events while remaining blinded to the group assignments. In cases of disagreement, a third physician will be consulted to reach a consensus.
Registered secondary clinical outcome
In this feasibility trial, the registered secondary clinical outcome will be collected as an exploratory clinical outcome. The secondary clinical outcome is the number of heart failure exacerbation events during the 24-week observation period following randomisation. Heart failure exacerbation events are defined according to the aforementioned operational definitions and will be adjudicated using the same standardised procedures described for the registered primary clinical outcome.
Sample size
This study is an exploratory feasibility trial; therefore, no formal statistical sample size calculation was performed owing to the lack of prior studies to inform the assumptions necessary for such an estimation. A target sample size of 60 participants was determined based on the expected number of eligible patients at the participating clinics.
Data collection and analysis
Data collection
Baseline demographic and clinical characteristics will include heart failure aetiology, previous hospitalisation for heart failure, comorbidities, medication use, natriuretic peptide data when available and echocardiographic findings when available. Laboratory tests are available in the home-based care setting; however, electrocardiography, echocardiography and chest radiography/CT are not routinely available at home. Therefore, when these imaging or diagnostic assessments are clinically indicated, they will be performed at the study-site clinics or at collaborating hospitals. Table 1 summarises the collected data according to a standardised format, including clinical symptoms, physical examination findings, laboratory test results and diagnostic assessments such as electrocardiography, echocardiography, chest radiography and phonocardiography.15
Table 1. Observation and examination items.
| Category | Details |
|---|---|
| Patient background (the attending physician measures items at baseline) |
Date of birth, sex, medical history and comorbidities (including history of cardiac surgery and the extent of comorbidities such as chronic kidney disease, atrial fibrillation, diabetes mellitus and chronic respiratory diseases); basis for heart failure diagnosis; medication history (specifically recording the use of sacubitril/valsartan (ARNI) and diuretics); vaccination history (pneumococcal, COVID-19 and influenza); and social history (smoking, alcohol consumption and primary caregiver). Past medical conditions existing prior to consent acquisition are also included |
| Phonocardiographic examination (intervention group only, the attending physician evaluates at each visit) | Phonocardiographic procedures and AI-guided analysis results |
| Symptoms (the attending physician evaluates at each visit) | Presence of the following symptoms: dyspnoea (shortness of breath), orthopnoea or malaise |
| Physical examination (the attending physician evaluates at each visit) | Vital signs: body temperature, blood pressure, pulse rate, respiratory rate and peripheral oxygen saturation. Physical findings: presence of characteristic signs such as the third heart sound or gallop rhythm and peripheral oedema |
| Treatment details (the attending physician evaluates as needed) | Detailed information on heart failure treatment, if administered |
| Laboratory tests (the attending physician orders tests as needed) | Haematology: white blood cell count, haemoglobin and platelet count. Biochemistry: prothrombin time, troponin T, NT-proBNP, BNP, C reactive protein, aspartate aminotransferase, alanine aminotransferase, total bilirubin, γ-glutamyl transpeptidase, blood urea nitrogen, creatinine, creatine kinase, lactate dehydrogenase, total protein, albumin, estimated glomerular filtration rate, total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, sodium, potassium, chloride, calcium and blood glucose |
| Electrocardiography (the attending physician orders tests as needed) | Atrial fibrillation (irregular R–R interval), ST–T changes and abnormal Q waves |
| Echocardiography (the attending physician orders tests as needed) | Left and right ventricular function, E/e′ ratio, valvular disease and pericardial effusion |
| Chest radiography/computed tomography (the attending physician orders tests as needed) | Presence of pulmonary congestion, pleural effusion or cardiomegaly |
Laboratory testing will be conducted by the laboratories originally contracted at each study site. For Japanese individuals, estimated glomerular filtration rate (eGFR) in men will be calculated using the following formula: eGFR=194×serum creatinine−1.094×age−0.287. For women, the calculated value is multiplied by 0.739.15
AI, artificial intelligence; ARNI, angiotensin receptor neprilysin inhibitor; BNP, B-type natriuretic peptide; NT-proBNP, N-terminal pro–B-type natriuretic peptide.
Statistical analysis
All analyses will be conducted according to the intention-to-treat principle, which will include all participants as originally randomised.
