Search

Significant developments are currently underway in the field of cancer research, particularly in Germany, regarding cancer registration and the use of medical information systems. The use of such systems contributes significantly to quality assurance and increased efficiency in data evaluation. The growing importance of artificial intelligence (AI) in cancer research is evident as these systems integrate AI for various purposes, i.e. to assist users in data analysis. This paper uses ensemble learning to classify the graphical user interface state of the medical information system CARESS. The results show that all ensemble learning models utilized achieved good performance. In particular, the gradient boosting algorithm performed the best with an accuracy of 97%. The results represent a starting point for further development of ensemble learning in medical data analysis, with the potential for integration into various applications such as recommender systems.

Approximately 5,000 of 6 million annual visitors of the Oktoberfest in Munich have to undergo medical treatment. Patients with alcohol intoxication without trauma or further complications are all treated in a specialized medical camp. We studied these patients in order to identify risk factors and to assess the relevance of the Glasgow Coma Score (GCS) and of ethanol blood concentrations for patient management. In 2004 totally 405 patients suffering from ethanol intoxication without trauma were treated in the medical camp. A complete set of the following data was obtained from all 405 patients: GCS, ethanol blood concentration, age, sex, blood pressure (mean, systolic and diastolic), body temperature, heart rate, blood sugar, GOT, gamma-GT, and CK. A multivariate logistic regression model was applied to identify risk factors predicting patients at increased risk of hospitalization. Low GCS (< or =8 vs. >8, OR: 4.18, CI: 1.96-8.65) low age (20-29 vs. > or =30 years, OR: 2.35, CI: 1.05-5.65) and male gender (male vs. female, OR: 3.58, CI: 1.36-9.34) independently predicted patients that had to be hospitalized. All other parameters including ethanol blood concentrations were not explanatory. Patients with GCS < or = 8 (n = 66) had a lower median blood pressure (P = 0.0312) and showed a smaller increase in blood pressure during the observation period compared to patients with GCS > 8 (P < 0.001), suggesting that this subgroup may require longer recovery periods. Men aged 20-29 years were at highest risk for hospital admission. Increased risk could not be explained by higher ethanol blood concentrations in this subgroup. Importantly, GCS < 6 does not justify endotracheal intubation in ethanol intoxicated patients, when further complications, such as trauma, can be excluded.

Background: Urokinase-type plasminogen activator (uPA) and plasminogen activator inhibitor type 1 (PAI-1) contribute to the invasiveness of many carcinomas. Here, we studied a possible association between cytosolic uPA and PA-1 concentrations in tumor tissue with prognosis in patients with endometrial cancer., Methods: Cytosolic concentrations of uPA and PAI-1 were determined in 69 primary endothelial adenocarcinomas using an enzyme-linked immunoassay (ELISA). A possible influence of uPA and PAI-1 was studied by multivariate Cox regression adjusting for the established clinical prognostic factors FIGO-stage, grading, depth of invasion, diabetes mellitus and age., Results: Both uPA (p=0.011) and PAI-1 (p=0.003) were associated with relapse free time using the multivariate proportional hazards model. Association with overall survival was less pronounced with p=0.021 for uPA and p=0.358 for PAI-1. Concentrations of PAI-1 increased with FIGO stage (p=0.003) and with histological grading (p=0.005). Both uPA and PAI-1 concentrations were negatively correlated with estrogen and progesterone receptor levels., Conclusion: The combination of high cytosolic concentrations of uPA (>5 ng/mg total protein) and high PAI-1 (>20 ng/mg total protein) may reveal a group of patients with increased risk of progression.