NutriScan: An AI-Driven System for Child Malnutrition Detection and Region-Aware Dietary Recommendation

NutriScan: An AI-Driven System for Child Malnutrition Detection and Region-Aware Dietary Recommendation | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article NutriScan: An AI-Driven System for Child Malnutrition Detection and Region-Aware Dietary Recommendation Disha Shirbad This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8744190/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 12 You are reading this latest preprint version Abstract Child malnutrition has been a severe issue of population health especially in third world countries where prompt diagnosis and access to nutritional intervention result in chronic physical and intellectual disability. Traditional methods of malnutrition evaluation are based on manual anthropometric measurements and do not in most cases give effective dietary advice. In this paper, the author is introducing the AI-based web system called NutriScan, which will allow early detection of malnutrition in children and dietary recommendations depending on their location. The suggested system involves the straightforward anthropometric data of age, height, and weight that are incorporated to calculate the standardized nutritional indicators based on the World Health Organization (WHO) growth standards. The models used are machine learning models (Logistic Regression and random forest) which categorize children into normal, stunted, wasted, and underweight and yield a risk score that is used to show severity. To overcome ethical and privacy limitations of real world child health data, a synthetic data set was created in Python without much distortion of realistic anthropometric distributions. Besides automated malnutrition data, NutriScan has a diet recommendation module which uses the Indian Council of Medical Research (ICMR) Food Composition Table to suggest region-specific diets in Maharashtra. The experimental analysis shows that ensemble-based models present consistent performance in classification. NutriScan promotes an effective application of nutritional intervention as malnutrition identification is combined with localized dietary advice. Child Malnutrition Artificial Intelligence Machine Learning Anthropometric Indicators WHO Growth Standards Risk Assessment Region-Specific Diet Recommendation Web-Based Healthcare System Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 I. INTRODUCTION 1.1 Background of Child Malnutrition Child malnutrition remains as one of the most urgent issues in the global public health, especially in low- and middle-income countries. Reports show that malnutrition contributes to the morbidity and mortality of children below the age of five according to reports by international health organizations. Stunting, wasting, and underweight are relatively common conditions that severely impact physical development, cognitive, immunological, and future productivity levels [1], [2]. Child malnutrition has been a thorn in the flesh in India even with a series of government programs and nutrition plans. Such surveys as the National Family Health Survey (NFHS) show that a significant number of children experience chronic and acute malnutrition and that they should be better screened and the intervention mechanisms should be improved [3]. Insufficient dietary consumption is not the only cause of malnutrition, which is also caused by socio-economic factors, sanitation, maternal education, and access to health care [4]. The need to identify malnutrition early before it is too late is necessary to ensure that there is no irreversible loss of vitality in the young during a crucial time of child development. However, traditional assessment tools tend to be manual, time consuming, and require a trained health care practitioner, which constrain their resource use in resource limited settings. 1.2 Artificial Intelligence and its role in healthcare. The latest development of the Artificial Intelligence (AI) and Machine Learning (ML) has revolutionized several fields of healthcare, such as disease diagnostics, risk forecasting, and clinical decision support. The use of AI-based systems is able to analyze health information automatically, which can uncover diseases that otherwise cannot be detected [5], [6]. Machine learning models have been specifically useful in uncovering non-linear and complicated relationships in healthcare data. Within the framework of child malnutrition, the ML methods have been effectively used in predicting nutritional status by the analysis of anthropometric and demographic data [7]. Random Forest and other types of the ensemble models have been shown to be more effective because of their strength and capability to interact among features [8]. Regardless of these benefits, most current AI-based health solutions pay more attention to the accuracy of prediction and do not consider practical intervention plans. In the case of malnutrition management, prediction cannot be conducted without actionable guidance that can guide caregivers and healthcare workers to help in managing this condition. 1 .3 Motivation for the Study The impetus to base this research on the fact that there is a gap between diagnosis and nutritional intervention of malnutrition. Although a number of studies have revealed the efficacy of machine learning in forecasting malnutrition, majority of the systems conclude with classification and fail to offer information on how to treat the identified problem [9]. Moreover, the diet advice tends to be generalized and does not reflect on the food supply in the region, the cultural eating lifestyles, and the national nutritional policies. Dietary planning based on regions is important in a diverse nation such as India, where the nutrition interventions must be adhered to and feasible [10]. The rising accessibility of web-based technologies has presented a possibility to create scalable and accessible healthcare solutions. The malnutrition monitoring systems can be improved in terms of their applicability in the real world with the integration of AI-based detection and region-specific dietary recommendations. The necessity of such an integrated, intelligent, and easy to use solution is the reason behind this study. 1.4 Problem Statement Despite the fact that a number of machine learning-based systems are suggested with the purpose of child malnutrition detection, several limitations still exist. The current strategies are all concerned with predictive performance and do not consider the post-diagnosis intervention mechanisms [11]. Also, there are numerous systems which are based on large-scale real-world data that presents ethical, privacy, and access issues. One more significant weakness is the lack of localized dietary recommendations. The generalized nutrition information fails to cover the local eating habits or the availability of local foodstuffs, which diminishes the impact of the intervention measures [12]. Hence, it is required that an AI-based system be able to not only detect child malnutrition properly, but also issue actionable and region-specific dietary recommendations in an ethical and scalable way. 1. 5 Objectives of the Study · The main aim of the study is as follows: · The purpose of the research is to design and develop an AI-based system to early detect malnutrition in children. To categorise children to nutritional groups, which include normal, stunted, wasted, and underweight. · To produce a risk score of the degree of malnutrition. · To give localized nutritional advice using local nutritional guidelines. · To create a web-based platform which is accessible, scalable, and user friendly. · These goals shall help fill the gap in recognizing malnutrition and providing nutritional intervention through clever decision-making processes. 1.6 Organization of the Paper The rest of this paper is structured in the following way. Section 2 contains an extensive literature review of the existing solutions on the topic of child malnutrition detection and healthcare technologies. Section 3 explains the proposed NutriScan system and the general architecture. Section 4 gives out the methodology which includes the preparation of the datasets, machine learning models, and diet recommendation strategy. Section 5 deals with the performance evaluation and results of the experiment. Findings are discussed in detail in section 6. Lastly, Section 7 concludes the paper and gives future research directions. II. LITERATURE REVIEW The study will involve machine learning methods to identify child malnutrition. A number of researchers have conducted studies to assess the use of machine learning to predict child malnutrition with the help of anthropometric and demographic information. The objectives of these methods are to enhance the early diagnosis through automating health indicators like age, height and weight. In their study, Bhagya Jyothi Rao and colleagues conducted a meta-analysis of the performances of different machine learning models to predict child malnutrition on their demographic and health surveys data. [13]. The paper established a comparison with the Logistic Regression, Random Forest and Support Vector machine models and indicated that the ensemble-based approaches yielded an accuracy of about 88. The authors emphasized the fact that tree-based models are good because they are capable of non-linear relationships between features. Nevertheless, their study was more concerned with the predictive accuracy and has not considered the dietary intervention mechanisms. In the same way, Das and Rahman used machine learning classifiers to establish the malnutrition status by anthropometric indicators and household survey data.