Page 160 - 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
(combining text and image features via TF-IDF vectorization F1-score, as well as confidence scores and comparative
and ResNet-18 extraction) and the testing set used to analyses. Results indicate high prediction accuracy, robust
evaluate generalization. The model is trained using confidence levels, and resource-efficient scalability, affirming
optimization techniques to minimize cross-entropy loss and the model’s reliability and adaptability.
evaluated using metrics like accuracy, precision, recall, and

Fig: Model Architecture
V. PERFORMACE EVALUATION
The performance of the proposed hybrid medical prediction model has been rigorously evaluated using a diverse set of metrics
to ensure its accuracy, reliability, and scalability in real-world medical applications. The following subsections highlight the key
aspects of the evaluation:
A. Overall Accuracy and Reliability
1. The model achieved a training accuracy of 100% and a test accuracy of 99.52%, demonstrating its robust learning
capabilities and excellent generalization performance. These metrics indicate the model's effectiveness in classifying
diseases accurately across diverse datasets, instilling confidence in its reliability for clinical decision-making.
B. Category-Specific Performance
1. For each disease category, the model's precision, recall, and F1-score were evaluated to provide a granular understanding
of its classification capabilities:
2. Macro Average Precision: 99.70%
3. Macro Average Recall: 99.78%
4. Macro Average F1-Score: 99.73% These metrics affirm that the model maintains consistent high performance across all
disease categories, minimizing the risk of misclassification. For example:
5. Diseases like Alzheimer's, Arthritis, and Diabetes achieved perfect precision, recall, and F1-scores of 1.0, reflecting their
flawless classification.
6. Slight variations in metrics, such as for coronary artery disease (F1-score = 94.73%) and Hypertension (F1-score =
96.30%), highlight areas for potential improvement.

IJTSRD | Special Issue on Emerging Trends and Innovations in Web-Based Applications and Technologies Page 150