Region-Based Semi-Two-Stream Convolutional Neural Networks for Pressure Ulcer Recognition.

Pressure ulcers are a common, painful, costly, and often preventable complication associated with prolonged immobility in bedridden patients. It is a significant health problem worldwide because it is frequently seen in inpatients and has high treatment costs. For the treatment to be effective and to ensure an international standardization for all patients, it is essential that the diagnosis of pressure ulcers is made in the early stages and correctly. Since invasive methods of obtaining information can be painful for patients, different methods are used to make a correct diagnosis. Image-based diagnosis method is one of them. By using images obtained from patients, it will be possible to obtain successful results by keeping patients away from such painful situations. At this stage, disposable wound rulers are used in clinical practice to measure the length, width, and depth of patients' wounds. The information obtained is then entered into tools such as the Braden Scale, the Norton Scale, and the Waterlow Scale to provide a formal assessment of risk for pressure ulcers. This paper presents a novel benchmark dataset containing pressure ulcer images and a semi-two-stream approach that uses the original images and the cropped wound areas together for diagnosing the stage of pressure ulcers. Various state-of-the-art convolutional neural network (CNN) architectures are evaluated on this dataset. Our experimental results (test accuracy of 93%, the precision of 93%, the recall of 92%, and the F1-score of 93%) show that the proposed semi-two-stream method improves recognition results compared to the base CNN architectures.

Pressure Injury Prevalence and Risk Factors: A National Multicenter Analytical Study.

PURPOSE: The purpose of this study was to determine the point prevalence (PP) of general pressure injuries (PIs), hospital-acquired PIs, PI-related risk factors, and PI preventive interventions performed by nurses. DESIGN: Descriptive, multicenter, prospective, analytical study. SUBJECTS AND SETTING: The sample comprised 5088 patients cared for in 13 hospitals in 12 geographic regions of Turkey. Data were collected between November 5, 2018, and July 17, 2019. METHODS: The study was carried out in 2 stages. First, nurses who collected data were trained in the diagnosis of PI, risk assessment, staging, and prevalence studies, and informed about the purpose and methods of the study, including data collection. Second, nurses and researchers who had received training related to data collection for this study conducted a PP study for PIs in their inpatient clinics using the ASSIST II method. The PI Prevalence Study Tool and the Braden Scale for Predicting Pressure Sore Risk were also used during data collection. RESULTS: The PP of general PIs was 9.5%; the prevalence of PIs with hospitalization in intensive care units was 43.2%; medical device-related pressure injuries prevalence was 10.7%. We found that 65.1% of the PIs were acquired after hospital admission. CONCLUSIONS: Similarities exist between PI prevalence in Turkey and reported PI prevalence rates worldwide. However, the prevalence of nosocomial PIs related to intensive care units and the prevalence of all nosocomial injuries were higher than rates previously reported. Based on results, there is a need to develop strategies to reduce the prevalence of nosocomial PIs.

Impact of tailored training about pressure injuries on nurses' knowledge levels and pressure injury point prevalence: The case of Turkey.

AIM: This study was conducted to determine the impact of tailored training provided to nurses for preventing pressure injuries (PIs) on nurses' knowledge levels and the PI point prevalence (PP). MATERIALS AND METHODS: This interventional study was carried out in a university hospital with a bed capacity of 1114 in an urban center in Turkey. Ethics committee approval (28.06.2018/31) and institutional permission were obtained for the study, in addition to the nurses' written, informed consent. The study was completed in three stages. In the first stage an initial PP study was conducted in the clinics with the participation of the nurses and the members of the research team (n = 422 patients). In the second stage the knowledge levels of 194 nurses were measured before training was given on following-up and preventing PIs. The nurses then participated in the tailored training and their knowledge levels were re-measured afterwards. All the nurses were given individual advice related to the prevention of PIs for 30 days after they had completed the training. In the third stage a second PP study was conducted four months after the first PP study (n = 454 patients). The data were collected using the Pressure Injury Prevalence Form, the Braden Pressure Ulcer Risk Assessment Tool and the Knowledge Level Measurement Form. Descriptive values, the paired samples t-test, Pearson's chi-squared test and Fisher's Exact test were used to evaluate the data. RESULTS: The nurses' pretest mean knowledge score was 55.36% ± 14.40 and their posttest mean score was 69.92% ± 9.73. The difference between these scores was statistically significant (p < 0.05). The study found no significant difference between the first PP ratio and the second PP ratio (p > 0.05), and the nurses were better able to evaluate skin and PIs after the training. CONCLUSION: The study determined that the tailored training given to the nurses increased their knowledge; however, it had no impact on the PP after four months. It is recommended that any training programs using this model be continued and that PP studies of institutions be conducted annually.