[14]. Their findings showed that the random forest models are stronger than the linear classifiers in the classification task. Though it was more accurate, the system did not have customized or area specific nutritional suggestions and thus, could not be utilized practically. These works show the usefulness of machine learning in the detection of malnutrition and also indicate a persistent weakness, namely most methods conclude with classification and do not include any actionable healthcare recommendation. Sr. No. Title, Authors & Year Objective & Method Key Findings Remarks 1 Machine Learning in Predicting Child Malnutrition – Rao et al. (2025) Meta-analysis using Logistic Regression, Random Forest, SVM Achieved ~88% accuracy in classification Ensemble models perform best 2 ML for Acute Child Malnutrition – Sánchez-Martínez et al. (2024) Random Forest with feature selection Identified feeding frequency, birth weight as key factors Helps optimize treatment 3 Deep Learning-Based Malnutrition Detection – Biradar et al. (2024) CNN with transfer learning Achieved 92% accuracy Requires image data 4 Blockchain-Based Healthcare Framework – Tahir et al. (2024) Secure EHR using blockchain Improved data integrity No prediction module 5 Determinants of Child Malnutrition in India – Singh et al. (2023) Logistic regression using NFHS-5 Socio-economic factors significant Lacks automation 2.2 Machine Learning of Acute Malnutrition and Treatment Results. In addition to the detection, a few researchers have also concentrated on predicting acute malnutrition and treatment outcomes based on machine learning. SANTE-only models and feature selection methods were used by the authors to forecast acute malnutrition and weight gain after the treatment in children. [15]. Some of the predictors identified in the study include feeding frequency, birth weight and parental education. The authors have shown that machine learning models could be used to assist clinical decision-making effectively by approaching children who might be subject to poor treatment results. Nonetheless, the system was not applied to nutritional planning or dietary recommendation, which focused on prediction and not intervention. These results show that machine learning can be valuable in enhancing treatment monitoring and also emphasize the necessity of systems that integrate the detection and dietary assistance. 2.3 Deep Learning and Image-Based Detection of Malnutrition. Recent studies have delved into the deep learning approaches to detecting malnutrition through the use of images. Vidyadevi G. Biradar et al. defined a convolutional neural network-based system with visual malnutrition detection based on transfer learning models (VGG16 and SqueezeNet) applied. [16]. The research attained a rate of about 92 in terms of classification. On the one hand, deep learning methods are characterized by high predictive abilities, but on the other hand, they can be very demanding, in terms of large labeled data, acquisition infrastructure of images, and computer resources. Such needs restrict their use in rural or low-resource medical institutions. Moreover, the image-based systems are more likely to provoke privacy issues and cannot be used to perform a rapid field-level screening. This means that the anthropometric based approaches based on numeric data are more useful in scalable systems of malnutrition detection. 2.4 Child Malnutrition Socio-Economic Determinants. A number of studies have highlighted that the socio-economic and environmental factors have a role to play in child malnutrition besides physical growth indicators. Singh et al. performed a cross-sectional study on the basis of NFHS-5 data to determine the factors that determine stunting, wasting, and underweight among Indian children. [17]. The researchers discovered that mother education, household earnings, sanitation amenities, and maternal body mass index have a significant impact on malnutrition. Another observation made by the authors was that the children in economically weaker households were almost twice as likely to suffer stunting as compared to better socio-economic backgrounds. These types of studies are informative on risk factors of malnutrition; however, they are based on statistical analysis and do not contain automated data detection or intervention models. These results support the rationale of the intelligent systems capable of integrating anthropometric examination and promoting feasible nutritional intervention measures. 2.5 Technology-based Healthcare Systems. The improvement of digital healthcare technologies has resulted in the creation of mobile health (m-health) and Web-based monitoring systems. Kumar et al. suggested an m-health-based child nutrition monitoring system based on mobile applications with machine learning models. [18]. The system enhanced accessibility among rural populations yet it depended on broad dietary recommendations as opposed to localized recommendations. Towards the other end, Tahir et al. proposed a health record management framework based on blockchain, which is used to provide safe and irrevocable storage of medical records. [19]. Although the framework improved the integrity and reliability of data, it failed to introduce predictive analytics and nutritional evaluation.These articles prove the emerging role of online platforms in healthcare and emphasize the absence of combined AI-based models of nutritional decision-support. 2.6 Standard Guidelines of Assessment of nutrition. The international bodies, like the World Health Organization (WHO) and UNICEF, have come up with standardized recommendations on how to measure malnutrition in children using anthropometric measurements like Height-for-Age Z-score (HAZ), Weight-for-Age Z-score (WAZ) and Weight-for-Height Z-score (WHZ). [20]. These indicators are the basis of clinical and survey based nutritional assessment in the world. Though these standards offer good evaluation parameters, they do not give prescriptive automated or AI-based evaluations to mass screening or individualized dietary intercession. A combination of these guidelines and smart systems is likely to play a significant role in the initial detection and decision-making. 2.7 Literature review summary. According to the literature reviewed, machine learning and deep learning methods prove to be useful in identifying child malnutrition based on anthropometric and demographic data. Ensemble models are always seen to have better predictive power and the deep learning methods are very accurate in detecting images. The multi-dimensionality of malnutrition is also brought out by socio-economic studies. Most of the systems available are however targeted at detection or analysis but lack region specific dietary advice. Moreover, a good number of approaches have ethical or privacy or scale issues. 2.8 Research Gap Analysis Based on the literature review, there are a number of research gaps. Current mechanisms focus more on characterizing malnutrition and incorporating dietary treatment in a large scale. Standardized nutrition information does not consider the availability of foods in regions and cultural tastes in food. Moreover, the use of real-world datasets is ethically and privately questionable. In order to overcome these shortcomings, a system based on AI should be developed which combines the precise malnutrition detection with localized dietary advice with the help of ethically generated data. The proposed NutriScan system will address this gap and serve as a solution, specifically, machine learning-based detection and localized nutritional advice, incorporated into a large-scale web system. An easy way to comply with the conference paper formatting requirements is to use this document as a template and simply type your text into it. III. PROPOSED SYSTEM: NUTRISCAN 3.1 System Overview NutriScan is an AI-driven web application that helps to identify child malnutrition at an early stage and recommend a particular diet based on the current region. The system takes machine learning models, coupled with regularized nutritional assessment guidelines, to give precise and acts as an actioner. Compared to the traditional systems that only classify the malnutrition, NutriScan has gone a notch higher by offering appropriate diets depending on the availability of foods in the region. The proposed system will be user-friendly, expandable, and available to the caregivers, the health care workers and nutritionists. It focuses on the use of ethical data, its interpretability, and practicality. 