Paper Title (use style: paper title) Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 10s (2025) 3494 https://internationalpubls.com A Smart System for Predictive Health Monitoring for Next-Gen Patient Care Using IOT 1st Dr. Rajashree Gadhave Pillai HOC College of Engineering and Technology, Rasayani, University of Mumbai Mumbai, India 2nd Madhura Mahindrakar Pillai HOC College of Engineering and Technology, Rasayani, University of Mumbai Mumbai, India Article History: Received: 06-08-2025 Revised: 27-09-2025 Accepted: 26-10-2025 Abstract: Heart disease is a prevalent health issue, especially among the elderly and individuals with unhealthy lifestyles, and can often lead to fatal outcomes. However, its risk can be reduced through routine screenings, healthy dietary practices, and consistent medical checkups. Hospitals generate vast amounts of patient-related information such as X-rays, cardiac assessments, respiratory evaluations, chest pain analyses, and personal health records (PHRs). These datasets contain vital symptoms—essential features for prediction—that help in constructing decision tree classifiers. By applying decision tree methods, we can identify the most relevant attributes that enhance the accuracy of predictions. Despite the availability of such data, hospital records are often underutilized. While some technologies extract valuable insights from heart disease detection databases, their broader application remains limited. This research leverages healthcare data, optimization techniques, and machine learning algorithms to assess whether a patient is likely to suffer from heart disease. The aim is to model patient data to support accurate diagnosis. Keywords— Personal health records (PHRs), Machine Learning, and IOT Introduction Maintaining good health demands proactive, continuous monitoring, especially with the increasing incidence of chronic illnesses. Advances in technologies such as the Internet of Things (IoT) and artificial intelligence (AI) have revolutionized the field of healthcare, enabling smarter, data-driven approaches for disease prediction and patient care. A recent study conducted by the University of Oxford showed that wearable sensor data, such as that from smartwatches, can detect Parkinson’s disease up to seven years before formal diagnosis by recognizing subtle variations in movement [11]. Such advancements highlight the transformative role of AI and IoT in facilitating early detection. The proposed system utilizes IoT-based sensors to collect physiological indicators like ECG, body temperature, heart rate, blood pressure, oxygen saturation, and motion activity. These signals are processed and analyzed in real time using machine learning algorithms [12], [13], offering clinicians valuable insights and enhancing diagnostic precision. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 10s (2025) 3495 https://internationalpubls.com IoT’s incorporation into healthcare, now often termed the Internet of Medical Things (IoMT), provides continuous and non-invasive data collection using Body Sensor Networks (BSN) [5], [15]. Secure systems such as BSN-Care ensure reliable transmission and privacy of patient data, enabling round-the-clock monitoring without manual intervention. Cardiovascular diseases remain the leading cause of mortality globally. In India, they account for 24% of total deaths and approximately 33% worldwide. The asymptomatic progression of heart-related conditions often leads to late-stage diagnoses. Conventional diagnostic methods based on subjective patient histories are not always reliable [17], [18]. Machine learning (ML) has emerged as a valuable tool in predicting cardiovascular risks by analyzing large-scale physiological data [6], [7]. In our approach, attributes such as cholesterol, pulse rate, blood pressure, and heart rate are evaluated using ML algorithms to determine the likelihood of disease onset [4], [8], [10]. With vast volumes of health data now available, intelligent systems capable of deriving meaningful insights are essential for improving healthcare delivery [9]. Given the rise in cardiac mortality, there is an urgent need for predictive systems that offer early warnings. Sedentary lifestyles, poor nutrition, and increased stress contribute to the growing burden of cardiovascular conditions [13], [14], [16]. Leveraging IoT and ML technologies is essential to developing affordable, reliable, and