A scalable solution for multi-symptom disease prediction and lifestyle recommendations using machine learning.

Area of Science:

• Intelligent healthcare systems
• Machine learning applications in medicine
• Clinical decision support systems

Background:

• Accurate disease prediction from symptoms is crucial for intelligent healthcare.
• Machine learning (ML) offers potential for early symptom-based risk screening.
• Clinical adoption of ML is hindered by data quality, validation, interpretability, and diagnostic claims.

Purpose of the Study:

• To present a mobile decision-support prototype for multi-symptom disease prediction.
• To offer rule-based lifestyle recommendations linked to predicted diseases.
• To evaluate ML model performance on a public dataset for benchmark purposes.

Main Methods:

• Utilized the Kaggle Medicine Recommendation System Dataset (4,920 records, 41 disease classes).
• Evaluated seven supervised ML models: Decision Tree, Random Forest, Naive Bayes, Logistic Regression, XGBoost, SVM, and KNN.
• Developed an Android-Flask prototype integrating prediction with a recommendation layer.

Main Results:

• The best-performing ML models achieved high classification accuracy on the benchmark dataset.
• The prototype successfully linked predicted disease classes to relevant information (descriptions, precautions, diet, workouts).
• Results are presented as benchmark performance, not clinical diagnostic validity.

Conclusions:

• The developed system is a decision-support and educational prototype.
• It is not a clinically validated diagnostic or prescribing tool.
• Future work requires external validation, clinician assessment, and real-world testing.