3.2 System Architecture NutriScan architecture is based on a client server architecture, which is divided into three significant parts: Frontend Layer: The frontend, which is developed in React, contains an intuitive interface where users can input child anthropometric information and receive a result. Backend Layer: The backend is implemented in Flask and developers process the data, infer machine-learned algorithms, and delivers data to the frontend and the machine-learned models. Machine Learning Layer: This layer houses trained machine learning models that were created in Python and Scikit-learn, and charged with malnutrition classification and risk score generation. The scalability, flexibility and ease of maintenance are guaranteed by the modular design. 3.3 Input Parameters NutriScan makes use of simple anthropometrics which are simple to obtain within clinical and community context. The following parameters are the input: Age (in months) Height (in centimeters) Weight (in kilograms) These are adequate inputs to calculate standardized nutritional indicators that are prescribed by international health organizations. 3.4 Output Parameters NutriScan provides the following outputs based on the input data: Classification of nutritional status: Normal Stunted Wasted Underweight Risk score showing cases of malnutrition severity. Dietary prescription depending on nutritional deficiency. The outputs are also in an easy to understand format to help in making the decisions. IV. METHODOLOGY The overall methodology of the proposed NutriScan system is illustrated in Fig. 1. 4.1 Dataset Preparation On the ethical and privacy grounds of real child health data, a synthetic dataset was produced in Python. A dataset that was realistic in terms of anthropometric distributions was created as per WHO child growth standards. This would allow adherence to the ethical standards but allow the effective training and evaluation of the models. 4.2 Feature Selection The most dominant factors picked to train the model are age, height and weight. sThe number of features to be selected was kept to a minimum in order to make the system simple and relevant to the real world. 4.3 Machine Learning Models utilized . To analyze the risks and benefits associated with the ageing population, the authors employed machine learning models as discussed below. In this study, two machine learning models were used: Logistic Regression: It is a basic and easy to understand classifier that is used as a baseline classifier. Random Forest: A model that will be an ensemble and will be able to address non-linear relationships and feature interactions and lead to a better prediction accuracy. The standard classification metrics were used to determine the model performance. Fig. 2 shows the comparison between actual and predicted weight using the Linear Regression model. 4.4 Classification Process of Malnutrition . This is a classification process that entails the categorization of malnutrition based on the health status of the affected individual subheading: Malnutrition Classification Process This is a classification process, which involves categorising malnutrition according to the health condition of the affected person. The classification of malnutrition process entails the determination of standardized Z-score using WHO standards. The machine learning models are then fed with these scores to determine nutritional status. The classification logic guarantees the correspondence to the accepted medical standards and the advantage of AI-influenced automation. The classification accuracy achieved by the Random Forest model is presented in Fig. 3. 4.5 Diet Recommendation Module Diet Recommendation Module: The diet recommendation module enables users to enter their data to generate a nutritional plan by considering a variety of factors (such as health, age, and physical activity). The diet recommendation module relies on the Indian Council of medical Research (ICMR) Food Composition Table, which provided suggestions on nutritionally suitable foods. Dietary practices and food availability in the Maharashtra region are taken into account and are tailored to the local recommendations. Such location-focused strategy increases compliance and nutritional recommendations. V. RESULTS The experimental analysis shows that the random forest model has a greater classification accuracy when compared to the logistic regression. The system is effective in classifying the children into normal, stunted, wasted, and underweight. The system generates risk scores which are an added severity assessment. Sample results show that the region-specific diet recommendation module is effective in proposing the correct food items in line with the nutritional deficiencies detected. The NutriScan web dashboard used for entering child anthropometric and demographic details is shown in Fig. 4. The output generated by the proposed NutriScan system for a sample child case is illustrated in Fig. 5. VI. DISCUSSION The findings underscore the usefulness of machine learning combining with the conventional nutritional assessment. NutriScan is better than the existing systems because it integrates the detection and intervention assistance. In contrast to the image-based deep learning methods, the given system is based on the use of simple numeric inputs, so it is applicable in the case of low-resource settings. The aspect of synthetic data use also guarantees the ethical aspect, but restricts the exposure to variability in the real world. However, the system has great potential to actual implementation. VII. CONCLUSION This article described NutriScan, an artificial intelligence web application that allows early child malnutrition and regional nutritional prescription. NutriScan can offer a unified and practical solution to monitoring child health by using machine learning models, WHO growth standards, and national nutritional guidelines. The system is used to fill in key gaps in current solutions as it goes beyond the classification to offer nutritional intervention, which increases practicality in the real world. FUTURE SCOPE The next generation development of NutriScan can involve: · Expansion into additional areas where the dietary databases can be customized. · Real life clinical and government health data integration. · Creation of cell phone application to be more accessible. · Insight of socio-economic and environmental consideration. · Implementation in state health and nutritional services. Abbreviations AI: Artificial Intelligence ML: Mahine Learning WHO: World Health Organization UNICEF United Nations International Children’s Emergency Fund ICMR: Indian Council of Medical Research HAZ: Height-for-Age Z-score WAZ: Weight-for-Age Z-score WHZ: Weight-for-Height Z-score NFHS: National Family Health Survey Declarations Ethics approval and consent to participate Ethics approval was not required for this study as no real human participants were involved. The dataset used in this research was synthetically generated using WHO growth standards to simulate realistic anthropometric distributions. Therefore, no human subject data were collected or analysed. Consent for publication Not applicable. Availability of data and materials The dataset used in this study was synthetically generated in Python based on WHO Child Growth Standards. The synthetic dataset and implementation details are available from the corresponding author upon reasonable request. Competing Interests The authors declare that they have no competing interests. Funding The authors received no specific funding for this study. Authors’ contributions D.S. conceptualized and designed the NutriScan system, developed the methodology, implemented the machine learning models, performed experimental evaluation, and drafted the manuscript. K. T. V. R. supervised the research, provided technical guidance, reviewed the manuscript critically, and approved the final version. Acknowledgements The authors would like to thank the Department of Artificial Intelligence and Data Science, Faculty of Engineering and Technology, for providing the academic support and infrastructure necessary to conduct this research. References World Health Organization. Levels and Trends in Child Malnutrition. Geneva, Switzerland: WHO; 2020. UNICEF, World Health Organization, and World Bank Group. Joint Child Malnutrition Estimates, 2021. World Health Organization. WHO Child Growth Standards: Methods and Development. Geneva, Switzerland: WHO; 2006. International Institute for Population Sciences (IIPS). and ICF, National Family Health Survey (NFHS-5), India , Mumbai, India, 2021. Rao B, Kulkarni S, Patil R. Machine learning-based prediction of child malnutrition: A meta-analysis. J Biomed Inform. 