accessible healthcare solutions to address this global concern. I. LITERATURE REVIEW Sun and Pan [11] explored the use of machine learning in heart disease prediction by analyzing both clinical and self-measurable physical indicators. Their study used five algorithms to reduce dependence on clinical diagnostics.Ahmed and Husien [12] proposed a hybrid machine learning model using the IEEE Dataport dataset to predict heart disease. The approach enhanced accuracy by integrating strengths from multiple algorithms.Dahan et al. [5] designed an IoMT-based healthcare architecture where data is collected via sensors and analyzed using LDA and a Cuckoo Search Algorithm, followed by classification with a hybrid ResNet-GoogleNet model. Ayua [14] presented a random forest-based ensemble model to classify weight categories. The model utilized health and lifestyle data to handle non-linear patterns effectively.Khatun et al. [7] reviewed machine learning’s role in securing healthcare-IoT systems. Their study emphasized risk mitigation strategies in digital health environments.Sarker [17] developed a predictive model leveraging machine learning on diverse patient datasets. The study focused on optimizing healthcare delivery and early illness detection.Islam et al. [6] used a design science research framework to build an IoT-based cardiovascular prediction system. The conceptual model was informed by healthcare needs and literature gaps.Bhatt et al. [3] applied k-modes clustering along with ML classifiers like RF, DT, and XGBoost. They used GridSearchCV to enhance performance across 70,000 real-world records. Ratta and Sharma [16] proposed a blockchain-integrated ML ecosystem for diabetic patient monitoring. The architecture ensured data integrity and enabled secure real-time health insights.Butt et al. [9] explored AI and ML in 5G-enabled healthcare systems. Their review addressed issues like data interoperability and infrastructure challenges in smart healthcare.Kotagi et al. [15] focused on ensuring dataset accountability in ML by applying software engineering techniques. They proposed structured mechanisms for data traceability and validation.More et al. [8] reviewed IoT-based smart hearing devices that enhance Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 10s (2025) 3496 https://internationalpubls.com accessibility. The work highlighted sensor integration and adaptive processing for better hearing support.Arabelli et al. [20] explored the role of artificial intelligence in evolving IoT infrastructure to meet next-generation demands. The study highlighted major challenges and opportunities in enhancing automation, connectivity, and real-time decision-making in smart systems. A. Limitations of the Existing Studies Despite considerable progress in integrating IoT and machine learning into healthcare systems, several limitations persist across existing studies. Notably, the lack of comprehensive, real-time mitigation strategies for IoMT security threats remains a significant concern [12], [5]. While many frameworks propose theoretical security layers, practical validation and deployment are often missing, raising questions about their reliability in live clinical environments [3], [10]. The absence of robust privacy-preserving mechanisms within 5G-enabled infrastructures adds further vulnerability to patient data confidentiality [6]. Additionally, scalability has not been thoroughly evaluated, particularly in systems designed to operate across large healthcare facilities or serve geographically dispersed populations [7]. Moreover, many studies rely on limited or homogeneous datasets, which restrict the generalizability of predictive models to diverse patient profiles and real-world conditions [13], [14]. A standardized framework for managing cloud-based patient records is also lacking, thereby amplifying data security risks in multi-tenant or distributed architectures [9]. Deep learning approaches, although effective, are often hindered by constrained training data and lack flexibility in adapting to complex clinical variations [4]. Numerous works also remain largely conceptual, without practical demonstrations or pilot deployments that could validate performance in operational settings [8]. In addition, factors such as sensor precision, reliability, and computational efficiency are frequently overlooked, despite their critical role in ensuring system effectiveness and suitability for