Автоматизація діагностики медичного обладнання на основі семантичного аналізу текстових описів несправностей

Abstract

Decision-making in medical equipment service centers is complicated by uncertainty and unstructured fault descriptions. A decision support module integrated into an information system is proposed. The approach combines semantic vector representation of textual descriptions with machine learning methods for predicting repair parameters. Text data are transformed into numerical vectors to perform similarity search among historical repair cases using cosine similarity. Retrieved cases provide recommendations on possible causes and required repair operations, while machine learning models forecast repair cost and duration. The proposed solution improves repair planning justification while preserving the engineer’s final responsibility.