2025;132:104–18. Das M, Rahman M. Predicting child malnutrition using machine learning techniques. Health Inf J. 2022;28(2):1–15. Sánchez-Martínez S, Pérez J, Gómez L. Machine learning for predicting acute malnutrition and treatment outcomes. BMC Nutr. 2024;10(3):45–58. Biradar VG, Patil A, Desai S. Deep learning-based malnutrition detection using transfer learning. IEEE Access. 2024;12:45678–90. Singh R, Verma P, Kumar A. Socio-economic determinants of child malnutrition in India: Evidence from NFHS-5. Public Health Nutr. 2023;26(1):89–101. Indian Council of Medical Research. Indian Food Composition Tables. Hyderabad, India: National Institute of Nutrition; 2017. Kumar A, Mehta S, Sharma R. Mobile health system for monitoring child nutrition using machine learning. Int J Med Informatics. 2021;150:104–12. Tahir M, Khan F, Ali S. Blockchain-based secure healthcare data management system. IEEE Trans Eng Manage. 2024;69(4):2101–12. Breiman L. Random forests. Mach Learn. 2001;45(1):5–32. Hastie T, Tibshirani R, Friedman J. The Elements of Statistical Learning. 2nd ed. New York, NY, USA: Springer; 2009. Han J, Kamber M, Pei J. Data Mining: Concepts and Techniques. 3rd ed. San Francisco, CA, USA: Morgan Kaufmann; 2012. Obermeyer Z, Emanuel EJ. Predicting the future—big data, machine learning, and clinical medicine. N Engl J Med. 2016;375(13):1216–9. Topol E. Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again. New York, NY, USA: Basic Books; 2019. Esteva A, et al. A guide to deep learning in healthcare. Nat Med. 2019;25:24–9. Kaur P, Gupta S. Artificial intelligence in healthcare: Applications and challenges. Int J Eng Res. 2022;10(5):45–52. Kumar R, Singh S. Web-based decision support system for healthcare applications. Int J Comput Appl. 2020;176(25):12–8. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 07 Apr, 2026 Reviews received at journal 27 Mar, 2026 Reviews received at journal 21 Mar, 2026 Reviews received at journal 16 Mar, 2026 Reviewers agreed at journal 16 Mar, 2026 Reviewers agreed at journal 14 Mar, 2026 Reviewers agreed at journal 12 Mar, 2026 Reviewers invited by journal 11 Mar, 2026 Editor assigned by journal 09 Mar, 2026 Editor invited by journal 19 Feb, 2026 Submission checks completed at journal 19 Feb, 2026 First submitted to journal 19 Feb, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8744190","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":604839132,"identity":"daee450b-4559-4a8f-8d1d-c48276176191","order_by":0,"name":"Disha Shirbad","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9ElEQVRIiWNgGAWjYDCCwwxsQPIAD4MEA4PEByCTjZ0ULZIzQFqYCWk5ANHCANIizQMSIaSF7zj7swc//tyRMZ/d/PC2za9t8nzMDIwfPubg1iJ5mMfcsLftGY/MnWPG1rl9tw3bmBmYJWduw63F4DAPmwRvw2EeCYkEM+ncntuMQC1szLx4tbA/k/zzB6Ql/Zu0Zc9teyK0MJhJ87CBtOSYSTP8uJ1IUAvQL2bSsm1ALTJnii17G24ntzEzNuP1C9/5488k3/w5bC8h3b7xxo8/t23ntzcf/PARjxZUwNgGJhuIVQ8Cf0hRPApGwSgYBSMFAACUF08qyk18qAAAAABJRU5ErkJggg==","orcid":"","institution":"Datta Meghe Institute of Medical Sciences","correspondingAuthor":true,"prefix":"","firstName":"Disha","middleName":"","lastName":"Shirbad","suffix":""}],"badges":[],"createdAt":"2026-01-30 18:38:29","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8744190/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8744190/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104753630,"identity":"45c0c653-2e29-4dce-92ed-33e9018f92c3","added_by":"auto","created_at":"2026-03-16 21:02:50","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":80852,"visible":true,"origin":"","legend":"\u003cp\u003eMethodology of the Proposed NutriScan System\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8744190/v1/e009ced626e6c215211d03d0.jpg"},{"id":104783078,"identity":"57ee3437-8b01-41ee-91ef-872a05ca1211","added_by":"auto","created_at":"2026-03-17 07:58:11","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":57864,"visible":true,"origin":"","legend":"\u003cp\u003eActual vs. Predicted Weight using Linear Regression Model\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8744190/v1/f938110b7958172173e0d575.jpg"},{"id":104753633,"identity":"c2b13591-3aa6-4d62-b34a-6468458c5436","added_by":"auto","created_at":"2026-03-16 21:02:51","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":44009,"visible":true,"origin":"","legend":"\u003cp\u003eClassification Accuracy of Random Forest Model for Malnutrition Indicators\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8744190/v1/938279fefbbb910788b0d983.jpg"},{"id":104753631,"identity":"5be6a731-370a-497a-aa26-d00d341e9ac5","added_by":"auto","created_at":"2026-03-16 21:02:50","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":91568,"visible":true,"origin":"","legend":"\u003cp\u003eNutriScan Web Dashboard for Child Data Entry\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8744190/v1/687e5efcb5a6af3bef1a8ee9.jpg"},{"id":104783014,"identity":"d3f14985-5965-4954-abf9-89dee179e8f2","added_by":"auto","created_at":"2026-03-17 07:58:06","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":66073,"visible":true,"origin":"","legend":"\u003cp\u003eSample Output Showing Malnutrition Classification, Risk Score, and Dietary Recommendation\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8744190/v1/902ab2bbc77422db45959275.jpg"},{"id":104785012,"identity":"e57f910b-55ba-4c32-a57b-3f61ca2ee778","added_by":"auto","created_at":"2026-03-17 08:09:14","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1081395,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8744190/v1/0a9f67b4-cd2d-4cdd-a655-8c249a1fdb2a.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"NutriScan: An AI-Driven System for Child Malnutrition Detection and Region-Aware Dietary Recommendation","fulltext":[{"header":"I.\tINTRODUCTION","content":"\u003cp\u003e\u003cstrong\u003e1.1\u0026nbsp;\u0026nbsp;Background of Child Malnutrition\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eChild malnutrition remains as one of the most urgent issues in the global public health, especially in low- and middle-income countries. Reports show that malnutrition contributes to the morbidity and mortality of children below the age of five according to reports by international health organizations. Stunting, wasting, and underweight are relatively common conditions that severely impact physical development, cognitive, immunological, and future productivity levels [1], [2].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eChild malnutrition has been a thorn in the flesh in India even with a series of government programs and nutrition plans. Such surveys as the National Family Health Survey (NFHS) show that a significant number of children experience chronic and acute malnutrition and that they should be better screened and the intervention mechanisms should be improved [3]. Insufficient dietary consumption is not the only cause of malnutrition, which is also caused by socio-economic factors, sanitation, maternal education, and access to health care [4]. The need to identify malnutrition early before it is too late is necessary to ensure that there is no irreversible loss of vitality in the young during a crucial time of child development. However, traditional assessment tools tend to be manual, time consuming, and require a trained health care practitioner, which constrain their resource use in resource limited settings.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.2\u0026nbsp;\u0026nbsp;Artificial Intelligence and its role in healthcare.\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe latest development of the Artificial Intelligence (AI) and Machine Learning (ML) has revolutionized several fields of healthcare, such as disease diagnostics, risk forecasting, and clinical decision support. The use of AI-based systems is able to analyze health information automatically, which can uncover diseases that otherwise cannot be detected [5], [6].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMachine learning models have been specifically useful in uncovering non-linear and complicated relationships in healthcare data. Within the framework of child malnutrition, the ML methods have been effectively used in predicting nutritional status by the analysis of anthropometric and demographic data [7]. Random Forest and other types of the ensemble models have been shown to be more effective because of their strength and capability to interact among features [8].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eRegardless of these benefits, most current AI-based health solutions pay more attention to the accuracy of prediction and do not consider practical intervention plans. In the case of malnutrition management, prediction cannot be conducted without actionable guidance that can guide caregivers and healthcare workers to help in managing this condition.