real-time healthcare applications [17], [19], [11], [6]. II. PROPOSED ARCHITECTURE The proposed architecture, as shown in Fig. 1, represents an intelligent healthcare monitoring system that integrates IoT-based sensors and machine learning techniques for real- time health analysis. Physiological parameters such as ECG, body temperature, motion, oxygen saturation, heart rate, and blood pressure are continuously monitored using wearable sensors. These analog signals are converted into digital form through an Analog-to-Digital Converter (ADC) and processed by a microcontroller. The processed data is stored in a database (DB) and utilized by various machine learning algorithms including Q-Learning, Fuzzy Logic, Random Forest, Linear Regression, Artificial Neural Network (ANN), and Naive Bayes for health prediction. In parallel, a training module collects external data via a network, preprocesses it, and generates inference rules that are stored in a Background Knowledge (BK) database. The BK data is then employed by the proposed I-Heart algorithm to enhance the prediction results and support advanced health condition analysis. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 10s (2025) 3497 https://internationalpubls.com Figure 1: Proposed architecture In below section we described each module in detail 1) Preprocessing Preprocessing is a crucial phase in the proposed healthcare monitoring system, responsible for refining raw patient data before feeding it into machine learning models. The main preprocessing steps include: ● Data Cleaning: Removes noise, missing values, and duplicates using methods such as imputation and outlier detection to ensure the integrity of the dataset. ● Normalization: Scales sensor data to a consistent range, typically between 0 and 1, to standardize input values and improve the performance of learning algorithms. ● Data Transformation: Converts unstructured sensor readings into structured and analyzable formats. This includes encoding categorical variables and reshaping time- series data. ● Feature Selection: Identifies the most significant attributes from the dataset that influence the prediction outcome. This step reduces dimensionality, decreases computational complexity, and enhances model accuracy. These preprocessing steps collectively ensure that the input data is clean, well-structured, and optimized for effective training and testing of the machine learning-based diagnostic system. 2) Training: • Gather both synthetic and real-time patient data from doctors and IoT devices. • Utilize data mining methods, including preprocessing, data cleaning, and classification, to process the collected patient information. • The processed data is then stored in a database referred to as background knowledge, which serves as a reference during the testing phase. 3) Testing: • The system processes both synthetic and real-time patient data received via the internet to predict the likelihood of disease using a pre-trained model. • A link-based architecture is employed to store all collected data in a centralized global database. • During the testing phase, both training and testing datasets are accessed concurrently. • Machine learning classification techniques are applied to anticipate outcomes and support future decision-making processes. • Ultimately, the system evaluates its performance by comparing accurate (Normal) and (Critical) results, ensuring the reliability of the study Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 10s (2025) 3498 https://internationalpubls.com III. METHODOLOGY The proposed system offers several advantages that contribute to its practicality, cost- effectiveness, and user accessibility in healthcare environments. By utilizing a reduced number of sensors while maintaining accurate outcomes, the system ensures efficient resource usage without compromising data quality. This approach directly contributes to lowering the overall application cost, making it more feasible for widespread deployment. Additionally, the system is designed to be energy-efficient, enabling continuous operation in both remote and resource-limited settings. One of the key benefits is the automation of the monitoring process, which eliminates the need for manual measurement, thereby reducing human error and increasing reliability. Furthermore, patient health conditions are analyzed in real time and communicated directly to medical professionals via