Architecture, Implementation and Experimental Study of the Health Passport Intellectualized Software Package for the Diagnosis of Clinical and Hematological Syndromes
Abstract
The article presents the architecture, software implementation, and experimental validation of an intellectualized software package developed within the framework of the "Health Passport" digital platform for automated diagnosis of clinical and hematological syndromes, in particular, anemias of various morphological and biochemical nature. The proposed system is implemented in the form of a modular software and hardware complex that covers the entire analysis cycle: from the collection and pre-processing of laboratory data to the calculation of diagnostic indices, morphological classification, probabilistic forecasting and the formation of a diagnostic conclusion. The key modules of the platform include: Data Acquisition (structured data import), Data Preprocessing (data cleaning and normalization), Mathematical Evaluation (calculation of MCV, MCH, MCHC indices and aggregated anemia score), Morphological Classification (anemia presence, type and character determination), as well as Ensembling and Neural modules for building and training machine learning models and neural networks. As part of the experimental study, classification accuracy was achieved up to 98.1% (XGBoost) and 100% (neural network classifier with probabilistic output). The system demonstrated the stability, reproducibility and clinical interpretability of decisions. The results obtained indicate the high applied importance of the complex as a tool for standardizing laboratory diagnostics, reducing the burden on doctors and integrating into digital healthcare ecosystems.