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1\u003cstrong\u003e.3 Motivation for the Study\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe impetus to base this research on the fact that there is a gap between diagnosis and nutritional intervention of malnutrition. Although a number of studies have revealed the efficacy of machine learning in forecasting malnutrition, majority of the systems conclude with classification and fail to offer information on how to treat the identified problem [9].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMoreover, the diet advice tends to be generalized and does not reflect on the food supply in the region, the cultural eating lifestyles, and the national nutritional policies. Dietary planning based on regions is important in a diverse nation such as India, where the nutrition interventions must be adhered to and feasible [10].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe rising accessibility of web-based technologies has presented a possibility to create scalable and accessible healthcare solutions. The malnutrition monitoring systems can be improved in terms of their applicability in the real world with the integration of AI-based detection and region-specific dietary recommendations. The necessity of such an integrated, intelligent, and easy to use solution is the reason behind this study.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.4\u0026nbsp;\u0026nbsp;Problem Statement\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDespite the fact that a number of machine learning-based systems are suggested with the purpose of child malnutrition detection, several limitations still exist. The current strategies are all concerned with predictive performance and do not consider the post-diagnosis intervention mechanisms [11]. Also, there are numerous systems which are based on large-scale real-world data that presents ethical, privacy, and access issues.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOne more significant weakness is the lack of localized dietary recommendations. The generalized nutrition information fails to cover the local eating habits or the availability of local foodstuffs, which diminishes the impact of the intervention measures [12]. Hence, it is required that an AI-based system be able to not only detect child malnutrition properly, but also issue actionable and region-specific dietary recommendations in an ethical and scalable way.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.\u003cstrong\u003e5 Objectives of the Study\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e· The main aim of the study is as follows:\u003c/p\u003e\n\u003cp\u003e· The purpose of the research is to design and develop an AI-based system to early detect malnutrition in children. To categorise children to nutritional groups, which include normal, stunted, wasted, and underweight.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e· To produce a risk score of the degree of malnutrition.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e· To give localized nutritional advice using local nutritional guidelines.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e· To create a web-based platform which is accessible, scalable, and user friendly.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e· These goals shall help fill the gap in recognizing malnutrition and providing nutritional intervention through clever decision-making processes.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1.6 Organization of the Paper\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe rest of this paper is structured in the following way. Section 2 contains an extensive literature review of the existing solutions on the topic of child malnutrition detection and healthcare technologies. Section 3 explains the proposed NutriScan system and the general architecture. Section 4 gives out the methodology which includes the preparation of the datasets, machine learning models, and diet recommendation strategy. Section 5 deals with the performance evaluation and results of the experiment. Findings are discussed in detail in section 6. Lastly, Section 7 concludes the paper and gives future research directions.\u003c/p\u003e"},{"header":"II.\tLITERATURE REVIEW","content":"\u003cp\u003eThe study will involve machine learning methods to identify child malnutrition.\u003c/p\u003e\n\u003cp\u003eA number of researchers have conducted studies to assess the use of machine learning to predict child malnutrition with the help of anthropometric and demographic information. The objectives of these methods are to enhance the early diagnosis through automating health indicators like age, height and weight.\u003c/p\u003e\n\u003cp\u003eIn their study, Bhagya Jyothi Rao and colleagues conducted a meta-analysis of the performances of different machine learning models to predict child malnutrition on their demographic and health surveys data. [13].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe paper established a comparison with the Logistic Regression, Random Forest and Support Vector machine models and indicated that the ensemble-based approaches yielded an accuracy of about 88. The authors emphasized the fact that tree-based models are good because they are capable of non-linear relationships between features. Nevertheless, their study was more concerned with the predictive accuracy and has not considered the dietary intervention mechanisms.\u003c/p\u003e\n\u003cp\u003eIn the same way, Das and Rahman used machine learning classifiers to establish the malnutrition status by anthropometric indicators and household survey data.[14]. Their findings showed that the random forest models are stronger than the linear classifiers in the classification task. Though it was more accurate, the system did not have customized or area specific nutritional suggestions and thus, could not be utilized practically.\u003c/p\u003e\n\u003cp\u003eThese works show the usefulness of machine learning in the detection of malnutrition and also indicate a persistent weakness, namely most methods conclude with classification and do not include any actionable healthcare recommendation.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"98%\" class=\"fr-table-selection-hover\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eSr. No.\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eTitle, Authors \u0026amp; Year\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eObjective \u0026amp; Method\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eKey Findings\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eRemarks\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eMachine Learning in Predicting Child Malnutrition \u0026ndash; Rao et al. (2025)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eMeta-analysis using Logistic Regression, Random Forest, SVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAchieved ~88% accuracy in classification\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eEnsemble models perform best\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eML for Acute Child Malnutrition \u0026ndash; S\u0026aacute;nchez-Mart\u0026iacute;nez et al. (2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRandom Forest with feature selection\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eIdentified feeding frequency, birth weight as key factors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eHelps optimize treatment\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDeep Learning-Based Malnutrition Detection \u0026ndash; Biradar et al. (2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCNN with transfer learning\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAchieved 92% accuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRequires image data\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eBlockchain-Based Healthcare Framework \u0026ndash; Tahir et al. (2024)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSecure EHR using blockchain\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eImproved data integrity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNo prediction module\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDeterminants of Child Malnutrition in India \u0026ndash; Singh et al. (2023)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLogistic regression using NFHS-5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSocio-economic factors significant\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLacks automation\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003e2.2 Machine Learning of Acute Malnutrition and Treatment Results.