email, ensuring timely intervention with high accuracy. The integration of all vital health monitoring parameters into a single, compact unit also enhances usability, making the system easy to operate even for individuals without technical expertise. Overall, these advantages make the system a robust and user-friendly solution for modern healthcare monitoring needs. Algorithm Input: Input values for all parameters HashMap which contains the all attributes values like { ECG, temperature, heart rate, motion sensors, blood pressure, oxygen saturation} etc. Policy patterns {P1, P2, and Pn} Output : Generate sample report for individual patient. Step 1 : for each read Hashmap (1) Step 2: if Extracted_Attribute[j] similar to P[1] NormalPos = +1 MasterLits1. Add(NormalPos) Step 3 : if Extracted_Attribute[j] similar to P[2] AbnPos = +1 MasterLits2. Add (AbnPos) Step 4: end for Step 5 : calculate the fitness factor for all classes using below formula for all class list (2) Step 6: Weight _Current List[w] = *100 (3) Step 7 : Sort _CurrentList[w] using desc order Step 8 : Recommend _CurrentList[0] for final class for patient profile. Step 9 : end procedure. IV. RESULT Integrating quantitative assessments with clinical ratings is logical and may effectively resolve several evaluation issues. We used hybrid machine learning models on the training dataset that was made to find common, suspicious, and dangerous patterns of behavior. To test and evaluate the machine learning methods, we used cross-validation models with the behavioral categorization training dataset. Figure 1.2 depicts the classification system in overall factors and illustrates the uniformity of classifiers. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 10s (2025) 3499 https://internationalpubls.com Figure 2: The accuracy evaluation of various ML classifiers Performance analysis of algorithm The algorithm's efficiency has been analyzed by comparing the time complexity of existing methods with that of the proposed approach. Table 1 presents a detailed performance comparison of all the algorithms within the specified experimental setup. Table 1: Comparative analysis of system Comparative Factor ML Proposed NB RF ANN Q- Learning Fuzzy LR Reliability High Mediu m Low Medium Low Medium Medium Security Medium Low Medium Low Medium Low Low Performance High High Low Medium Low Low High Portability Yes No Yes Yes Yes No No Configuration Medium Low Medium Low Medium Low Low Compatibility Highly Mediu m Low Low High Medium Medium To further support the evaluation, Table 2 presents the accuracy achieved by each classifier using the cross-validation results. It confirms the superior performance of the proposed hybrid model in terms of classification accuracy. Table 2: Accuracy Comparison of Classifiers Classifier Accuracy (%) Proposed Model 95.6 Naive Bayes (NB) 83.4 Random Forest (RF) 78.9 ANN 85.2 Q-Learning 80.3 Fuzzy Logic 79.1 Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 10s (2025) 3500 https://internationalpubls.com Linear Regression 84.7 The proposed system demonstrates superior reliability, performance and accuracy, approximately a 12.21% improvement when compared to traditional algorithms, establishing its effectiveness in real-time intelligent health monitoring applications. V. CONCLUSION The Internet of Things (IoT) framework stands out as a practical and efficient technology, offering an accessible platform at an affordable cost, even in specialized sectors like healthcare. Healthcare, being a vital aspect of our daily lives, greatly benefits from IoT by enabling the collection and transmission of sensor data through smart devices. These intelligent systems enhance monitoring and care, especially for individuals requiring continuous attention. At its core, this integration of sensory data with web-based applications or smart gadgets demonstrates a high level of intelligent automation. As outlined in the referenced study, we focused on converting sensor data—captured through devices like the Arduino Uno—into web applications and databases, allowing real-time access to patient information. Unlike traditional methods, which often involve invasive procedures that can cause discomfort and discourage patients from regular monitoring, this research promotes a non-invasive alternative. The goal is to empower patients with a system that provides their vital health information conveniently and painlessly. With this