Objectives:The objectives of this study included evaluating and reporting on the initial impact of the COVID-19 pandemic and preventive measures in the form of a lockdown on self-poisoning tendencies in northern Poland. Material and Methods: The authors retrospectively analyzed medical records of all patients (N = 2990) admitted to the Pomeranian Center of Toxicology in 2018-2020. Of those, further analysis included 2140 patients who had been admitted because of a suicide attempt by self-poisoning. The authors also selected a group of 40 patients on the basis of a self-reported direct relationship of the suicide attempt with the COVID-19 pandemic or the imposed lockdown. Results: The rates of suicide attempts in toxicological patients over the years ranged 68.18-75.3%. The patients were predominantly female, with age between M±SD 33.2±16.9 and 36.0±16.4. Each year, over 60% of patients were admitted during their first attempt and were treated psychiatrically prior to their attempt, with differences observed in the COVID-19-related group. The alcohol intoxication during the suicide attempt was confirmed in 37.40-43.53% of the patients, with a higher rate of 52.50% observed in the COVID-19-related group. The main self-reported reason for the suicide was a romantic relationship conflict or breakup, and a conflict and/or violence in the family. The most frequent agents were over-the-counter painkillers, antidepressants, antipsychotics and benzodiazepines or Z-drugs. Conclusions: During the initial year of the COVID-19 pandemic, there was a fall of suicide attempts by self-poisonings in northern Poland, significant only in the case of women. The self-reported reasons were similar in all years, with mainly minor changes. There was also an increase in attempts made using benzodiazepines or Z-drugs seen in 2020 and in the COVID-19-related group. The authors believe that there is a need for multi-center, large-scale prospective studies that would provide better insight into the pandemic-related suicidal trends.

Rhabdomyolysis is the process of striated muscle cell lysis, during which proteins and microelements such as myoglobin are released into the bloodstream. It is important to diagnose rhabdomyolysis as soon as possible and start the treatment according to severity, as it is a state that significantly increases the mortality of the patients. The current gold standard of rhabdomyolysis diagnosis is the creatine kinase plasma concentration test, but it can be also diagnosed with imaging techniques, such as ultrasound (US). This review aims to gather previously published information regarding sonographic appearance of rhabdomyolysis. We searched through PubMed and ScienceDirect databases for studies using designed queries. After the selection process we were left with 13 studies containing a description of US appearance of rhabdomyolysis confirmed with a CK plasma level test. Findings described in the majority of the cases were muscle thickening, ground glass opacity, traits of edema and anechoic areas. Other than these, there were several less often reported findings. As a conclusion, rhabdomyolysis seems to have its own US appearance, but for now it cannot be precisely specified and needs further research for clarification.

Machine learning techniques play an important role in building predictive models by learning from Electronic Health Records (EHR). Predictive models building from Electronic Health Records still remains as a challenge as the clinical healthcare data is complex in nature and analysing such data is a difficult task. This paper proposes prediction models built using random forest ensemble by using three different classifiers viz. J48, C4.5 and Naïve Bayes classifiers. The proposed random forest ensemble was used for classifying four stages of liver cancer. Using a feature selection method the reliable features are identified and this subset serves as input for the ensemble of classifiers. Further a majority voting mechanism is used to predict the class labels of the liver cancer data. Experiments were conducted by varying the number of decision trees generated using the J48, C4.5 and Naïve Bayes classifiers and compared with the classification made using decision stump and Adaboost algorithms.

No abstract