Researchers developed artificial intelligence–based models to assess dehydration severity for patients with acute diarrhea: the DHAKA models for children under five and the NIRUDAK models for patients aged five and older. In prior large studies in Bangladesh these models were reported by the authors to outperform WHO IMCI and IMAI guidelines in accuracy and reliability for dehydration assessment. The models were incorporated into a mobile health clinical decision support tool (FluidCalc) intended to support frontline health workers in improving fluid management and overall care for patients with acute diarrhea.
The present study aimed to identify barriers and facilitators to uptake and use of FluidCalc in both a low-resource setting (Tanzania) and a high-resource setting (United States). The Consolidated Framework for Implementation Research (CFIR) guided data collection and analysis to elicit perceptions from healthcare providers, stakeholders, and policymakers about integrating the CDST into clinical practice.
Qualitative data were collected through focus group discussions (FGDs) with healthcare providers and in-depth interviews (IDIs) with stakeholders and policymakers. FGDs and IDIs occurred in Tanzania from February to July 2025 and in the United States from February to March 2026. In total, 35 providers from varied cadres participated in FGDs and 13 stakeholders took part in IDIs.
Audio recordings of FGDs and IDIs were transcribed, and Swahili-language recordings were translated to English where applicable. The research team analyzed transcripts using framework matrix analysis mapped to CFIR domains to identify common themes regarding intervention characteristics, outer and inner setting influences, characteristics of individuals, and implementation processes.
Participants identified multiple facilitators that could enable adoption of FluidCalc in clinical settings. Key facilitator themes included:
Simplicity and ease of use: Providers emphasized that the app’s straightforward interface and functionality would lower training burden and support rapid uptake across cadres.
Offline capability: The ability to function without continuous internet connectivity was viewed as critical in low-resource settings and during outbreaks when networks may be strained.
Workflow integration: Participants reported that FluidCalc could streamline clinical workflows by standardizing dehydration assessment and fluid rehydration calculations, potentially reducing variability and error.
Support for guideline adherence and stewardship: Respondents believed the tool could promote adherence to recommended diarrhea-management protocols and support antibiotic stewardship by aiding decision-making about when antibiotics were indicated.
Task shifting and training: FluidCalc was seen as a valuable teaching tool and a support for less experienced healthcare workers and trainees, enabling task shifting while maintaining a consistent approach to patient assessment.
Outbreak utility: Several participants noted the app’s potential usefulness during diarrheal disease outbreaks for rapid, standardized assessments across settings.
Overall, stakeholders and providers considered these features to increase the likelihood of beneficial clinical impact if the tool were implemented effectively.
Participants also described several barriers that could impede implementation:
Digital infrastructure requirements: Reliable access to mobile devices, dependable electricity, and internet connectivity were repeatedly cited as prerequisites that are not consistently available in all settings, particularly in low-resource environments.
Institutional approval processes: Lengthy facility-level and regulatory approval pathways were perceived as a significant barrier to timely integration into routine care.
Need for high-level endorsement: Endorsement and approval from the Ministry of Health and health facility leadership were considered essential for widespread adoption and scale-up; lack of such endorsement could hinder implementation.
Resource constraints: Although not quantified in the report, participants highlighted concerns about device procurement, maintenance, and long-term support for digital health tools.
These barriers were framed as addressable but requiring planning, investment, and coordination with health system authorities.
Across settings, the study participants recommended strategies to improve the feasibility of implementing FluidCalc:
Prioritize reliable digital health infrastructure including device availability, power solutions, and offline functionality where connectivity is limited.
Engage health facility leaders and Ministries of Health early to secure endorsement, streamline approvals, and align the tool with existing policies and workflows.
Use the app’s teaching potential to support training programs and enable task shifting to less experienced cadres while maintaining quality of care.
Leverage the tool during outbreak responses and quality-improvement initiatives to demonstrate clinical value and build institutional buy-in.
The authors report that addressing these considerations would be necessary to support effective integration of FluidCalc into patient care.
Ethical approval for the work was granted by the ethics committees/IRBs of Muhimbili University of Health and Allied Sciences and Brown University. The investigators confirm that participant consent procedures and relevant reporting guidelines were followed. The dataset, which includes individual transcripts, is not publicly available due to confidentiality policies. Funding was provided by the National Institutes of Health, specifically the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), under R01DK116163. The authors declared no competing interests.
Healthcare providers and stakeholders in both Tanzania and the United States perceived FluidCalc as a promising clinical decision support tool that could improve care for patients with acute diarrhea by standardizing dehydration assessment, guiding fluid rehydration calculations, promoting guideline adherence, supporting antibiotic stewardship, and aiding training. Key barriers to successful implementation include the need for dependable digital infrastructure, device access, reliable electricity and connectivity, and securing endorsement from Ministries of Health and facility leadership. The study suggests that addressing infrastructure and approval requirements is necessary to enable effective integration of FluidCalc into routine clinical practice.