Page 154 - 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

Fig 2: Proposed model of the disease predicting system
 Raw Data: V. PERFORMANCE EVALUATION
· This is the starting point, containing unprocessed The performance of the proposed model will be evaluated
information about patients and their health conditions. using the following metrics:
1. Accuracy: Measures the percentage of correct
 Preprocessing :
· The raw data is transformed into a suitable format for predictions.
machine learning algorithms. This involves tasks like: 2. Class-Wise Performance: Assesses precision, recall,
o Cleaning the data to handle missing values or and F1-score for each disease.
inconsistencies.
3. Confidence Scores: Analyzes the reliability of model
o Feature engineering to extract relevant information predictions.
from the raw data.
4. Comparative Analysis: Compares the hybrid model's
o Normalization or scaling of data to ensure features have
performance to single-modal ML or DL systems.
comparable ranges.
5. Scalability and Resource Efficiency: Evaluates the
 Training Set & Testing Set: computational efficiency and adaptability of the model
· The preprocessed data is divided into two subsets:
to diverse datasets.
o Training Set: Used to train the machine learning model.
The model learns patterns and relationships from this This evaluation ensures that the model provides not only
data. accurate predictions but also scalable and resource-efficient
solutions for real-world applications.
o Testing Set: Used to evaluate the model's performance
on unseen data. This helps assess how well the model VI. RESULT ANALYSIS
generalizes to new cases.  Metrics Analysis :
· Prediction Accuracy: Achieved high overall accuracy,
 Final Model:
· The diagram suggests two different approaches to demonstrating reliable performance in predicting
building the final model : Chronic diseases.
o Disease Prediction using CNN: Convolutional Neural · Confidence Scores: The model provided consistent
Networks (CNNs) are deep learning models well-suited confidence levels, indicating robust predictions.
for image analysis. They might be used to analyze
medical images (like X-rays or scans) to predict the · Probability Distribution: Predicted probabilities across
presence of a disease. diseases were balanced, reducing bias toward specific
diagnoses.
o Calculating Distance using KNN: K-Nearest Neighbors
(KNN) is a simpler algorithm that classifies new data · Per-Class Metrics: Precision, recall, and F1-scores
points based on the distances to their nearest neighbors highlighted strong performance across all disease
in the training set. categories, visualized using a confusion matrix.
 Predictive Model for Disease:  Comparative Analysis :
· The final output is a trained model that can be used to · Better Than Single-Modality Systems: The multi-modal
predict the presence or absence of a disease in new approach performed better than single-modality models,
patients based on their data. improving accuracy and diagnosis precision.

IJTSRD | Special Issue on Emerging Trends and Innovations in Web-Based Applications and Technologies Page 144