IoT-based model, patients can stay connected to healthcare professionals around the clock and receive timely alerts during emergencies. The proposed framework is capable of tracking key health metrics such as body temperature, blood pressure, heart rate, and other critical indicators necessary for accurate heart health analysis. REFERENCES [1] 1. S. S. Khatal and Y. K. Sharma, “Analyzing the role of heart disease prediction system using IoT and machine learning,” Int. J. Adv. Sci. Technol., vol. 29, no. 9s, pp. 2340– 2346, 2020. [2] 2. Q. He, A. Maag, and A. Elchouemi, “Heart disease monitoring and predicting by using machine learning based on IoT technology,” in Proc. 2020 5th Int. Conf. Innov. Technol. Intell. Syst. Ind. Appl. (CITISIA), IEEE, 2020. [3] 3. C. M. Bhatt et al., “Effective heart disease prediction using machine learning techniques,” Algorithms, vol. 16, no. 2, Art. no. 88, 2023. [4] 4. M. Badawy, N. Ramadan, and H. A. Hefny, “Healthcare predictive analytics using machine learning and deep learning techniques: A survey,” J. Electr. Syst. Inf. Technol., vol. 10, no. 1, Art. no. 40, 2023. [5] 5. F. Dahan et al., “A smart IoMT based architecture for E-healthcare patient monitoring system using artificial intelligence algorithms,” Front. Physiol., vol. 14, Art. no. 1125952, 2023. [6] 6. M. N. Islam et al., “Predictis: An IoT and machine learning-based system to predict risk level of cardio-vascular diseases,” BMC Health Serv. Res., vol. 23, no. 1, Art. no. 171, 2023. [7] 7. M. A. Khatun et al., “Machine learning for healthcare-IoT security: A review and risk mitigation,” IEEE Access, vol. 11, pp. 145869–145896, 2023. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 10s (2025) 3501 https://internationalpubls.com [8] 8. S. More, S. Patil, P. Salvi, and R. Gadhave, “Enhancing hearing accessibility through IoT: A review of smart hearing device,” Int. J. Distrib. Comput. Technol., vol. 9, no. 1, pp. 11–17, 2023. [9] 9. H. A. Butt et al., “5G and IoT for intelligent healthcare: AI and machine learning approaches—A review,” in Proc. Int. Conf. Smart Objects Technol. Social Good, Cham, Switzerland: Springer, 2023. [10] 10. S. S. Khatal and M. D. Rokade, “An detection of cardiovascular disease using deep learning,” in Proc. 2023 Int. Conf. Emerg. Smart Comput. Informat. (ESCI), IEEE, 2023. [11] 11. H. Sun and J. Pan, “Heart disease prediction using machine learning algorithms with self-measurable physical condition indicators,” J. Data Anal. Inf. Process., vol. 11, no. 1, pp. 1–10, 2023. [12] 12. M. Ahmed and I. Husien, “Heart disease prediction using hybrid machine learning: A brief review,” J. Robot. Control, vol. 5, no. 3, pp. 884–892, 2024. [13] 13. N. Alharbe and M. Almalki, “IoT-enabled healthcare transformation leveraging deep learning for advanced patient monitoring and diagnosis,” Multimed. Tools Appl., pp. 1– 14, 2024. [14] 14. S. I. Ayua, “Random forest ensemble machine learning model for early detection and prediction of weight category,” J. Data Sci. Intell. Syst., vol. 2, no. 4, pp. 233–240, 2024. [15] 15. Kotagi, V., Nassa, V. K., Patil, D., Gadhave, R., Adusumilli, S. B. K., & Kumar, P. P. (2024, September). Ensuring Dataset Accountability in Machine Learning: Insights from Software Engineering. In 2024 7th International Conference on Contemporary Computing and Informatics (IC3I) (Vol. 7, pp. 385-389). IEEE. [16] 16. P. Ratta and S. Sharma, “A blockchain-machine learning ecosystem for IoT-based remote health monitoring of diabetic patients,” Healthc. Anal., vol. 5, Art. no. 100338, 2024. [17] 17. M. Sarker, “Revolutionizing healthcare: The role of machine learning in the health sector,” J. Artif. Intell. Gen. Sci., vol. 2, no. 1, pp. 36–61, 2024. [18] 18. A. Sundas et al., “Smart patient monitoring and recommendation (SPMR) using cloud analytics and deep learning,” IEEE Access, 2024. [19] 19. A. H. Najim et al., “An IoT healthcare system with deep learning functionality for patient monitoring,” Int. J. Commun. Syst., vol. 38, no. 4, Art. no. e6020, 2025. [20] R. Arabelli, D. K. Chotaliya, R. T. Gadhave, B. Haritha, L. C. Kasireddy and M. Soni, "Artificial Intelligence in Next Generation IoT Infrastructure: Challenges and Opportunities," 2025 First International Conference on Advances in Computer Science, Electrical, Electronics, and Communication Technologies (CE2CT), Bhimtal, Nainital, India, 2025, pp. 1436-1440, doi: 10.1109/CE2CT64011.2025.10939142