Enhancing Medication Adherence in Smart Medical Reminder Systems using The AI-MA Algorithm in AI

Enhancing Medication Adherence in Smart Medical Reminder Systems using The AI-MA Algorithm in AI

Sasipriya S (Assistant Professor), Department of Information Technology,

K.S.R. College of Engineering, Tiruchengode,

Tamil Nadu, India.

Janani S (Student),

Department of Information Technology,

K.S.R. College of Engineering Tiruchengode,

Tamil Nadu, India.

Pradeepa D M (Student), Department of Information Technology,

K.S.R. College of Engineering, Tiruchengode,

Tamil Nadu, India.

AbstractThis paper presents a Smart Medical Reminder System (SMRS) powered by the AI-based Medication Adherence (AI-MA) algorithm to address medication noncompliance, a critical issue in chronic disease management. By integrating machine learning, predictive analytics, and IoT devices, the system personalizes reminder schedules, verifies medication intake, and triggers real-time alerts for missed doses. The AI-MA algorithm adapts to user behavior, optimizing reminder timing and modality while ensuring privacy through federated learning and encryption. The system enhances patient engagement, reduces hospitalizations, and aligns with modern healthcare innovation goals.

KeywordsMedication Adherence, AI-MA Algorithm, Smart Medical Reminder Systems, Machine Learning, IoT, Predictive Analytics.

Teenu V (Student), Department of Information Technology,

K.S.R. College of Engineering, Tiruchengode,

Tamil Nadu, India.

Loshini M (Student), Department of Information Technology,

K.S.R. College of Engineering, Tiruchengode,

Tamil Nadu, India.

1. INTRODUCTION

   Medication non-compliance is a critical challenge in healthcare, particularly for managing chronic diseases such as diabetes, hypertension, and asthma. Failure to adhere to prescribed regimens leads to severe health complications and increased economic burdens on healthcare systems worldwide. Traditional reminder systems, which rely on static schedules, often fall short due to forgotten alerts, lack of mechanisms to confirm medication intake, and limited user engagement. Additional drawbacks include high battery consumption, privacy vulnerabilities, and an inability to adapt to individual patient behaviors, reducing their effectiveness across diverse populations. To address these limitations, this paper proposes a Smart Medical Reminder System (SMRS) powered by the AI-based Medication Adherence (AI-MA) algorithm. By integrating artificial intelligence (AI), the Internet of Things (IoT), and mobile technology,

   SMRS offers a dynamic, user-centric solution to enhance medication adherence. The AI-MA algorithm leverages machine learning, deep learning, and natural language processing to analyze behavioral and biometric data, predict risks of non-adherence, and deliver personalized reminders tailored to each users lifestyle and preferences. The system interfaces with IoT devices, including smart pill dispensers, wearables, and mobile applications, to monitor real-time medication intake and verify compliance through sensor feedback or user interactions. In cases of missed doses, SMRS escalates notifications to caregivers or clinicians, ensuring timely interventions. Privacy is safeguarded through federated learning and encryption, aligning with standards like HIPAA and GDPR. Multi channel reminders, such as SMS, push notifications, and voice alerts, enhance accessibility, particularly for elderly or visually impaired users. By fostering stronger patient engagement and enabling real-time clinician monitoring, SMRS aligns with global healthcare innovation goals for adaptive and preventive care models.

2. LITERATURE SURVEY

Aldeer et al. (2018): Compared sensor, vision, and proximity systems using SVMs, Naïve Bayes, KNN, and template matching. Strengths: Diverse monitoring approaches. Weaknesses: High energy use, environmental noise, and multi-user authentication challenges.

Bohlmann et al. (2021): Used logisticregression, ANNs, SVMs, and random forest classifiers to predict medication adherence from self reported, sensor, and pharmacy data. Strengths: High accuracy (up to 93.75%), robust predictor detection, wearable sensor integration, and conversational AI. Weaknesses: Variable adherence measurement quality, poor smartwatch activity differentiation, intrusive reminders.

Singh & Varshney (2021): Proposed MDCAR for context-aware reminders, tested via expert judgment. Strengths: Improved adherence over conventional reminders, generalizable. Weaknesses: Needs precise contextual data, interoperability issues.