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn addition to the detection, a few researchers have also concentrated on predicting acute malnutrition and treatment outcomes based on machine learning. SANTE-only models and feature selection methods were used by the authors to forecast acute malnutrition and weight gain after the treatment in children.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e[15]. Some of the predictors identified in the study include feeding frequency, birth weight and parental education.\u003c/p\u003e\n\u003cp\u003eThe authors have shown that machine learning models could be used to assist clinical decision-making effectively by approaching children who might be subject to poor treatment results. Nonetheless, the system was not applied to nutritional planning or dietary recommendation, which focused on prediction and not intervention.\u003c/p\u003e\n\u003cp\u003eThese results show that machine learning can be valuable in enhancing treatment monitoring and also emphasize the necessity of systems that integrate the detection and dietary assistance.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3 Deep Learning and Image-Based Detection of Malnutrition.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRecent studies have delved into the deep learning approaches to detecting malnutrition through the use of images. Vidyadevi G. Biradar et al. defined a convolutional neural network-based system with visual malnutrition detection based on transfer learning models (VGG16 and SqueezeNet) applied.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e[16]. The research attained a rate of about 92 in terms of classification.\u003c/p\u003e\n\u003cp\u003eOn the one hand, deep learning methods are characterized by high predictive abilities, but on the other hand, they can be very demanding, in terms of large labeled data, acquisition infrastructure of images, and computer resources. Such needs restrict their use in rural or low-resource medical institutions. Moreover, the image-based systems are more likely to provoke privacy issues and cannot be used to perform a rapid field-level screening.\u003c/p\u003e\n\u003cp\u003eThis means that the anthropometric based approaches based on numeric data are more useful in scalable systems of malnutrition detection.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.4 Child Malnutrition Socio-Economic Determinants.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA number of studies have highlighted that the socio-economic and environmental factors have a role to play in child malnutrition besides physical growth indicators. Singh et al. performed a cross-sectional study on the basis of NFHS-5 data to determine the factors that determine stunting, wasting, and underweight among Indian children. [17]. The researchers discovered that mother education, household earnings, sanitation amenities, and maternal body mass index have a significant impact on malnutrition.\u003c/p\u003e\n\u003cp\u003eAnother observation made by the authors was that the children in economically weaker households were almost twice as likely to suffer stunting as compared to better socio-economic backgrounds. These types of studies are informative on risk factors of malnutrition; however, they are based on statistical analysis and do not contain automated data detection or intervention models.\u003c/p\u003e\n\u003cp\u003eThese results support the rationale of the intelligent systems capable of integrating anthropometric examination and promoting feasible nutritional intervention measures.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.5 Technology-based Healthcare Systems.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe improvement of digital healthcare technologies has resulted in the creation of mobile health (m-health) and Web-based monitoring systems. Kumar et al. suggested an m-health-based child nutrition monitoring system based on mobile applications with machine learning models.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e[18]. The system enhanced accessibility among rural populations yet it depended on broad dietary recommendations as opposed to localized recommendations.\u003c/p\u003e\n\u003cp\u003eTowards the other end, Tahir et al. proposed a health record management framework based on blockchain, which is used to provide safe and irrevocable storage of medical records.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e[19]. Although the framework improved the integrity and reliability of data, it failed to introduce predictive analytics and nutritional evaluation.These articles prove the emerging role of online platforms in healthcare and emphasize the absence of combined AI-based models of nutritional decision-support.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.6 Standard Guidelines of Assessment of nutrition.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe international bodies, like the World Health Organization (WHO) and UNICEF, have come up with standardized recommendations on how to measure malnutrition in children using anthropometric measurements like Height-for-Age Z-score (HAZ), Weight-for-Age Z-score (WAZ) and Weight-for-Height Z-score (WHZ).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e[20]. These indicators are the basis of clinical and survey based nutritional assessment in the world.\u003c/p\u003e\n\u003cp\u003eThough these standards offer good evaluation parameters, they do not give prescriptive automated or AI-based evaluations to mass screening or individualized dietary intercession. A combination of these guidelines and smart systems is likely to play a significant role in the initial detection and decision-making.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.7 Literature review summary.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAccording to the literature reviewed, machine learning and deep learning methods prove to be useful in identifying child malnutrition based on anthropometric and demographic data. Ensemble models are always seen to have better predictive power and the deep learning methods are very accurate in detecting images. The multi-dimensionality of malnutrition is also brought out by socio-economic studies.\u003c/p\u003e\n\u003cp\u003eMost of the systems available are however targeted at detection or analysis but lack region specific dietary advice. Moreover, a good number of approaches have ethical or privacy or scale issues.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.8 Research Gap Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBased on the literature review, there are a number of research gaps. Current mechanisms focus more on characterizing malnutrition and incorporating dietary treatment in a large scale. Standardized nutrition information does not consider the availability of foods in regions and cultural tastes in food. Moreover, the use of real-world datasets is ethically and privately questionable.\u003c/p\u003e\n\u003cp\u003eIn order to overcome these shortcomings, a system based on AI should be developed which combines the precise malnutrition detection with localized dietary advice with the help of ethically generated data. The proposed NutriScan system will address this gap and serve as a solution, specifically, machine learning-based detection and localized nutritional advice, incorporated into a large-scale web system.\u003c/p\u003e\n\u003cp\u003eAn easy way to comply with the conference paper formatting requirements is to use this document as a template and simply type your text into it.\u003c/p\u003e"},{"header":"III.\tPROPOSED SYSTEM: NUTRISCAN","content":"\u003cp\u003e\u003cstrong\u003e3.1 System Overview\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNutriScan is an AI-driven web application that helps to identify child malnutrition at an early stage and recommend a particular diet based on the current region. The system takes machine learning models, coupled with regularized nutritional assessment guidelines, to give precise and acts as an actioner. Compared to the traditional systems that only classify the malnutrition, NutriScan has gone a notch higher by offering appropriate diets depending on the availability of foods in the region.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe proposed system will be user-friendly, expandable, and available to the caregivers, the health care workers and nutritionists. It focuses on the use of ethical data, its interpretability, and practicality.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2 System Architecture\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNutriScan architecture is based on a client server architecture, which is divided into three significant parts:\u003c/p\u003e\n\u003cp\u003eFrontend Layer: The frontend, which is developed in React, contains an intuitive interface where users can input child anthropometric information and receive a result.