AtulyaU: A Context-Aware Mobile Health Architecture for Closed-Loop Dietary Deficit Telemetry and Clinically Guarded Precision Supplementation
- 1Birla Institute of Technology and Science, Pilani
Abstract
Preventive healthcare and chronic disease self-management (e.g., Type 2 Diabetes, Polycystic Ovary Syndrome [PCOS], iron-deficiency anemia, and hypertension) depend fundamentally on daily micronutrient adequacy, metabolic balance, and bioavailable nutrient absorption. However, contemporary mobile health (mHealth) solutions suffer from a critical clinical disconnect: commercial trackers provide passive caloric tallies without analyzing actual nutrient shortages, while consumer health culture promotes "blind supplementation" popping generic multivitamins without measuring dietary intake, risking both chronic deficiencies and micronutrient toxicity.
In this paper, we present AtulyaU, an end-to-end context-aware mHealth architecture built on a core clinical innovation: Closed-Loop Dietary Deficit Quantification and Dynamic Precision Supplementation. AtulyaU decomposes real-time meal telemetry from a 235+ regional food ontology into bioavailable macro- and micronutrient profiles, computes the patient's personalized clinical target vector based on chronic conditions and hormonal milestones, and derives an exact, dynamic Nutrient Shortage Gap Vector. Rather than prescribing generic regimens, AtulyaU's Clinical Context Injection (CCI) engine drives state-of-the-art LLMs (Gemini 2.5 Flash) to generate a prioritized, two-tier intervention: (1) food-first nutrient bridging within caloric envelopes, followed by (2) precision micro-supplementation calibrated strictly to the residual shortage.
Enforcing a Stable Prefix Protocol with disk-level KV caching and a Deterministic Safety Boundary (DSB), AtulyaU eliminates absorption contraindications (e.g., competitive calcium-iron inhibition at the DMT-1 transporter, lipid-carrier requirements for fat-soluble vitamins). Across 150 clinical scenarios in 5 disease cohorts, AtulyaU achieves 98.4% clinical guideline adherence and 100% contraindication safety, eliminating both micronutrient under-coverage and toxicity risks inherent in ungrounded consumer LLMs.
Key Contributions:
1. Closed-Loop Dietary Deficit Quantification: Decomposing consumed meals to compute real-time micronutrient shortages against clinical targets.
2. Dynamic Precision Supplementation: Prioritizing whole-food gap closure followed by micro-supplementation calibrated strictly to the residual shortage.
3. Type-Safe Edge-Cloud Architecture: End-to-end tRPC/Zod validation between React Native client and Gemini LLM backend.
4. Stable Prefix Caching & Safety Boundaries: 57.0% latency reduction via disk caching and 100% contraindication prevention.
Related
- Decentralized Smart-Hive IoT Architecture and Open Field Telemetry Dataset
- The effect of macular carotenoids and omega-3 supplementation on performance in precision sports: the sport nutrition intervention in performance exercise (SNIPE) study
- Associations Between Microscopic Balantioides coli Counts and Precision Dairy-Farming Indicators in Clinically Healthy Dairy Cows.
- Toward Precision Imaging in Lung NET: A Clinically Oriented Framework Integrating Dual Tracer PET and Radiomics.
- Genetics of Premature Ovarian Insufficiency: From Molecular Architecture to Precision Medicine.
BibTeX
@misc{singh2026atulyau,
title = {{AtulyaU: A Context-Aware Mobile Health Architecture for Closed-Loop Dietary Deficit Telemetry and Clinically Guarded Precision Supplementation}},
author = {SAMEER SINGH},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23659689},
url = {https://zenodo.org/records/23659689},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - AtulyaU: A Context-Aware Mobile Health Architecture for Closed-Loop Dietary Deficit Telemetry and Clinically Guarded Precision Supplementation
AU - SAMEER SINGH
PY - 2026
DA - 2026/10/26
PB - Zenodo
DO - 10.5281/zenodo.23659689
UR - https://zenodo.org/records/23659689
AB - Preventive healthcare and chronic disease self-management (e.g., Type 2 Diabetes, Polycystic Ovary Syndrome [PCOS], iron-deficiency anemia, and hypertension) depend fundamentally on daily micronutrient adequacy, metabolic balance, and bioavailable nutrient absorption. However, contemporary mobile health (mHealth) solutions suffer from a critical clinical disconnect: commercial trackers provide passive caloric tallies without analyzing actual nutrient shortages, while consumer health culture promotes "blind supplementation" popping generic multivitamins without measuring dietary intake, risking both chronic deficiencies and micronutrient toxicity. In this paper, we present AtulyaU, an end-to-end context-aware mHealth architecture built on a core clinical innovation: Closed-Loop Dietary Deficit Quantification and Dynamic Precision Supplementation. AtulyaU decomposes real-time meal telemetry from a 235+ regional food ontology into bioavailable macro- and micronutrient profiles, computes the patient's personalized clinical target vector based on chronic conditions and hormonal milestones, and derives an exact, dynamic Nutrient Shortage Gap Vector. Rather than prescribing generic regimens, AtulyaU's Clinical Context Injection (CCI) engine drives state-of-the-art LLMs (Gemini 2.5 Flash) to generate a prioritized, two-tier intervention: (1) food-first nutrient bridging within caloric envelopes, followed by (2) precision micro-supplementation calibrated strictly to the residual shortage. Enforcing a Stable Prefix Protocol with disk-level KV caching and a Deterministic Safety Boundary (DSB), AtulyaU eliminates absorption contraindications (e.g., competitive calcium-iron inhibition at the DMT-1 transporter, lipid-carrier requirements for fat-soluble vitamins). Across 150 clinical scenarios in 5 disease cohorts, AtulyaU achieves 98.4% clinical guideline adherence and 100% contraindication safety, eliminating both micronutrient under-coverage and toxicity risks inherent in ungrounded consumer LLMs. Key Contributions: 1. Closed-Loop Dietary Deficit Quantification: Decomposing consumed meals to compute real-time micronutrient shortages against clinical targets. 2. Dynamic Precision Supplementation: Prioritizing whole-food gap closure followed by micro-supplementation calibrated strictly to the residual shortage. 3. Type-Safe Edge-Cloud Architecture: End-to-end tRPC/Zod validation between React Native client and Gemini LLM backend. 4. Stable Prefix Caching & Safety Boundaries: 57.0% latency reduction via disk caching and 100% contraindication prevention.
ER -