Analysis of feasibility and safety evaluation
Descriptive statistics will be used to summarise the feasibility outcomes, including recruitment rate, overall data completeness and completion rate and protocol-concordant completion rates of phonocardiography. The incidence proportions of safety evaluation items will be tabulated for each group.
Analysis of primary endpoint
The incidence rates of primary time-to-event outcomes will be estimated using the Kaplan-Meier method for each group. Because this is an exploratory study, formal hypothesis testing will not be performed.
Analysis of secondary endpoints
The incidence of heart failure exacerbation during the study period, along with the corresponding 95% CI, will be estimated for each group. The numerator for the incidence rate represents the total number of events in each group, whereas the denominator represents the total observation period for that group. If multiple heart failure exacerbations occur in the same participant, all events will be aggregated and counted as incident events. The incidence rate ratio and 95% CI for heart failure exacerbation will be estimated using a negative binomial regression model to compare intervention and non-intervention groups. If the negative binomial model fails to converge, the Poisson regression model will be used.
Subgroup analysis
The incidence rates of the primary and secondary outcomes will be estimated for each subgroup categorised by age and sex. Additionally, as ARNI therapy may increase circulating BNP levels while affecting NT-proBNP differently, exploratory descriptive analyses of the diagnostic output of the device will be stratified by the baseline use of sacubitril/valsartan to account for the potential confounding effect of neprilysin inhibition on BNP biology.16
Data monitoring
A formal independent data monitoring committee was not established for this study because it is a small-scale feasibility trial with a short follow-up period and minimal anticipated risk to the participants. Instead, study monitoring is conducted in accordance with a predefined monitoring standard operating procedure developed under the Japanese Clinical Trials Act.
Monitoring will be performed by designated monitors who are independent of the study conduct and not directly involved in patient care or data collection. The principal investigator will appoint qualified monitors and define the scope and frequency of monitoring activities. Monitoring includes document- and case-based monitoring and may be conducted either onsite or remotely, depending on feasibility and study needs.
The monitors will review study conduct, informed consent procedures, protocol adherence, data completeness and accuracy and participant safety by verifying source documents and case report forms. Monitoring will be performed at predefined time points, including prior to study initiation, during periodic reporting and before study completion. Serious adverse events, protocol deviations or non-compliance identified during monitoring will be promptly reported to the principal investigator, who is responsible for implementing corrective and preventive actions.
Monitoring reports will be documented and submitted to the principal investigator and relevant site investigators, and all monitoring records will be retained in accordance with regulatory requirements. This monitoring framework ensures data integrity, participant safety and compliance with applicable regulations throughout the study period.
Ethics and dissemination
This study was approved by the Nagasaki University Hospital Clinical Research Ethics Committee (CRB24-028) on 18 February 2025. Written informed consent will be obtained from all the participants or their caregivers. The findings will be disseminated through publications in peer-reviewed academic journals and scientific conference presentations.
Data management and governance
AMI Inc. is responsible only for generating the standard phonocardiography report from uploaded examination data and will not be involved in study conduct, endpoint adjudication, statistical analysis, manuscript preparation or publication decisions. Only anonymised phonocardiographic signal data will be uploaded to the cloud-based platform. The clinical study data will be separately maintained by the investigators and the clinical research secretariat.
Trial registration
This trial was registered in the Japan Registry of Clinical Trials (jRCT; jRCT1072240118) on 24 February 2025. The study period is planned from 24 February 2025 to 31 December 2027. Participant recruitment began on 2 June 2025 and is planned to continue until 31 July 2026.
Funding and conflicts of interest
This study is funded by Nagasaki University. Additional funding may be secured if appropriate grant support becomes available. JM, TM and HY have received a research grant from AMI Inc for a separate study. In this study, the role of AMI Inc during study conduct is strictly limited to providing product information and delivering the standard result report analysis routinely performed by a testing company in standard clinical practice. They will not be involved in accessing the study data, endpoint assessment, statistical analysis, manuscript preparation or publication decisions.
Patient and public involvement
Patients and/or the public were not involved in the design, conduct, reporting or dissemination of this study.
