A Quantum-Inspired AI System for Malnutrition Prediction

Abstract

Malnutrition continues to threaten public health across many developing regions, where late identification of at-risk individuals often translates into avoidable suffering and long-term developmental harm. This paper introduces a quantum-inspired artificial intelligence framework that brings together natural language processing, classical machine learning, and parameterized quantum circuits to predict malnutrition risk from textual and socioeconomic information. Raw inputs are first cleaned, tokenized, and converted into compact numerical features. These features are then encoded into quantum states using Qiskit-based circuits, allowing the model to explore richer feature relationships than a purely classical pipeline can capture. A hybrid quantum classical classifier is trained using gradient-based optimization, and its quality is evaluated using accuracy, precision, recall, and F1score on held-out data. Experimental results show that the proposed pipeline produces stable, highconfidence predictions and behaves reliably on heterogeneous inputs. By combining established machine learning practices with quantum-inspired feature mapping, the system offers a practical, scalable, and easily deployable tool that can support early screening, timely intervention, and data driven public-health decision making.

Introduction

This paper proposes a hybrid quantum-classical machine learning framework for malnutrition risk prediction using demographic, nutritional, socio-economic, and textual health data. It addresses the challenge of delayed malnutrition detection by combining traditional data analytics with quantum-inspired feature encoding implemented through Qiskit simulations. The system preprocesses structured and unstructured data using NLP techniques such as text cleaning, TF-IDF vectorization, feature scaling, and feature selection before encoding features into parameterized quantum circuits. A hybrid architecture integrates quantum feature mapping with classical optimization and classification, enabling improved detection of complex, non-linear relationships among risk factors.

The proposed model is trained using gradient-based optimization and evaluated with accuracy, precision, recall, and F1-score metrics. Experimental results show that the hybrid approach outperforms classical models such as Random Forest, SVM, and XGBoost, achieving approximately 98% accuracy, 97.4% precision, 97.8% recall, and 97.6% F1-score, particularly improving identification of high-risk malnutrition cases. The framework is lightweight, reproducible, deployable on standard hardware, and designed as a decision-support tool for healthcare workers rather than a replacement for clinical judgment.

The study highlights advantages such as handling both numerical and textual inputs, modular design, compatibility with existing healthcare systems, and future scalability toward real quantum hardware. However, limitations include reliance on partially synthetic datasets, challenges in interpreting quantum representations, and the need for validation on larger and more diverse populations. Future work includes multilingual support, IoT integration, federated learning, explainable AI, and the incorporation of additional environmental and economic indicators to enhance malnutrition prediction and public-health decision-making.

Conclusion

This paper presented a quantum-inspired artificial intelligence framework for early malnutrition risk prediction. The system blends standard NLP preprocessing, Qiskit-based quantum feature encoding, and gradient-trained hybrid classification into a single deployable pipeline. Experimental results indicate strong accuracy and balanced precision-recall behaviour, with the hybrid model consistently outperforming purely classical baselines on records that require subtle reasoning across mixed signals. By keeping the architecture modular, transparent, and runnable on commodity hardware, the proposed framework is positioned as a practical building block for early-detection workflows in resource-constrained healthcare settings. The broader takeaway is that quantum-inspired components can be incorporated into real public-health pipelines today, providing a measurable representational benefit without waiting for fault-tolerant quantum hardware. Looking ahead, the most important next step is to validate the framework on real, prospectively collected datasets and to integrate it into the day-to-day workflow of community health programs. The authors hope that this work serves as a useful reference point for other teams exploring the intersection of quantum inspired computing and public-health analytics, and that it encourages further open, reproducible research at this intersection.

References

[1] L. J. Sánchez-Martínez et al., \"Using machine learning to fight child acute malnutrition,\" Nutrients, vol. 16, no. 23, p. 4213, 2024. [2] R. S. Gupta, \"A systematic review of quantum machine learning for healthcare,\" J. Healthcare Informatics Res., 2025. [3] D. Pomarico et al., \"A proposal of quantum-inspired machine learning for medical purposes,\" Mathematics, vol. 9, no. 4, p. 410, 2021. [4] D. Vatsavai et al., \"Quantum-inspired machine learning approach for high-dimensional pattern detection,\" Sci. Rep., 2025. [5] Sharma, \"Role of nutritional adequacy and machine learning in health-science prediction systems,\" J. Clin. Med., 2020. [6] M. Schuld and F. Petruccione, Supervised Learning with Quantum Computers. Cham, Switzerland: Springer, 2018. [7] J. Biamonte et al., \"Quantum machine learning,\" Nature, vol. 549, pp. 195–202, 2017. [8] F. Pedregosa et al., \"Scikit-learn: Machine learning in Python,\" J. Mach. Learn. Res., vol. 12, pp. 2825–2830, 2011. [9] World Health Organization, \"Levels and trends in child malnutrition,\" UNICEF/WHO/World Bank Joint Child Malnutrition Estimates, 2023. [10] Abbas et al., \"The power of quantum neural networks,\" Nat. Computational Science, vol. 1, pp. 403– 409, 2021. [11] V. Havlí?ek et al., \"Supervised learning with quantumenhanced feature spaces,\" Nature, vol. 567, pp. 209–212, 2019. [12] M. Cerezo et al., \"Variational quantum algorithms,\" Nat. Rev. Phys., vol. 3, pp. 625–644, 2021.

Copyright

Copyright © 2026 Balineni Yaswanth , Pasupuleti Naga Lakshmi Priyanka , Kandru Prabhu Kumar, Kasina Santha Lahari , D. Venkatesh. This 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.