A health-promoting system for analyzing health risks and predicting diseases

Abstract

A health recommendation system for disease diagnosis and feature selection, consisting of: a data acquisition unit configured to acquire initial raw data to be used for feature selection; a data preprocessing module configured to preprocess the acquired raw data, wherein the preprocessing module is configured to clean input health records, process missing values, and adjust record characteristics; a feature selection module based on multi-objective genetic Harris-Hawk optimization (MOGAHHO) configured to: initialize a population of search agents, perform genetic algorithm operations on a first subset of search agents, perform Harris-Hawk optimization on a second subset of search agents, evaluate the fitness of the generated solutions, maintain a non-dominated solution repository using the Pareto dominance principle, and select optimal features based on multiple objectives; a feature recommendation module connected to the feature selection module and configured to further evaluate the features selected by the feature selection module by implementing the TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) technique of the MCDM (Multi-Criteria Decision Making) algorithm for feature recommendation; and a model recommendation module connected to the feature selection module and configured to: implement automated machine learning using the Tree-based Pipeline Optimization Tool (TPOT), identify the best classifier for the reduced dataset, recommend a predictive model with the highest classification accuracy, and output the recommended features and the predictive model.

Images