Medical Expert Diagnostic Decision Support System: Results and Validation

Авторы

Благосклонов Н. А.

Аннотация

There are over 7,500 hereditary diseases worldwide, most of which are rare (orphan). Many manifests in early childhood, posing diagnostic challenges that require interdisciplinary collaboration (pediatricians, geneticists, neurologists). Early and accurate diagnosis remains critical due to the complexities of differential diagnosis, enabling timely initiation of specific or symptomatic treatment. To address this, a new version of the GenDiES medical expert system (Genetic Diagnostic Expert System) has been implemented to support diagnostic decision-making during the pre-laboratory evaluation of suspected hereditary diseases. Designed for geneticists, the system covers 30 clinical forms of lysosomal storage diseases: mucopolysaccharidoses, mucolipidoses, and gangliosidoses. Key improvements in this version include adjustments to the modality certainty factor scale, optimization of the decision-making algorithm, and expanded explanations for generated conclusions. The updated system is deployed as a freely accessible web application. Accuracy was evaluated in three stages: testing (50 cases from international “case report” publications), validation (54 electronic health record (EHR) extracts from a single clinic), and verification (38 EHR extracts from three genetic centers across different regions). Success was defined by inclusion of the confirmed diagnosis in the system’s top five hypotheses. Accuracy scores were 0.84 (testing), 0.87 (validation), 0.90 (verification).

Внешние ссылки

DOI: 10.1007/978-3-032-13612-1_37

Скачать сборник статей с сайта издательства Springer Nature (англ.): https://link.springer.com/content/pdf/10.1007/978-3-032-13612-1.pdf

ResearchGate: https://www.researchgate.net/publication/399128415_Medical_Expert_Diagnostic_Decision_Support_System_Results_and_Validation