Approaches to Medical Decision-Making Based on Big Clinical Data

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Big Data Analytics

Background:

• Developing effective medical decision support systems (MDSS) is crucial for improving patient care.
• Traditional MDSS often rely on disease-specific data, limiting their generalizability.
• Big data analytics offers potential for more robust and adaptable MDSS.

Purpose of the Study:

• To evaluate different big data processing methods for building medical decision support systems.
• To compare the accuracy of case-based reasoning, simple single-layer neural networks, and probabilistic neural networks for medical recommendations.
• To identify the most efficient approach for MDSS development using universal big data methods.

Main Methods:

• Utilized big data processing techniques without data reduction.
• Applied universal teaching methods independent of disease classification standards.
• Assessed and compared recommendation accuracy using case-based reasoning, simple single-layer neural networks, and probabilistic neural networks.

Main Results:

• Probabilistic neural networks demonstrated superior accuracy in recommendation compared to case-based reasoning and simple single-layer neural networks.
• The study confirmed the effectiveness of universal big data processing methods for MDSS.
• No data reduction was necessary, preserving the integrity of the big data.

Conclusions:

• Probabilistic neural networks are the most efficient approach for developing big data-driven medical decision support systems.
• Universal big data methods can be successfully applied to medical decision support without disease-specific constraints.
• The findings support the advancement of AI in healthcare through robust data analysis techniques.