Projects / Programmes source: ARIS

Model of artificial neural networks (ANN) - help for clinicians in selecting adjuvant treatment in breast cancer patients

Research activity

Code Science Field Subfield
3.04.00  Medical sciences  Oncology   

Code Science Field
B520  Biomedical sciences  General pathology, pathological anatomy 
B200  Biomedical sciences  Cytology, oncology, cancerology 
artificial neural networks, breast cancer
Evaluation (rules)
source: COBISS
Researchers (6)
no. Code Name and surname Research area Role Period No. of publicationsNo. of publications
1.  00702  PhD Marjan Budihna  Oncology  Researcher  2003 - 2005  125 
2.  15875  PhD Cvetka Grašič Kuhar  Oncology  Junior researcher  2003 - 2005  261 
3.  20055  PhD Erika Matos  Oncology  Researcher  2003 - 2005  197 
4.  15835  MSc Bojana Pajk  Oncology  Researcher  2003 - 2005  86 
5.  20176  PhD Uroš Smrdel  Oncology  Researcher  2003 - 2005  121 
6.  11747  PhD Branko Zakotnik  Oncology  Head  2003 - 2005  425 
Organisations (1)
no. Code Research organisation City Registration number No. of publicationsNo. of publications
1.  0302  Institute of Oncology Ljubljana  Ljubljana  5055733000  15,789 
Traditionally, breast cancer outcome prediction is based on TNM classification system (T - tumor size; N - nodal involvment; M - distant metastases). However, besides these classical pathologic variables, the outcome of the disease could be influenced by other markers of biological agressivness. Recently, cancer prediction models involve higher number of possible prognostic variables organised in artificial neural networks (ANN) model. ANN model processes information like neurons in the human brain. ANN model, which has been trained on the large amount of clinical examples, should be capable for accurate prediction of the outcome of the cancer patient. It is specially suitable for data sets with complex multidimensional non-linear functions and for prognostic variables which are changing during the time. With ANN model outcome prediction (risk of relapse or cancer) is more accurate than with TNM system or with the use of classical statistical methods (Cox's proportional hazards model or logistic regression). At the Institute of Oncology, Ljubljana, Slovenia, based on our own patients data sets, we want to establish ANN outcome prediction model for breast cancer patients. As input variables, we want to include classical clinico-pathological and some newer biological markers and additionally the data about adjuvant treatment of breast cancer patients. ANN models published by other researchers rarely include adjuvant treatment, or had a small set of patients only. By means of our own ANN model we would like obtain a complementar method for clinical decision making about the type of adjuvant treatment. Our patient could be supplied by accurate individual prognosis/prediction about risik of relapse or cancer survival with specific choice of adjuvant treatment.
Views history