Page 161 - Emerging Trends and Innovations in Web-Based Applications and Technologies

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
C. Prediction Reliability
1. The model's predictions were evaluated for confidence scores, which indicate the certainty of its classifications. The
uniformly high scores suggest a robust prediction mechanism, crucial for applications requiring a high degree of reliability,
such as clinical diagnostics.
D. Competitive Assessment
1. A comparative analysis was conducted to benchmark the hybrid model against traditional single-modal machine learning
(ML) and deep learning (DL) systems. Results highlight that integrating multiple data types significantly enhances
diagnostic accuracy and reliability, outperforming existing methodologies by:
2. Providing higher recall and precision.
3. Achieving improved classification in edge cases (e.g., rare diseases like Chronic Fatigue Syndrome and Ulcerative Colitis).
E. Efficiency and Scalability
1. The model's scalability was tested with increasing data sizes and complexities:
2. It demonstrated efficient handling of large datasets without degradation in performance.

3. Feature importance analysis revealed that attributes like "inhalers," "bronchodilators," and "family history" played a
significant role in predictions, enabling resource-efficient computation.
F. Confusion Matrix Analysis
1. The confusion matrix confirmed that the majority of classifications were accurate, with minimal misclassifications. For
instance:
2. Diseases like Asthma and Diabetes were classified perfectly with no false positives or negatives.
3. Rare misclassifications, such as for Hypertension, were minimal and did not significantly impact overall performance.
This evaluation validates the model's accuracy, scalability, and resource efficiency, establishing it as a reliable tool for
improving diagnostic workflows and patient outcomes in healthcare settings. This comprehensive evaluation ensures that the
hybrid medical prediction model provides not only accurate predictions but also scalable and resource-efficient solutions for
real-world applications. By using these metrics, the evaluation validates the model's effectiveness and applicability in
improving diagnostic accuracy and clinical workflows, ultimately enhancing patient care and outcomes.

Fig: Model Accuracy & Average Metrics

Here as we can as the disease I gets more rare the F1-score drops as there is very low level of label dataset available which can
be used to train such chronic disease prediction system models

IJTSRD | Special Issue on Emerging Trends and Innovations in Web-Based Applications and Technologies Page 151