Babel et al. (2023): Employed decision trees, Naïve Bayes, SVMs, and gradient boost on mobile platforms for non-communicable disease (NCD) adherence. Strengths: Improved adherence, patient engagement, tailored interventions, and better outcomes. Weaknesses: Digital inequality, depersonalized care, infrastructure dependency, and

user variability.

improved glycemic control, and reduced distress. Weaknesses: Limited sample size, single clinical setting.

Liang et al. (2023): Used rule-based scheduling, sensor processing (threshold detection, decision trees), and human-robot interaction (HMMs, CNNs) for geriatric medication management. Strengths: Real-time management, high usability, engagement. Weaknesses: Technology interaction dependency, long-term adoption issues.

Sumner et al. (2023): Applied reinforcement learning (Q-learning), clustering (K-means), and rule based AI for personalized nudge interventions.

Strengths: High acceptability, personalized delivery, low burden. Weaknesses: Low interest in incentives, balancing simplicity and personalization.

Sekandi et al. (2023): Developed deep learning CNNs (3D ResNet, VGG16, etc.) for video based tuberculosis adherence monitoring. Strengths: High sensitivity (92.895.8%), good F1-score (0.91 0.92). Weaknesses: Few public datasets, low specificity.

Almeman (2024): Utilized AI recommendation, blockchain (PoS, PBFT), and predictive modeling for pharmacy digital transformation. Strengths: Enhanced access, patient care, high-tech integration. Weaknesses: Regulatory gaps, drug safety issues, and digital infrastructure reliance.

Gargioni et al. (2024): Reviewed pill dispensers, emphasizing usability scoring and patient behavior modeling. Strengths: Focus on human centered design, accessibility. Weaknesses: No specific algorithm proposed.

Ghozali (2024): Highlighted AI tools like AiCure (computer vision, CNNs) and reinforcement learning for adherence. Strengths: Tracks medication via smartphones, adaptive reminders. Weaknesses: Data privacy, digital literacy gaps, and infrastructure issues.

Baker & Xiang (2021, 2024): Explored AIoT for healthcare, using deep learning, reinforcement learning, and federated learning. Strengths: Scalable, privacy-preserving, supports remote monitoring. Weaknesses: Data interoperability, real-time processing limits, and explainability issues.

Kanyongo & Ezugwu (2024): Reviewed ML (logistic regression, random forest, SVMs, XGBoost,

Nayak et al. (2023): Developed a voice-based etc.) for NCD adherence. Strengths: High potential for AI platform with NLP (intent recognition, sequence-to prediction and monitoring. Weaknesses: Limited by sequence models) and reinforcement learning for data quality and study variability.

insulin management in type 2 diabetes. Strengths:

Faster insulin optimization, better compliance, III. EXISTING SYSTEM

An advanced AI system enhances medication adherence for chronic disease managemen by employing machine learning, deep learning, natural language processing, and predictive analytics. Utilizing algorithms like decision trees, support vector machines, neural networks, and reinforcement learning (Q- learning, Markov Decision Processes), it customizes reminders based on patient behavior and real-time data from wearables and electronic medical records. Interactive chatbots and voice assistants, powered by natural language processing, engage patients, while IoT devices track adherence and detect side effects. The system offers tailored interventions and early risk identification and aligns with Saudi Vision 2030s healthcare goals. Challenges such as data privacy, user resistance, and regulatory compliance are addressed through explainable AI and adherence to HIPAA/GDPR standards, ultimately reducing medication errors and hospitalizations and improving patient outcomes.

1. PROPOSED SYSTEM

   An advanced AI-driven system boosts medication adherence by leveraging machine learning algorithms (Random Forest, Decision Trees, Neural Networks) and reinforcement learning to tailor schedules based on patient behavior, lifestyle, and real time data from IoT devices, mobile apps, and smartwatches. It predicts non-adherence risks and delivers adaptive reminders through SMS, voice assistants, or wearables, ensuring accessibility for elderly or visually impaired patients. Integrated with electronic health records, it provides healthcare providers with real-time compliance reports for timely interventions. Cloud-based predictive analytics optimize performance, minimizing errors and hospitalizations. Aligned with Saudi Vision 2030, it enhances patient safety and healthcare outcomes while addressing data privacy and user adoption challenges through secure systems and HIPAA/GDPR compliance.

