Archives

  • 2019-07
  • 2019-08
  • 2019-09
  • 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • 2021-03
  • br The majority of previous

    2019-10-10


    The majority of APTSTAT3-9R previous research on breast cancer has focused on expert systems approaches for breast cancer diagnosis, risk esti-mation, and prognosis (see, e.g., [13,–17]), while relatively little re-search has focused on the multi-criteria decision-making methods for determining an appropriate treatment for breast cancer. The National Comprehensive Cancer Network (NCCN) provides breast cancer treatment guidelines for specialists and patients in terms of decision trees involving various criteria that impact treatment options [18]. However, while multiple trees are presented, it is still not an easy task to derive a treatment plan for each specific pa-tient based on the NCCN guidelines.
    Adunlin et al. [19] stated that multi-criteria decision analysis (MCDA) is well suited to address issues related to benefit-risk trade-offs between treatment options for breast cancer. Given a set of explanatory variables that include the patient’s demograph-ics, health condition, and cancer treatment regimen Kibis et al.
    [20] studied different machine learning methods including neural networks, regression, and Bayesian belief network to predict the ten-year survivability of a breast cancer patient after initial diag-nosis. Carter et al. [21] compared the Markov model with the AHP and the analytical network process (ANP) to determine a treat-ment strategy for breast cancer, and they discuss the advantages and disadvantages of each. They considered observation, radiation, tamoxifen, a combination of radiation and tamoxifen, and simple mastectomy (MA) as the alternatives. They find that all three mod-els recommend the same preferred treatment choice. The Markov process provides more detailed results, whereas AHP and ANP give
    only rank APTSTAT3-9R of the alternatives but include more patient in-put. However, this work presents general results and does not se-lect a patient-specific treatment strategy. Çakır et al. [22] proposed a data mining approach to determine an appropriate treatment for breast cancer patients using data from 462 breast cancer patients. However, our approach differs from their approach by following an expert-based methodology and presenting specific treatment regi-mens based on the health condition of each patient.
    Different methods of cancer treatment are being used by on-cologists. For example, Giordano et al. [23] used a multivariate lo-gistic regression in order to determine the factors associated with CTx use. Their final model included patient age; year of diagno-sis; surveillance, epidemiology, and end results (SEER) region; eth-nicity; education level; poverty level; cancer stage; grade; ER sta-tus; tumor size; number of positive lymph nodes; surgery, radia-tion; and comorbidity indices. They also used the multivariate Cox proportional hazards model to calculate the hazard of death from breast cancer and overall mortality after adjusting for confound-ing variables. While the authors present a comprehensive study on breast cancer, they did not consider other therapies such as radio-therapy or hormone therapy.
    2.2. Limitations of literature and research contribution
    Selecting an effective therapy for breast cancer is a complex decision-making process. First, a large number of criteria exist in the decision process. Second, every patient’s case is unique, and thus a specific treatment plan is needed for each patient with breast cancer. Third, there is no standard procedure for each partic-ular stage. Given that there is a vast number of different possible cases and multiple conflicting criteria, it is a challenge for clini-cians to determine an appropriate and effective treatment plan.
    While prior work considers multi-criteria decision-making methods for cancer treatment, to our knowledge, none of the for-mer studies provides a detailed set of criteria and treatment al-ternatives to determine the most effective treatment strategy for breast cancer. In addition, the relative importance of factors in de-termining the appropriate breast cancer treatment and ranking of different treatment alternatives have not been systematically stud-ied using a multi-criteria approach. In addition, to the best of our knowledge, a patient-specific breast cancer treatment selection has not been studied using a multi-criteria approach including a hier-archical AHP model and a ranking algorithm.