Early diagnosis of skin melanoma in the samara region in 1995–2019 using multi-agent technology

Melanoma  of  the  skin  (MS)  is  a  tumor  with  a  predictable  course,  when  diagnosed  at  the  pT1a  stage, the patient has an almost 100 % chance of curing the disease. Diagnosis of MS is carried out mainly by doctors of primary contact, the accuracy of the diagnosis is low. In a large industrial region with more than three million people, a study was conducted that showed a significant increase in the number of cases. In this regard, a multi-agent technology for  melanoma  diagnostics  was  developed  in  the  Samara  region.  From  2010  to  2019,  an  experiment  was  carried  out to introduce a multi-agent system for early diagnosis of MS, the results were analyzed for the period of its use (2010– 2019) and in comparison with another period – 2000–2009.

The aim of the study is to assess the effectiveness of early diagnosis of MS in the period from 1995 to 2019 in the Samara region in connection with the introduction of multi-agent technology.

Materials and methods.A retrospective analysis of the quality of diagnostics was carried out using a multi-agent system, consisting of the author’s expert system based on artificial intelligence (patent for invention No 2018620399, 2018613016), and without its implementation for the periods from 2000–2009 and 2010–2019. The primary documentation was analyzed: form 7, 35, 025/u. The coefficients of reliability (t, p) of the proportion of stages I–II and “thin” skin melanoma, linear and polynomial trend approximations were calculated. The result of histological examination was used as a reference method.

Results.It was shown that the proportion of diagnosed MS at stages I–II significantly increased as a result of the introduction of a multi-agent system (t= 0.04113219, p<0.05), the proportion of “immature” MS did not give statistically significant changes (t= 0.07303817, p>0.05).

Conclusion.It was concluded that for effective diagnosis of MS at stages I–II, the use of multi-agent technology is sufficient, but for diagnosis at the pre-invasive stage it is ineffective and requires an increase in the number of trained expert doctors.