   1. MODULE DIAGRAM

   2. SYSTEM DESIGN & DATA ACQUISITION

      system architecture & data collection framework, enabling secure and efficient gathering of patient information from multiple sources. This is followed by AI-Powered Medication Recognition, where machine learning algorithms identify medications using visual and contextual cues. Simultaneously, IoT-based real time monitoring captures live behavioral and physiological data to track adherence.

   3. AI MODEL DEVELOPMENT & OPTIMIZATION

      Collected data is utilized for AI model training & deployment, enabling predictive decision making regarding medication intake. Continuous AI optimization & future enhancements ensure the system evolves based on feedback and new patterns, improving performance and adaptability.

   4. SECURITY, APPLICATION, AND INTEGRATION

      To support real-world use, a mobile app development process provides users with an interactive platform. This layer integrates security & compliance measures to protect personal data and ensures regulatory alignment (HIPAA/GDPR). Moreover, the system facilitates caregiver and healthcare integration, allowing professionals to access relevant data for patient oversight.

   5. EXPLAINABILITY & PERSONALIZATION

      Explainable AI (XAI) modules are embedded to provide transparency in AI decisions, making them interpretable by both users and healthcare staff. Building on this, AI-based personalized coaching delivers adaptive guidance and behavioral nudges

      tailored to individual users, enhancing adherence and health outcomes.

   6. EDGE INTELLIGENCE & EMERGENCY RESPONSE

      Advanced intelligence is pushed to the edge through wearables & edge AI, supporting real-time processing on low-power devices. This is complemented by Federated AI & Adaptive Learning, enabling local model training without compromising privacy. Finally, emergency handling protocols are triggered in critical scenarios, such as missed doses or health anomalies, ensuring timely interventions.

2. ALGORITHM

   The AI-driven system, powered by the AI MA algorithm, enhances medication adherence for chronic disease management by integrating machine learning like Random Forest, Decision Trees, Neural Networks, reinforcement learning like Q-learning, Markov Decision Processes, and deep learning like RNNs,

   The foundation begins with building a robust LSTMs to predict non-adherence risks and personalize

   reminder schedules using real-time patient behavior and IoMT data, while NLP enables interactive chatbots and voice assistants, predictive analytics assigns risk scores, federated learning ensures privacy

   preserving model updates, and edge AI and blockchain enhance local processing and secure, auditable adherence logs.

3. RESULT ANALYSIS

   The AI-MA system enhances medication adherence using a mobile-based platform with the AI MA algorithm, integrating ML like Random Forest, Neural Networks, NLP, and predictive analytics. It collects data via apps and wearables, dynamically optimizing reminders based on behavior and biometric inputs. Real-time monitoring tracks compliance, triggering alerts for missed doses or emergencies. Evaluated for adherence rate (92.3% vs. 65.4%), response time (300.6 ms vs. 500.4 ms), and user satisfaction (9.1/10 vs. 6.5/10), it outperforms existing systems. Battery consumption is reduced (50% vs. 80%), and predictive accuracy is improved (92% vs. 60%). Emergency alerts and intake verification enhance safety. Aligned with Saudi Vision 2030, it faces privacy and adoption challenges, mitigated by HIPAA/GDPR compliance.

4. CONCLUSION

   The Smart Medical Reminder System enhances medication adherence through AI-driven analytics and personalized reminders. By addressing limitations of traditional systems, our AI-MA Algorithm optimizes alerts, verifies intake, and triggers emergency notifications, improving compliance and healthcare outcomes. Evaluated through key performance metrics, it minimizes ignored notifications and enhances doctor-patient interaction. As AI advances, future improvements will further refine adaptive reminders and health analytics, setting a new standard for AI-driven healthcare solutions.

5. FUTURE ENHANCEMENT

Future advancements in the Smart Medical Reminder System will focus on enhancing adaptability, security, and user engagement. Implementing reinforcement learning and adaptive algorithms will refine reminder accuracy by learning from user behavior. The integration of multimodal health analytics, including biometric data and wearable sensors, will personalize medication schedules while

detecting potential health risks. AI-driven predictive adherence modeling will anticipate missed doses and adjust reminders proactively. Additionally, voice based interactions and chatbot assistance will improve accessibility, particularly for elderly or visually impaired users. To ensure data security, blockchain technology will be explored for secure medication tracking and adherence records. Lastly, seamless integration with electronic health records (EHRs) will enable real-time updates for healthcare providers, ensuring more effective treatment plans. These advancements will

XI. REFERENCES

1. Almeman, "The Digital Transformation in Pharmacy: Embracing Online Platforms and the Cosmeceutical Paradigm Shift," Journal of Health, Population and Nutrition, vol. 43, no. 60, pp. 122, 2024.

