Her Health: An AI-Based Multilingual Symptom Assessment System for Early Screening of Women's Health Disorders

International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.18042955 Original Article @2025 RS Publication, [email protected] 251 Her Health: An AI-Based Multilingual Symptom Assessment System for Early Screening of Women’s Health Disorders Dharshini R Dept of Computer Science and Engineering, East Point College of Engineering and Technology, Bangalore, India Indu Shree S C Dept of Computer Science and Engineering, East Point College of Engineering and Technology, Bangalore, India Keerthana V Dept of Computer Science and Engineering East Point College of Engineering and Technology, Bangalore, India Kilari Bhavana Dept of Computer Science and Engineering, East Point College of Engineering and Technology, Bangalore, India Mrs. Divya U H Dept of Computer Science and Engineering, East Point College of Engineering and Technology, Bangalore, India International Journal of Computer Application https://rspublication.com/ijca/ijca_index.htm ISSN 2250-1797 ARTICLE INFO ABSTRACT ©2025 RS Publication Paper ID: IJCA694B55C52785A Published: 2025-12-24 DOI: https://dx.doi.org/ 10.5281/zenodo.1804 2955 Page No: 251-257 Limited healthcare awareness and low literacy levels make it difficult for women to identify and manage menstrual disorders such as PCOS, PCOD, and Breast Cancer. This paper presents an AIpowered mobile application that uses a chatbot to collect symptoms, classify potential health risks, and provide personalized recommendations. Multilingual support, voice-based input, text-tospeech functionality, and SMS notifications ensure accessibility for low-literacy users. A simple icon-based interface, emergency contact support, and health awareness content further enhance usability. Strong data security measures protect user information, making the system a reliable tool for early screening and effective management of women’s health conditions. Keywords— Women’s health, PCOS, PCOD, Breast Cancer, AI chatbot, Multilingual interface, Machine Learning, Logistic Regression, Random Forest, Mobile Health (mHealth) Cite This Paper: Dharshini R., Indu Shree S. C., Keerthana V.,Kilar I. Bhavana and Mrs. Divya U. H. (2025). "Her Health: An AI-Based Multilingual Symptom Assessment System for Early Screening of Women’s Health Disorders". INTERNATIONAL JOURNAL OF COMPUTER APPLICATION (IJCA), vol. 15, no. 6, 2025, pp. 251-257. DOI: https://dx.doi.org/10.5281/zenodo.18042955 International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.18042955 Original Article @2025 RS Publication, [email protected] 252 I. I NTRODUCTION Women often face difficulties in understanding and managing menstrual and reproductive health due to limited awareness, cultural taboos, and hesitation to discuss sensitive issues. Conditions such as PCOS, PCOD, and Breast Cancer frequently remain undetected because of delayed medical guidance and irregular health screenings. Lack of accurate information and fear of social judgment further contribute to postponed treatment, leading to preventable complications. Access to reliable healthcare tools remains a major challenge, particularly for women in rural and economically disadvantaged regions. Many existing solutions require high literacy levels, continuous internet access, or frequent medical consultations, which are not feasible for all users. Limited digital literacy also restricts the use of text-heavy and complex applications, forcing women to depend on incomplete information or traditional beliefs. To address these challenges, the proposed system, Her Health: AI-Powered Health Companion, provides a simple and accessible mobile application that supports symptom analysis and early guidance. The system integrates artificial intelligence, multilingual support, voice interaction, and personalized recommendations to assist users without requiring medical expertise. By combining symptom assessment, health education, reminders, and accessible communication, the application enables women with low literacy or limited healthcare access to receive meaningful and timely health support. LITERATURE REVIEW Several research works have explored the use of artificial intelligence, chatbots, and machine learning techniques to support women’s reproductive health. The following studies are most relevant to the proposed Her Health system. In the paper [1], Noureen et al. proposed an AI-based chatbot named Sahacharza to assist Bengali-speaking women with reproductive health disorders. The system employs natural language processing techniques, keyword matching, disease categorization, and deep learning models such as LSTM to analyze symptoms and respond to user queries. An unsupervised spelling correction technique using Word2Vec was incorporated to improve interaction accuracy. Limitation: The system supports only a single language, and its performance is highly dependent on the quality of user input and the limited size of the training dataset. In the paper [2], Ajil et al. presented a supervised machine learning approach for the automated detection of Polycystic Ovary Syndrome (PCOS) using clinical, hormonal, and lifestyle-related features. Multiple algorithms including Decision Tree, Logistic Regression, Random Forest, AdaBoost, and XGBoost were evaluated, with XGBoost achieving the highest accuracy. Limitation: The model’s performance is datasetdependent, computationally intensive, and may suffer from overfitting when applied to diverse populations. In the paper [3], Patel et al. proposed a holistic telemedicine system that integrates machine learning models for PCOS and PCOD detection, with