Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{209081,
        author = {AMAN GUPTA and Himanshu Gupta and Anshu and Dr. Tripti Choudhary},
        title = {MatruCare-AI: A RAG-Augmented Multi-Agent Decision Support & Framework for ASHA-Mediated High-Risk Pregnancy Detection in Rural India},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {5},
        pages = {813-833},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=209081},
        abstract = {Maternal mortality in rural India remains a persistent public health challenge, with the country recording 97 deaths per 100,000 live births [13]. Geo-graphic inaccessibility of tertiary care facilities means that a large proportion of pregnant women in remote areas complete their entire antenatal period without consulting a physician [15]. Accredited Social Health Activists (ASHAs) constitute the primary interface between rural pregnant women and the formal health system, yet a meta-analysis of 37 studies established that ASHAs respond correctly to maternal health queries at a rate of approximately 62% [1]. Epidemiological surveys indicate that around 40% of rural pregnancies carry some degree of risk, whereas national programme records flag only 14% as high-risk [14], implying that the majority of at-risk mothers are not identified before complications develop.
Existing digital tools deployed for ASHAs, including e-SAATHI and ASHA-Soft, rely on pre-programmed rule sets that cannot reason beyond their fixed parameters, detect evolving risk patterns across consecutive antenatal visits, or adapt when national clinical guidelines are revised [16], [17].
This paper proposes MatruCare-AI, a three-layer framework designed specifically for ASHA workers in rural India. Layer 1 is a MERN-based Progressive Web Application that records antenatal parameters and computes a four-tier risk classification. Layer 2 is a Retrieval-Augmented Generation (RAG) clinical engine that grounds advisory responses in India’s own governmental health protocols, including NHM Maternal Health guidelines, PMSMA clinical criteria, and WHO ANC 2016 recommendations. Layer 3 operates three parallel decision-support agents responsible for danger-sign detection, referral routing to the nearest Primary Health Centre, and antenatal compliance scheduling. This paper makes three contributions. First, to the best of the authors’ knowledge, MatruCare-AI represents the first framework to combine RAG-based clinical advisory with multi-agent decision support built specifically for ASHA-mediated high-risk pregnancy detection in rural India. Second, the proposed clinical knowledge base draws exclusively from Indian governmental protocols, ensuring contextual alignment with the NHM programme environment. Third, a systematic gap matrix confirms that no existing framework simultaneously satisfies all five required dimensions: ASHA focus, RAG advisory, HRP detection, rural India context, and multilingual support.},
        keywords = {ASHA workers, high-risk pregnancy, ma-ternal health, rural India, retrieval-augmented generation, multi-agent system, clinical decision support, antenatal care, mobile health, large language models},
        month = {October},
        }