Discussion
This feasibility trial will evaluate the implementation of an AI-assisted phonocardiography system in home-based medical care for older adults with heart failure and assess its potential role in the early detection of heart failure exacerbations in home settings. A significant proportion of patients receiving home-based care in Japan are older individuals with multiple comorbidities who often lack access to conventional diagnostic modalities such as chest radiography, electrocardiography or echocardiography. Simple and practical diagnostic tools are essential for the timely detection of early clinical deterioration of heart failure or early symptomatic heart failure.9
Effective intervention programmes in primary care, particularly home-based medical care, remain undetermined and pose challenges for primary care physicians in providing optimal heart failure management. Among patients discharged after hospitalisation for chronic heart failure, readmission rates generally remain high globally, exceeding 20% within 30 days.17 Although disease management interventions for heart failure have been shown to reduce readmissions, their effectiveness varies by intervention type. In particular, case management interventions significantly reduce heart failure readmissions (risk ratio 0.64, 95% CI 0.53 to 0.78), whereas clinic-based interventions may have little or no effect (1.01, 0.87 to 1.18).18 However, many existing programmes remain predominantly hospital-based.
A qualitative study of Belgian primary care physicians has emphasised their recognition of their essential role in home-based heart failure care, as well as the challenges they encounter, including difficulties in assessing disease status through symptoms and physical findings, limited experience in heart failure management and barriers to collaboration with cardiologists.19 These findings underscore the need for practical assessment tools that enable early detection of heart failure exacerbations in home-based medical care settings.
Although biomarkers such as NT-proBNP are widely used, their interpretation in older patients is complicated by the high prevalence of renal dysfunction,6 20 21 which can result in elevated NT-proBNP levels even without acute worsening of heart failure, and the results are not always immediately available. Given these challenges, diagnostic tools that can be used easily and repeatedly in home-based medical care settings are urgently needed. Conventional cardiac auscultation is highly operator-dependent, demonstrating only fair and variable interobserver agreement for murmur classification and wide variability in diagnostic accuracy across studies.7 8 In contrast, the AMI-SSS01 device provides an objective and quantitative evaluation of heart sounds through AI-based analysis. This capability is particularly advantageous for primary care physicians who lack extensive experience in specialised cardiovascular care as it could facilitate the early identification of heart failure exacerbations during the early stages of clinical deterioration in the home care environment.
Previous remote patient monitoring programmes for heart failure, including the TIM-HF2 trial, have demonstrated clinical benefits using comprehensive telemonitoring systems involving the daily transmission of multiple physiological parameters and centralised telemedicine support.22 23 In contrast, the AMI-SSS01 system is designed as a lower-infrastructure, provider-driven, point-of-care tool integrated into routine physician home visits. Rather than replacing comprehensive remote monitoring systems, the present intervention is intended to be a complementary bedside screening approach using AI-assisted phonocardiographic assessment. This simplified workflow may be particularly relevant for older adults receiving home-based medical care, including patients with frailty or cognitive impairment who may have difficulty participating in intensive daily telemonitoring programmes.
This study protocol describes a randomised controlled feasibility trial designed to evaluate the implementation of the AMI-SSS01 phonocardiography system in home-based medical care. This study will explore the potential for early detection of heart failure exacerbations in home settings and evaluate the feasibility of the device in enhancing collaboration between primary and specialised care. The findings of this feasibility trial will inform the design of future large-scale studies and support the development of practical heart failure management strategies for home-based medical care.
Strengths and limitations
This study has several strengths. First, it will be conducted in a real-world home-based medical care setting where access to standard diagnostic modalities, such as echocardiography or chest radiography, is limited. This pragmatic design enhances the clinical relevance of these findings. Second, the study will evaluate a non-invasive AI-assisted phonocardiography system that enables objective and reproducible assessment of heart sounds, addressing the limitations of conventional auscultation. Finally, the intervention was designed to be implemented by primary care physicians without specialised cardiovascular expertise, supporting its potential for scalability and applicability in routine home-based care.