a focus on improving healthcare accessibility for women in rural areas. The system demonstrated effective classification performance using supervised learning techniques and emphasized the importance of AI-driven screening in reducing healthcare disparities. Limitation: The system relies on continuous internet connectivity and primarily focuses on PCOS and PCOD, without incorporating screening for other conditions such as breast health or emergency support features. The analysis of these studies highlights the need for an integrated, multilingual, and accessible AI-based health system that combines chatbot interaction, machine learning classification, and support features such as voice interaction and alerts. These requirements are addressed by the proposed Her Health application. PROPOSED SYSTEM The proposed system presents an AI-powered women’s health companion designed to support early screening and awareness of PCOS, PCOD, and breast-related abnormalities. The solution integrates a multilingual chatbot, machine learning–based symptom analysis, voice-enabled interaction, and SMS notifications into a single mobile application. The system emphasizes accessibility by providing a simple interface supported by speech technologies and health education content, making it suitable for users with limited literacy. International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.18042955 Original Article @2025 RS Publication, [email protected] 253 A. Problem Statement Women often face difficulties in identifying and understanding symptoms related to menstrual and reproductive health disorders such as PCOS, PCOD, and Breast Cancer due to limited awareness, social hesitation, and lack of accessible medical guidance. Many existing digital health solutions are not suitable for users with low literacy levels, provide limited language options, and lack personalized symptom-based analysis. As a result, early diagnosis is frequently delayed, leading to avoidable health complications. Hence, there is a need for a simple, multilingual, and AI-assisted mobile application that enables women to report symptoms comfortably and receive early health guidance in a private and accessible manner. B. Objectives of the Proposed System The primary objectives of the proposed Her Health system are as follows: • To develop an AI-powered chatbot capable of collecting symptoms and identifying risks related to PCOS, PCOD, and Breast Cancer. • To provide multilingual support that allows users to interact with the system in their preferred language. • To enable voice-based symptom input and text-to-speech output for users with limited literacy skills. • To generate personalized health recommendations and awareness content based on symptom analysis. • To ensure secure handling and confidentiality of user health data. • To support reminders and alerts that encourage timely health follow-ups. C. Key Features of the Proposed System The proposed system incorporates the following key features to enhance usability and accessibility: • AI-Based Chatbot: Interacts with users to collect symptoms and provide preliminary health assessment and guidance. • Multilingual Interaction: Supports multiple regional languages to ensure wider accessibility. • Voice Input and Text-to-Speech: Allows users to speak symptoms and receive audio responses, supporting lowliteracy users. • Symptom Classification Engine: Uses machine learning algorithms such as Logistic Regression and Random Forest to predict health risks. • Simple Icon-Based Interface: Ensures easy navigation for non-technical users. • Health Tips and Awareness Content: Provides educational guidance related to menstrual and reproductive health. • SMS Notifications and Alerts: Sends reminders, followups, and important health notifications to users. D. Advantages of the Proposed System The proposed Her Health system offers several advantages over existing solutions: • Enables early identification of PCOS, PCOD, and Breast Cancer through AI-based symptom analysis. • Provides an accessible and user-friendly platform for women with low literacy or limited technical skills. • Supports personalized health guidance based on individual symptoms and risk levels. • Ensures privacy and confidentiality, encouraging users to discuss sensitive health concerns comfortably. • Reduces dependency on frequent hospital visits by offering preliminary digital health support. • Improves health awareness and proactive monitoring, especially for women in rural and underserved regions. II. S YSTEM A RCHITECTURE The Her Health system integrates multilingual communication, AI-based symptom analysis, voice assistance, and personalized health guidance within a unified mobile framework. The architecture is designed to ensure accessibility for users with varied literacy levels while delivering reliable early health screening. International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.18042955 Original Article @2025 RS Publication, [email protected] 254 Fig. 1. Functional Architecture of the Her Health Application A. UI and Interaction Layer The UI layer provides a simple, icon-based interface for user interaction. It manages navigation, collects user inputs through touch or forms, and controls the conversational flow with the chatbot, ensuring ease of use for first-time and low-literate users. B. Voice Processing and TTS Module This module enables natural interaction by converting spoken symptoms into text using speech processing and delivering system responses through text-to-speech (TTS). Multilingual support ensures effective communication for users who face reading difficulties. C. Chatbot Module