Comparison of Artificial Neural Networks and Logistic Regression for 30-days Survival Prediction of Cancer Patients

Abstract

Introduction: A machine learning technique that imitates neural system and brain can provide better than traditional methods like logistic regression for survival prediction and create an algorithm by determining influential factors. Aim: To determine the influential factors on survival time of palliative care cancer patients and to compare two statistical methods for better prediction of survival. Methods: One-year data is gathered from the patients that we followed in the palliative care clinic of our hospital (2017-2018) (n = 189). All data were retrospectively evaluated. After descriptive statistics, we used Pearson and Spearman correlations for parametric and non-parametric variables. The Artificial Neural Networks (ANN) and logistic regression model were applied to parameters which have a significant correlation with short survival. Results: Significantly correlated variables with short survival were Palliative Performance Scale (PPS), Edmonton Symptom Assessment System (ESAS), Karnofsky Performance Scale (KPS), brain, liver, and distant metastasis, hemogram parameters, cero-reactive protein (CRP) and albumin (ALB). ANN model showed 89.3% prediction accuracy while the logistic regression model showed 73.0%. ANN model achieved a better AUC value of 0.86 than logistic regression model (0.76). Discussion: There are several prognostic evaluation tools such as PPS, KPS, CRP, albumin, leukocytes, neutrophil were reported several studies as survival-related parameters in logistic regression models, also. Many studies compare ANN with logistic regression. When we evaluated these parameters totally, we observed the same relations with survival then we used the same parameters in the ANN model. The effectivity of the survival prediction models can be improved with the use of ANN. Conclusion: ANN provides a more accurate estimation than logistic regression. ANN model is an important statistical method for survival prediction of cancer patients.

Comparative Analysis of Web of Science and Pubmed Indexed Medical Journals Published in Former Yugoslav Countries

“…Number and percent of the evaluated studies that failed to meet each of the methodological criteria tested, number of the evaluated criteria not satisfied per database and number of studies that satisfied more than 4 criteria per database are presented in the Table 2 . References of the analyzed studies referred in both WoS and MEDLINE could be found in our previous publication ( 6 ), while the references of the studies covered only by WoS or only by Pubmed/MEDLINE ( 10 - 96 ) medical journals are given at the end of this article ( Table 3 and 4 ). When explanatory potential of journal impact factor, number of citations, time elapsed from publication and a database where a journal is referred were tested by linear regression in regard to the number of methodological criteria satisfied per study, the linear regression model was obtained by backward deletion method and achieved R2 adjusted of 0.166 (F=13.827, df1 = 2, df2 = 127, p=0.000).…”

Impact of Augmented Intelligence on Utilization of Palliative Care Services in a Real-World Oncology Setting

“…20,21 The shortage of PC specialists and our aging population underscore the growing need for a tool to address the challenge of accurately identifying patients at risk for death in the short term. 22 Although several studies have evaluated ML tools to predict mortality in patients with cancer, [3][4][5][6] to our knowledge, thus far, only one has been shown to influence clinical practice, by increasing the number of Serious Illness Conversations. 23 However, the intervention in this trial combined the ML insights into 180-day mortality with behavioral nudges, using text message reminders, and performance reports and data on peer comparisons for Serious Illness Conversations.…”

“…0.9 at both 30 and 90 days), indicating accuracy on par with or better than other tools in development. [3][4][5][6] Third, this study examined the deployment of the AI tool in community practice, a setting where most patients with cancer in the United States receive treatment. 24 This setting, combined with our large sample size, which included patients with many different cancer types, may make the results more generalizable to the overall population.…”

Machine learning in oncology—Perspectives in patient-reported outcome research

“…We present two studies that show the potential of PROs and machine learning in oncology. Firstly, Arkin et al [2] used neural networks to predict the overall survival of palliative patients with cancer. In addition to clinical variables, the authors used a PRO instrument, the Edmonton Symptom Assessment Scale (ESAS), as a predictor of survival.…”