Intelligent Disease Prediagnosis Only Based on Symptoms

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Computational Biology

Background:

• Understanding the relationship between symptoms and diseases is crucial for medical advice.
• Existing diagnostic methods can be time-consuming and require expert interpretation.
• The need for efficient and accurate preliminary disease identification is growing.

Purpose of the Study:

• To develop and validate computational methods for disease recognition based solely on patient symptoms.
• To explore the efficacy of different machine learning algorithms in classifying diseases at various levels of specificity.
• To assess the potential of symptom-based prediction for intelligent medical triage and common disease diagnosis.

Main Methods:

• Medical data were categorized into main disease groups, subclass types, and specific diseases.
• Two distinct symptom-based disease recognition methods were implemented: a neural network and a support vector machine (SVM).
• The accuracy of both the neural network and SVM algorithms was rigorously tested and compared for disease identification.

Main Results:

• The study successfully demonstrated the feasibility of automatic disease prediction using only symptom data.
• Both the neural network and support vector machine (SVM) algorithms showed promising results in identifying diseases.
• Comparative analysis confirmed the accuracy of symptom-based prediction for different disease classification levels.

Conclusions:

• Automatic disease prediction from symptoms is achievable, supporting intelligent medical triage.
• Symptom-based diagnostic tools can aid in the efficient identification of common diseases.
• Machine learning approaches offer a viable pathway for enhancing preliminary medical assessments.