The chatbot acts as the primary communication interface, guiding users through structured questions, collecting symptom information, and preparing inputs for further analysis. It supports both text and voice-based interaction to encourage comfortable symptom reporting. D. Classification Engine The classification engine processes encoded symptom features and applies machine learning models to predict potential conditions such as PCOS, PCOD, and early breast abnormalities. It supports data-driven early screening and assists users in identifying when medical attention may be required. E. Language Selection and Recommendation Module The language selection module manages user language preferences across the application. Based on classification results, the recommendation module generates personalized lifestyle suggestions, health awareness content, and next-step guidance. F. SMS Notification System The SMS notification system delivers reminders, follow-up alerts, and health notifications, ensuring continued engagement even when the application is inactive. G. System Workflow The system workflow includes symptom input through text or speech, speech-to-text processing, chatbot-based data collection, machine learning–based classification, multilingual output generation using TTS, and delivery of personalized recommendations and SMS alerts. International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.18042955 Original Article @2025 RS Publication, [email protected] 255 Together, these components form a seamless, multilingual, and user-friendly health assessment system that supports women in understanding symptoms, evaluating risks, and accessing timely health guidance. III. I MPLEMENTATION M ETHODOLOGY The implementation of the Her Health application follows a modular and algorithm-driven approach that integrates artificial intelligence, machine learning, multilingual interaction, and voice-based accessibility. The system is implemented using a chatbotdriven workflow combined with supervised machine learning models to enable early screening of PCOS, PCOD, and Breast Cancer. A. Symptom Collection and Chatbot Interaction The implementation begins with an AI-based chatbot that interacts with users through text or voice. The chatbot follows a predefined question flow related to menstrual patterns, lifestyle factors, hormonal symptoms, and breast health indicators. User responses are collected sequentially and stored temporarily for processing. Voice inputs are converted into text using Speech-toText (STT) technology to ensure uniform data handling. B. Data Preprocessing and Feature Encoding Collected symptom inputs undergo preprocessing to convert raw responses into structured features. Numerical values such as age or cycle length are normalized, while categorical responses such as “Yes/No” are encoded into binary values. This preprocessing stage generates a machine-learning-ready feature vector that represents the user’s health profile. C. Machine Learning Algorithms Used The processed feature vector is passed to the classification engine, which applies supervised machine learning algorithms for health risk prediction: • Logistic Regression: Used for symptom-based classification of PCOS and PCOD due to its efficiency, interpretability, and suitability for binary and multi-class prediction. • Random Forest Classifier: Used primarily for Breast Cancer risk prediction, as it combines multiple decision trees to improve accuracy and reduce overfitting. The output of the classification engine determines whether the user is at risk of PCOS, PCOD, Breast Cancer, or shows normal health indicators. D. Result Generation and Recommendation Delivery Based on the predicted class, the system generates personalized health recommendations, including lifestyle suggestions, dietary guidance, and awareness information. The results are presented to the user through text output and converted into speech using Text-to-Speech (TTS) technology to support users with limited literacy. E. Multilingual and Voice-Based Support Language preferences selected by the user are applied across the chatbot, recommendations, and audio outputs. Multilingual support ensures that all text and speech responses are delivered in the user’s preferred language, enhancing accessibility and comfort. F. Notification and Follow-Up Mechanism An SMS notification module is integrated to send reminders, health alerts, and follow-up messages. These notifications ensure continuous engagement and encourage timely health monitoring even when the application is inactive. This implementation methodology ensures that the Her Health system operates as an efficient, accurate, and userfriendly AIpowered health companion, combining machine learning algorithms with accessible interaction techniques for early women’s health screening. International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.18042955 Original Article @2025 RS Publication, [email protected] 256 IV. RESULTS AND ANALYSIS The performance of the proposed Her Health system was evaluated using standard classification metrics including Accuracy, Precision, Recall, and F1-Score for Breast Cancer, PCOS, and PCOD prediction models. Fig. 2 illustrates a comparative analysis of these models. The graph presents a comparative metrics analysis of two classification algorithms—Logistic Regression and Random Forest— based on accuracy, precision, and F1-score. From the visualization, it is evident that Logistic Regression outperforms the Random Forest classifier across all three evaluation metrics. Logistic Regression achieves an accuracy of 0.97, indicating a high overall correctness in predictions. Its