Page 155 - AIH-2-3

Artificial Intelligence in Health Explainable solutions from AI for HSSs

heart diseases, diseases characterized by increased blood 6. Comparing different approaches to the
pressure, and chronic rheumatic heart disease. Usually, automation of medical activities
experts form a clinical picture of diseases with dozens of
dynamic symptoms. The knowledge about the diagnostic Today, specialists from different medical teams are using
signs of biliary tract disease was built inductively, based medical IACPaaS services to test the capabilities of AI
on the data set of their surgical department. In addition, in solving various problems for formalized health case
from the knowledge built by the linguistic processor histories. For general practitioners, differential diagnosis
Ontosminer (http://ontosminer.opkrt.ru/) based on the services are being created, and for gastroenterologists,
analysis of millions of documents from the free resource a complex for diagnosis, treatment, and prognosis of
PubMed, some clinical guidelines were selected for which recovery. For cardiologists, the complexes are made so that
it was possible to carry out validation based on real case the risk assessment is carried out in two ways: based on
histories. Specialists in mucopolysaccharidoses manually ML and formalized knowledge. In comparison, advice on
created them, tested them on real examples of patients diagnostics is given in two other ways: based on knowledge
from different countries, and refined the knowledge. and formalized precedents.
Developing knowledge bases and intelligent software Medical software assistants (solving different tasks) are
components for medical services was carried out in built using IACPaaS tools and components of the IACPaaS
parallel. Such services are intended to help medical teams medical portal. All assembled systems work based on a
and institutions support the solution of the problems single terminological base of symptoms and facts (more
of their intellectual activity, providing a “third opinion” than 20 thousand) with synonyms.
through cloud means. Their hypothesis explanations rely
on formalized knowledge (Figure 6). The overall knowledge base is large and maintained
by multiple experts. Updates are carried out according
The labor costs are fully justified because each solver- to procedures that keep the cloud services running. For
interpreter works for more than one profile, and each example, knowledge on some digestive system diseases has
nosological base is used to solve several intellectual been expanded, information on the regional manifestations
problems (diagnosis, treatment, and prognosis). The of fever caused by rodents has been clarified, and new, unique
“cloud” implementation of ClDSS allows the monitoring knowledge on metabolic disorders of glycosaminoglycans,
of relevance and evolving a single clinical guidelines mucolipids, and gangliosides has been added.
base for several profiles and classes of tasks. The general
base accumulates the experience of several professional Knowledge base editing and verification tools are used
communities and teams in addition to universal knowledge. to update current diagnostic and therapeutic knowledge.

Figure 6. The fragment of hypotheses explanatory in diagnosing the Intelligent Adaptive Clinical Platform service

Volume 2 Issue 3 (2025) 149 doi: 10.36922/aih.5736