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eBackend Layer: The backend is implemented in Flask and developers process the data, infer machine-learned algorithms, and delivers data to the frontend and the machine-learned models.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMachine Learning Layer: This layer houses trained machine learning models that were created in Python and Scikit-learn, and charged with malnutrition classification and risk score generation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe scalability, flexibility and ease of maintenance are guaranteed by the modular design.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.3 Input Parameters\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNutriScan makes use of simple anthropometrics which are simple to obtain within clinical and community context. The following parameters are the input:\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAge (in months)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHeight (in centimeters)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWeight (in kilograms)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThese are adequate inputs to calculate standardized nutritional indicators that are prescribed by international health organizations.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.4 Output Parameters\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNutriScan provides the following outputs based on the input data:\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eClassification of nutritional status:\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNormal\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eStunted\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWasted\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eUnderweight\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eRisk score showing cases of malnutrition severity. Dietary prescription depending on nutritional deficiency.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe outputs are also in an easy to understand format to help in making the decisions.\u003c/p\u003e"},{"header":"IV.\tMETHODOLOGY","content":"\u003cp\u003eThe overall methodology of the proposed NutriScan system is illustrated in Fig. 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.1 Dataset Preparation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOn the ethical and privacy grounds of real child health data, a synthetic dataset was produced in Python. A dataset that was realistic in terms of anthropometric distributions was created as per WHO child growth standards. This would allow adherence to the ethical standards but allow the effective training and evaluation of the models.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.2 Feature Selection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe most dominant factors picked to train the model are age, height and weight. sThe number of features to be selected was kept to a minimum in order to make the system simple and relevant to the real world.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.3 Machine Learning Models utilized\u003c/strong\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo analyze the risks and benefits associated with the ageing population, the authors employed machine learning models as discussed below.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn this study, two machine learning models were used:\u003c/p\u003e\n\u003cp\u003eLogistic Regression: It is a basic and easy to understand classifier that is used as a baseline classifier.\u003c/p\u003e\n\u003cp\u003eRandom Forest: A model that will be an ensemble and will be able to address non-linear relationships and feature interactions and lead to a better prediction accuracy.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe standard classification metrics were used to determine the model performance.\u003c/p\u003e\n\u003cp\u003eFig. 2 shows the comparison between actual and predicted weight using the Linear Regression model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.4 Classification Process of Malnutrition\u003c/strong\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis is a classification process that entails the categorization of malnutrition based on the health status of the affected individual subheading: Malnutrition Classification Process This is a classification process, which involves categorising malnutrition according to the health condition of the affected person.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe classification of malnutrition process entails the determination of standardized Z-score using WHO standards. The machine learning models are then fed with these scores to determine nutritional status. The classification logic guarantees the correspondence to the accepted medical standards and the advantage of AI-influenced automation.\u003c/p\u003e\n\u003cp\u003eThe classification accuracy achieved by the Random Forest model is presented in Fig. 3.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e4.5 Diet Recommendation Module\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDiet Recommendation Module: The diet recommendation module enables users to enter their data to generate a nutritional plan by considering a variety of factors (such as health, age, and physical activity).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe diet recommendation module relies on the Indian Council of medical Research (ICMR) Food Composition Table, which provided suggestions on nutritionally suitable foods. Dietary practices and food availability in the Maharashtra region are taken into account and are tailored to the local recommendations. Such location-focused strategy increases compliance and nutritional recommendations.\u003c/p\u003e"},{"header":"V.\tRESULTS","content":"\u003cp\u003eThe experimental analysis shows that the random forest model has a greater classification accuracy when compared to the logistic regression. The system is effective in classifying the children into normal, stunted, wasted, and underweight. The system generates risk scores which are an added severity assessment. Sample results show that the region-specific diet recommendation module is effective in proposing the correct food items in line with the nutritional deficiencies detected.\u003c/p\u003e\n\u003cp\u003eThe NutriScan web dashboard used for entering child anthropometric and demographic details is shown in Fig. 4.\u003c/p\u003e\n\u003cp\u003eThe output generated by the proposed NutriScan system for a sample child case is illustrated in Fig. 5.\u003c/p\u003e"},{"header":"VI. DISCUSSION","content":"\u003cp\u003eThe findings underscore the usefulness of machine learning combining with the conventional nutritional assessment. NutriScan is better than the existing systems because it integrates the detection and intervention assistance. In contrast to the image-based deep learning methods, the given system is based on the use of simple numeric inputs, so it is applicable in the case of low-resource settings.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe aspect of synthetic data use also guarantees the ethical aspect, but restricts the exposure to variability in the real world. However, the system has great potential to actual implementation.\u003c/p\u003e"},{"header":"VII. CONCLUSION","content":"\u003cp\u003eThis article described NutriScan, an artificial intelligence web application that allows early child malnutrition and regional nutritional prescription. NutriScan can offer a unified and practical solution to monitoring child health by using machine learning models, WHO growth standards, and national nutritional guidelines.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe system is used to fill in key gaps in current solutions as it goes beyond the classification to offer nutritional intervention, which increases practicality in the real world.\u003c/p\u003e\n\u003ch3\u003eFUTURE SCOPE\u003c/h3\u003e\n\u003cp\u003eThe next generation development of NutriScan can involve:\u0026nbsp;\u003c/p\u003e\u003cp\u003e· Expansion into additional areas where the dietary databases can be customized.\u0026nbsp;\u003c/p\u003e\u003cp\u003e· Real life clinical and government health data integration.\u0026nbsp;\u003c/p\u003e\u003cp\u003e· Creation of cell phone application to be more accessible.\u0026nbsp;\u003c/p\u003e\u003cp\u003e· Insight of socio-economic and environmental consideration.\u0026nbsp;\u003c/p\u003e\u003cp\u003e· Implementation in state health and nutritional services.