However, this study has several limitations. First, although the AI-assisted phonocardiography system shows promise, the diagnostic accuracy and prognostic predictive performance of the AMI-SSS01 device have not been sufficiently validated against conventional diagnostic standards. Second, this trial will be conducted at a limited number of sites within a single region, which may limit the generalisability of the findings to other healthcare settings. Third, blinding of physicians and participants will not be feasible because of the nature of the intervention, introducing the potential for performance and observation bias. Fourth, the outcome assessments, including the identification of heart failure exacerbation events, rely partly on clinical judgement based on a review of medical records, which may involve a degree of subjectivity. Fifth, as this feasibility trial is designed for a home-based medical care setting, baseline standardised echocardiographic and biomarker assessments are not mandatory. Accordingly, participant phenotyping may be less precise than that in hospital-based heart failure studies, potentially resulting in greater clinical heterogeneity and variable interpretation of phonocardiographic findings across heart failure phenotypes. However, this pragmatic design was intentionally adopted to reflect the real-world constraints and diagnostic practices of home-based medical care for older adults. Sixth, because standardised echocardiographic phenotyping is not mandatory in this home-based feasibility study, the interpretation of phonocardiographic findings across heart failure phenotypes may be heterogeneous. Seventh, the interpretation of AI-derived findings may be influenced by sacubitril/valsartan therapy, which can increase circulating BNP concentrations through neprilysin inhibition, because the AI-assisted system estimates BNP-related physiological signals. Finally, formal health economic evaluations are beyond the scope of this feasibility trial. Future studies should assess implementation costs, reimbursement considerations and cost-effectiveness in home-based medical care settings.
This trial will clarify the feasibility of integrating this device into a comprehensive home-based heart failure management model. Further large-scale, multicentre, collaborative studies will be needed to evaluate the clinical utility of this approach.
Supplementary material
Acknowledgements
We are deeply grateful to all the members of the ROEN-PBRN study group. We extend our sincere gratitude to Takashi Kambara and Yuka Shigemura for their technical assistance. We would like to thank Editage (www.editage.jp) for English language editing.
Footnotes
Funding: This work was supported by a research grant from Nagasaki University, Japan (no grant number). The funder did not influence the results/outcomes of the study despite author affiliations with the funder.
Prepublication history for this paper is available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2026-117953).
Patient consent for publication: Not applicable.
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Collaborators: the ROEN-PBRN Study Group: Tetsuro Matsushita, Hideki Mori, Atsushi Taira, Ichiro Yoshii
Contributor Information
the ROEN-PBRN Study Group:
Tetsuro Matsushita, Hideki Mori, Atsushi Taira, and Ichiro Yoshii
References
- 1.Yan T, Zhu S, Yin X, et al. Burden, Trends, and Inequalities of Heart Failure Globally, 1990 to 2019: A Secondary Analysis Based on the Global Burden of Disease 2019 Study. J Am Heart Assoc. 2023;12:e027852. doi: 10.1161/JAHA.122.027852. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Savarese G, Becher PM, Lund LH, et al. Global burden of heart failure: a comprehensive and updated review of epidemiology. Cardiovasc Res. 2023;118:3272–87. doi: 10.1093/cvr/cvac013. [DOI] [PubMed] [Google Scholar]
- 3.Shimokawa H, Miura M, Nochioka K, et al. Heart failure as a general pandemic in Asia. Eur J Heart Fail. 2015;17:884–92. doi: 10.1002/ejhf.319. [DOI] [PubMed] [Google Scholar]
- 4.Okura Y, Ramadan MM, Ohno Y, et al. Impending epidemic: future projection of heart failure in Japan to the year 2055. Circ J. 2008;72:489–91. doi: 10.1253/circj.72.489. [DOI] [PubMed] [Google Scholar]