2. Bohlmann, J. Mostafa, and M. Kumar, "Machine Learning and Medication Adherence: Scoping Review," JMIRx Med, vol. 2, no. 4, pp. e26993, 2021.

3. Babel, R. Taneja, F. M. Malvestiti, A. Monaco, and S. Donde, "Artificial Intelligence Solutions to Increase Medication Adherence in Patients With Non-communicable Diseases," Frontiers in Digital Health, vol. 3, Article 669869 pp. 1

   9, 2021.

4. J. Sumner, A. Bundele, H. W. Lim, P. Phan, M. Motani, and A. Mukhopadhyay, "Developing an Artificial Intelligence-Driven Nudge Intervention to Improve Medication Adherence: A Human-Centered Design Approach," Journal of Medical Systems, vol. 48, no. 3, pp. 110, 2024.

5. J. N. Sekandi, W. Shi, R. Zhu, P. Kaggwa, E. Mwebaze, and S. Li, "Application of Artificial Intelligence to the Monitoring of Medication Adherence for Tuberculosis Treatment in Africa: Algorithm Development and Validation," JMIR AI, vol. 2, no. 1, pp. e40167, 2023.

6. L. Gargioni, D. Fogli, and P. Baroni, "A Systematic Review on Pill and Medication Dispensers from a Human-Centered Perspective," Journal of Healthcare Informatics Research, vol. 8, no. 3, pp. 244 285, 2024.

7. M. T. Ghozali, "The Role of Artificial Intelligence in Enhancing Medication Adherence: A Mini Review," Jurnal JurnalCendekiaIlmiah (J-CEKI), vol. 3, no. 6,

   pp. 87718781, Oct. 2024.

8. M. Aldeer, M. Javanmard, and R. P. Martin, "A Review of Medication Adherence Monitoring Technologies," Applied System Innovation, vol. 1, no. 2, pp. 127, 2018.

9. Nayak, S. Vakili, K. Nayak, M. Nikolov, M. Chiu, P. Sosseinheimer, S. Talamantes, S. Testa, S. Palanisamy, V. Giri, and K. Schulman, "Use of Voice-Based Conversational Artificial Intelligence for Basal Insulin Prescription Management Among Patients With Type 2 Diabetes: A Randomized Clinical Trial," JAMA Network Open, vol. 6, no. 12, pp. e2340232, 2023.

10. N. Singh and U. Varshney, "Medication adherence: A method for designing context aware reminders," International Journal of Medical Informatics, vol. 132, p. 103980,

    2019.

11. S. Baker and W. Xiang, "Artificial Intelligence of Things for Smarter Healthcare: A Survey of Advancements, Challenges, and Opportunities," IEEE Communications Surveys & Tutorials, vol. 25, no. 2, pp. 1261

    1274, 2023.

12. Sheik Abdullah, R. Kumar, and N. Venkatesh, "Design of Automated Model for Inspecting and Evaluating Handwritten Answer Scripts: A Pedagogical Approach with NLP and Deep Learning," Educational Data Mining Journal, vol. 16, no. 2, pp. 7692, 2024.

13. V. Peddisetti, P. K. Kandregula, J. A. John, S. Poomdla, K. George, and A. Panangadan, "Smart Medication Management: Enhancing Medication Adherence with an IoT-based Pill Dispenser and Smart Cup," unpublished manuscript, California State University, Fullerton, 2023.

14. W. Kanyongo and A. E. Ezugwu, "Machine learning approaches to medication adherence amongst NCD patients: A systematic literature review," Informatics in Medicine Unlocked, vol. 38, p. 101210, 2023.

15. Y. Gu, A. Zalkikar, M. Liu, L. Kelly, A. Hall,

K. Daly, and T. Ward, "Predicting medication adherence using ensemble learning and deep learning models with large-scale healthcare

data," Scientific Reports, vol. 11, no. 18961, 2021.