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eAI: Artificial Intelligence\u003c/p\u003e\n\u003cp\u003eML: Mahine Learning\u003c/p\u003e\n\u003cp\u003eWHO: World Health Organization\u003c/p\u003e\n\u003cp\u003eUNICEF United Nations International Children\u0026rsquo;s Emergency Fund\u003c/p\u003e\n\u003cp\u003eICMR: Indian Council of Medical Research\u003c/p\u003e\n\u003cp\u003eHAZ: Height-for-Age Z-score\u003c/p\u003e\n\u003cp\u003eWAZ: Weight-for-Age Z-score\u003c/p\u003e\n\u003cp\u003eWHZ: Weight-for-Height Z-score \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNFHS: National Family Health Survey\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEthics approval was not required for this study as no real human participants were involved. The dataset used in this research was synthetically generated using WHO growth standards to simulate realistic anthropometric distributions. Therefore, no human subject data were collected or analysed.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe dataset used in this study was synthetically generated in Python based on WHO Child Growth Standards. The synthetic dataset and implementation details are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors received no specific funding for this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors’ contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eD.S. conceptualized and designed the NutriScan system, developed the methodology, implemented the machine learning models, performed experimental evaluation, and drafted the manuscript. K. T. V. R. supervised the research, provided technical guidance, reviewed the manuscript critically, and approved the final version.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors would like to thank the Department of Artificial Intelligence and Data Science, Faculty of Engineering and Technology, for providing the academic support and infrastructure necessary to conduct this research.\u003c/p\u003e\n\n\n\n\n\n\n"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eWorld Health Organization. Levels and Trends in Child Malnutrition. Geneva, Switzerland: WHO; 2020.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUNICEF, World Health Organization, and World Bank Group. Joint Child Malnutrition Estimates, 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization. WHO Child Growth Standards: Methods and Development. Geneva, Switzerland: WHO; 2006.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eInternational Institute for Population Sciences (IIPS). and ICF, \u003cem\u003eNational Family Health Survey (NFHS-5), India\u003c/em\u003e, Mumbai, India, 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRao B, Kulkarni S, Patil R. Machine learning-based prediction of child malnutrition: A meta-analysis. J Biomed Inform. 2025;132:104\u0026ndash;18.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDas M, Rahman M. Predicting child malnutrition using machine learning techniques. Health Inf J. 2022;28(2):1\u0026ndash;15.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS\u0026aacute;nchez-Mart\u0026iacute;nez S, P\u0026eacute;rez J, G\u0026oacute;mez L. Machine learning for predicting acute malnutrition and treatment outcomes. BMC Nutr. 2024;10(3):45\u0026ndash;58.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBiradar VG, Patil A, Desai S. Deep learning-based malnutrition detection using transfer learning. IEEE Access. 2024;12:45678\u0026ndash;90.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSingh R, Verma P, Kumar A. Socio-economic determinants of child malnutrition in India: Evidence from NFHS-5. Public Health Nutr. 2023;26(1):89\u0026ndash;101.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIndian Council of Medical Research. Indian Food Composition Tables. Hyderabad, India: National Institute of Nutrition; 2017.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar A, Mehta S, Sharma R. Mobile health system for monitoring child nutrition using machine learning. Int J Med Informatics. 2021;150:104\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTahir M, Khan F, Ali S. Blockchain-based secure healthcare data management system. IEEE Trans Eng Manage. 2024;69(4):2101\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreiman L. Random forests. Mach Learn. 2001;45(1):5\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHastie T, Tibshirani R, Friedman J. The Elements of Statistical Learning. 2nd ed. New York, NY, USA: Springer; 2009.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHan J, Kamber M, Pei J. Data Mining: Concepts and Techniques. 3rd ed. San Francisco, CA, USA: Morgan Kaufmann; 2012.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eObermeyer Z, Emanuel EJ. Predicting the future\u0026mdash;big data, machine learning, and clinical medicine. N Engl J Med. 2016;375(13):1216\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTopol E. Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again. New York, NY, USA: Basic Books; 2019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEsteva A, et al. A guide to deep learning in healthcare. Nat Med. 2019;25:24\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaur P, Gupta S. Artificial intelligence in healthcare: Applications and challenges. Int J Eng Res. 2022;10(5):45\u0026ndash;52.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar R, Singh S. Web-based decision support system for healthcare applications. Int J Comput Appl. 2020;176(25):12\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Child Malnutrition, Artificial Intelligence, Machine Learning, Anthropometric Indicators, WHO Growth Standards, Risk Assessment, Region-Specific Diet Recommendation, Web-Based Healthcare System","lastPublishedDoi":"10.21203/rs.3.rs-8744190/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8744190/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eChild malnutrition has been a severe issue of population health especially in third world countries where prompt diagnosis and access to nutritional intervention result in chronic physical and intellectual disability. Traditional methods of malnutrition evaluation are based on manual anthropometric measurements and do not in most cases give effective dietary advice. In this paper, the author is introducing the AI-based web system called NutriScan, which will allow early detection of malnutrition in children and dietary recommendations depending on their location. The suggested system involves the straightforward anthropometric data of age, height, and weight that are incorporated to calculate the standardized nutritional indicators based on the World Health Organization (WHO) growth standards. The models used are machine learning models (Logistic Regression and random forest) which categorize children into normal, stunted, wasted, and underweight and yield a risk score that is used to show severity. To overcome ethical and privacy limitations of real world child health data, a synthetic data set was created in Python without much distortion of realistic anthropometric distributions. Besides automated malnutrition data, NutriScan has a diet recommendation module which uses the Indian Council of Medical Research (ICMR) Food Composition Table to suggest region-specific diets in Maharashtra. The experimental analysis shows that ensemble-based models present consistent performance in classification. NutriScan promotes an effective application of nutritional intervention as malnutrition identification is combined with localized dietary advice.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e","manuscriptTitle":"NutriScan: An AI-Driven System for Child Malnutrition Detection and Region-Aware Dietary Recommendation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-16 21:02:45","doi":"10.21203/rs.3.rs-8744190/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"263967656523936581947896996435451097555","date":"2026-04-07T09:31:06+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-27T05:13:19+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-21T08:32:30+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-16T09:52:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"301395919013456282252683999344348196003","date":"2026-03-16T07:24:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"211257911471472455361821877634538887806","date":"2026-03-14T08:35:09+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"12999350298097117153032334391002591584","date":"2026-03-12T06:00:11+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-03-11T06:22:34+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-09T11:09:46+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-02-19T09:46:44+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-02-19T06:12:49+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Informatics and Decision Making","date":"2026-02-19T06:08:04+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"9e5e01ff-12c7-43fa-80e2-67931fa31c8c","owner":[],"postedDate":"March 16th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-03-16T21:02:46+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-16 21:02:45","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8744190","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8744190","identity":"rs-8744190","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}