- 5.Bauersachs J, de Boer RA, Lindenfeld J, et al. The year in cardiovascular medicine 2021: heart failure and cardiomyopathies. Eur Heart J. 2022;43:367–76. doi: 10.1093/eurheartj/ehab887. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Nishikimi T, Nakagawa Y. Potential pitfalls when interpreting plasma BNP levels in heart failure practice. J Cardiol. 2021;78:269–74. doi: 10.1016/j.jjcc.2021.05.003. [DOI] [PubMed] [Google Scholar]
- 7.Davidsen AH, Andersen S, Halvorsen PA, et al. Diagnostic accuracy of heart auscultation for detecting valve disease: a systematic review. BMJ Open. 2023;13:e068121. doi: 10.1136/bmjopen-2022-068121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Andersen S, Davidsen AH, Schirmer H, et al. Interrater and intrarater agreement on heart murmurs. Scand J Prim Health Care. 2022;40:491–7. doi: 10.1080/02813432.2022.2159204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Heidenreich PA, Bozkurt B, Aguilar D, et al. 2022 AHA/ACC/HFSA Guideline for the Management of Heart Failure: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. J Am Coll Cardiol. 2022;79:e263–421. doi: 10.1016/j.jacc.2021.12.012. [DOI] [PubMed] [Google Scholar]
- 10.Ogawa S, Namino F, Mori T, et al. AI diagnosis of heart sounds differentiated with super StethoScope. J Cardiol. 2024;83:265–71. doi: 10.1016/j.jjcc.2023.09.007. [DOI] [PubMed] [Google Scholar]
- 11.Ogawa S, Ishii M, Saito S, et al. Deep Learning for Cardiac Overload Estimation - Predicting B-Type Natriuretic Peptide (BNP) Levels From Heart Sounds and Electrocardiogram. Circ J. 2025;89:1684–92. doi: 10.1253/circj.CJ-25-0098. [DOI] [PubMed] [Google Scholar]
- 12.Yamanashi H, Dania A, Kaneko M, et al. Practice-Based Research Network Activities and Future Nationwide Plans in Japan. J Gen Fam Med. 2026;27:e70091. doi: 10.1002/jgf2.70091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hicks KA, Mahaffey KW, Mehran R, et al. 2017 Cardiovascular and Stroke Endpoint Definitions for Clinical Trials. Circulation. 2018;137:961–72. doi: 10.1161/CIRCULATIONAHA.117.033502. [DOI] [PubMed] [Google Scholar]
- 14.Bozkurt B, Coats AJ, Tsutsui H, et al. Universal Definition and Classification of Heart Failure. J Card Fail. 2021;27:387–413. doi: 10.1016/j.cardfail.2021.01.022. [DOI] [PubMed] [Google Scholar]
- 15.Matsuo S, Imai E, Horio M, et al. Revised equations for estimated GFR from serum creatinine in Japan. Am J Kidney Dis. 2009;53:982–92. doi: 10.1053/j.ajkd.2008.12.034. [DOI] [PubMed] [Google Scholar]
- 16.Vasquez N, Carter S, Grodin JL. Angiotensin Receptor-Neprilysin Inhibitors and the Natriuretic Peptide Axis. Curr Heart Fail Rep. 2020;17:67–76. doi: 10.1007/s11897-020-00458-y. [DOI] [PubMed] [Google Scholar]
- 17.Feltner C, Jones CD, Cené CW, et al. Transitional care interventions to prevent readmissions for persons with heart failure: a systematic review and meta-analysis. Ann Intern Med. 2014;160:774–84. doi: 10.7326/M14-0083. [DOI] [PubMed] [Google Scholar]
- 18.Takeda A, Martin N, Taylor RS, et al. Disease management interventions for heart failure. Cochrane Database Syst Rev. 2019;1:CD002752. doi: 10.1002/14651858.CD002752.pub4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Smeets M, Zervas S, Leben H, et al. General practitioners’ perceptions about their role in current and future heart failure care: an exploratory qualitative study. BMC Health Serv Res. 2019;19:432. doi: 10.1186/s12913-019-4271-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Vergaro G, Januzzi JL, Jr, Cohen Solal A, et al. NT-proBNP prognostic value is maintained in elderly and very elderly patients with chronic systolic heart failure. Int J Cardiol. 2018;271:324–30. doi: 10.1016/j.ijcard.2018.04.006. [DOI] [PubMed] [Google Scholar]
- 21.Pfister R, Schneider CA. Natriuretic peptides BNP and NT-pro-BNP: established laboratory markers in clinical practice or just perspectives? Clin Chim Acta. 2004;349:25–38. doi: 10.1016/j.cccn.2004.06.018. [DOI] [PubMed] [Google Scholar]
- 22.Koehler F, Koehler K, Deckwart O, et al. Efficacy of telemedical interventional management in patients with heart failure (TIM-HF2): a randomised, controlled, parallel-group, unmasked trial. Lancet. 2018;392:1047–57. doi: 10.1016/S0140-6736(18)31880-4. [DOI] [PubMed] [Google Scholar]
- 23.Koehler F, Koehler K, Deckwart O, et al. Telemedical Interventional Management in Heart Failure II (TIM-HF2), a randomised, controlled trial investigating the impact of telemedicine on unplanned cardiovascular hospitalisations and mortality in heart failure patients: study design and description of the intervention. Eur J Heart Fail. 2018;20:1485–93. doi: 10.1002/ejhf.1300. [DOI] [PubMed] [Google Scholar]