precision of 0.98 reflects a strong ability to correctly identify positive cases with very few false positives, while the F1-score of 0.964 demonstrates a wellbalanced performance between precision and recall. These values indicate that Logistic Regression is highly consistent and effective in handling the classification task with minimal misclassification. The Random Forest model also demonstrates strong and reliable performance, with an accuracy of 0.96, precision of 0.961, and an F1-score of 0.954. Although its performance is slightly lower than that of Logistic Regression, the metrics still confirm that Random Forest provides robust and stable predictions. The marginal difference suggests that while Random Forest is effective, it may not be as well-optimized for this specific dataset. Overall, the comparative analysis confirms that both machine learning models perform efficiently, with Logistic Regression emerging as the superior classifier for the given dataset. The consistently high accuracy and F1-score values validate the effectiveness of the proposed AI-based classification framework. Such reliable performance supports the applicability of the system for early disease screening and decision support in women’s health applications, where accurate and consistent predictions are crucial. Fig. 2. Comparative Metrics Analysis of Classification Algorithm V. C ONCLUSION The Her Health application proves that an AI-powered, multilingual support system can greatly improve early detection and management of menstrual health conditions such as PCOS, PCOD, and Breast Cancer. By combining machine learning–based symptom analysis with a chatbot, voice input, and real-time processing, the system offers accessible and meaningful health guidance for women. International Journal of Computer Application ISSN 2250-1797 Available online on https://rspublication.com/ijca/ijca_index.htm Volume 15 Number. 6, 2025 DOI: 10.5281/zenodo.18042955 Original Article @2025 RS Publication, [email protected] 257 Features such as SMS alerts, emergency contacts, and personalized recommendations enhance reliability and ensure timely follow-ups. With its simple and culturally sensitive interface, the application reduces literacy barriers and encourages private, confident health consultations. Overall, the system empowers women to proactively manage their health and forms a strong foundation for future enhancements including deeper AI analytics and expanded language support. VI. F UTURE S COPE In the future, the Her Health system can be further enhanced by introducing complete offline functionality, allowing users to access the chatbot, symptom analysis, recommendations, and health resources even without an active internet connection. By storing essential data locally and enabling lightweight ondevice processing, the application can operate reliably in areas with limited or unstable network access. This improvement will significantly increase accessibility for women in rural and underserved regions, ensuring they receive continuous health support, timely guidance, and essential information at any time. Such enhancements will strengthen the system’s overall usability and make it a more dependable health companion regardless of network availability. R EFERENCES [1] H. Noureen, M. Deb, and E. Hassan, “Sahacharza: Fostering Women’s Wellness via AI-Enhanced Bengali Conversations,” 2023. [2] A. Ajil, M. Sundaram, A. Ali, R. H. V., T. Ahamed, and A. A. Nadaf, “Enhancing the Healthcare by an Automated Detection Method for PCOS Using Supervised Machine Learning Algorithm,” 2023. [3] D. Patel, R. Modi, M. Mhaskar, R. Sarvade, R. Shendage, A. Ram, and V. Bhandiwad, “Machine Learning-Enhanced Solutions for Mitigating PCOS/PCOD-Related Women’s Health Disparities in Rural Areas: A Holistic Telemedicine System Design and Implementation,” 2024. [4] M. Kowsher, F. S. Tithi, M. A. Alam, M. N. Huda, M. M. Moheuddin, and M. G. Rosul, “Doly: Bengali Chatbot for Bengali Education,” Proc. 1st Int. Conf. Adv. Sci. Eng. Robot. Technol. (ICASERT), pp. 1–6, 2019, doi: 10.1109/ICASERT.2019.8934592. [5] P. Dutta, S. Paul, and M. Majumder, “An Efficient SMOTE-Based Machine Learning Classification for Prediction and Detection of PCOS,” 2021. [6] H. Elmannai et al., “Polycystic Ovary Syndrome Detection Machine Learning Model Based on Optimized Feature Selection and Explainable Artificial Intelligence,” Diagnostics, vol. 13, no. 8, 2023. [7] G. Sinthia, T. Poovizhi, and R. Khilar, “Analysis on Polycystic Ovarian Syndrome and Comparative Study of Different Machine Learning Algorithms,” Advances in Intelligent Computing and Communication: ICAC 2021, pp. 191–196, Springer, 2022. [8] S. Rethinavalli and M. Manimekalai, “A Hypothesis Analysis on the Proposed Methodology for Prediction of Polycystic Ovarian Syndrome,” Int. J. Science, Engineering and Computer Technology, vol. 6, no. 11, pp. 396, 2016. [9] S. M. Zahid, Shaik Mohammed, et al., “A Multi Stage Approach for Object and Face Detection using CNN,” in Proc. 8th Int. Conf. Communication and Electronics Systems (ICCES), IEEE, 2023. [10] K. A. Glendining and R. E. Campbell, “Recent advances in emerging PCOS therapies,” Current Opinion in Pharmacology, vol. 68, p. 102345, 2023. [11] M. S. K. Inan et al., “Improved sampling and feature selection to support extreme gradient boosting for PCOS diagnosis,” in Proc. IEEE 11th Annual Computing and Communication Workshop and Conference (CCWC), IEEE, 2021. [12] H. Elmannai et al., “Polycystic Ovary Syndrome Detection Machine Learning Model Based on Optimized Feature Selection and Explainable Artificial Intelligence,” Diagnostics, vol. 13, no. 8, p. 1506, 2023.